I see you advertising that robotic vacuum cleaner…. My question is does this company sell my information about my house and floor layout with the information to collect from cleaning my house? Just a question if you haven’t looked into it, you’re probably should. What are the biggest markets today and growing is data?
When the roborock can pick up paper and put it in the recycling, pick up discarded socks and put them in the laundry, and do so without nagging me for untidiness, then I'll be interested.😊
there has to be some betrayal and revenge. people are involved. on analog vs digital, i saw a digital and analog paperclip-legged walker. the only change was bending one of the legs so it didn't touch the ground. the analog handled it much better than the digital. analog can't take over everything, but it's underutilized.
My first experience of digital computing was in 1962 when the U.S. Navy trained me for their first three ships at sea to be equipped with them. While those early computers were much smaller than an IBM mainframe of the time, they were the size of large double door refrigerators, and needed chilled water to cool the processors. Energy costs were high despite the transistors they used, and processing times were much slower than the simple digital devices we use today. The digital computers were used as an air traffic control system, for combat and it fed some of the information about potential targets for our missiles to the fire control technicians on a different set of display consoles. Their computers and consoles were analog. At that time the speed of their analog resolvers and differentiators was superior to any digital simulation our computers could produce. But they were constantly having to fine tune the analog elements to their system to maintain accuracy. As you note, digital can be very precise and repeatable.
The processing power of this little old Galaxy 9 cell phone is quite the Quantum Leap in technology isn't it ? I remember playing T.V Pong on a black and white T.V 40 years ago. Now we're colliding particles ? Come on. Men just stopped using bows and arrows to rule the world and now we have "Rods from God" ? Perhaps C.A.R.A.T [Commercial Applications for Recovered Alien Technology] isn't a mythical organization....after all.
When I toured the USS Misouri they told us that when they took the ship out of mothballs for the Iraq war the dig computers were not as accurate as the original annalog computers, so they re installed the annologs to aim her 16 inch guns. Really interesting.
I started my career working on analogue computers used in industrial and manufacturing environments. They were mostly used for controlling motors and used proportional, integration and differential calculations at very high speed. They worked well in environments that were not computer friendly and were very robust and reliable. That was 40 years ago and they lasted for years because they just worked. I think the challenge was that to program them you needed to understand electronics and mathematics and that isn’t so necessary with modern digital computers.
Yep, when I was taking electronics at vocational school over 40 years, and our instructor talked about analog computers to include computers that used RC time constants to solve log calculations.
Now chat GPT is doing code and now people are gonna have to learn electronics. I personally am happy with analog. It's like reality vs. artificiality with digital vs analog.
LOl we moved away from analogue for a very valid reason, moving parts because heat and wear eventually caused them to fail. Could someone explain what this guy is saying, as there is way too much filler.
Having grown up in the golden age of record players I fall on the side of digital. After a single play on a turntable a record will lose that frequency response curve. It really doesn't matter how expensive the turntable the wear and tear on the record is a function of the design. A physical needle riding grooves wears a bit of those grooves away with each play of the record. The quality of the digital recording is there forever.
A bigger issue with vinyl records is that high frequency response diminishes the closer you get to the center of the record, since the circumference (and hence, linear velocity) decreases. Essentially, you're trying to fit the same amount of information into an increasingly smaller area, resulting in decreased resolution. Inner groove distortion, as it's called, can be mitigated to a large extent with advanced (i.e. expensive) stylus profiles, such as Shibata or Microlinear, as opposed to the more conventional conical or elliptical styli found on most setups, where distortion will be noticeable, more or less. But even then, the fidelity comes nowhere close to that of even then cheapest DAC.
A few decades ago my local newspaper ran a story about an audiophile. Because of the degradation you mentioned, he played his vinyl records once - to record to open reel tape. He listened to the open reel tape and archived the vinyl. When CDs were new, most were created from analog multitrack masters. There were stories of classic albums suddenly revealing studio air conditioning noise on CDs, not heard on vinyl.
I'm a trained singer. The must authentic and believable representation of a human voice I have ever heard was a recording of Gigli played on an old 78 record played on a wind up gramophone with a horn. You could actually feel the voice like you can at a live performance.
@@LagMasterSam Could be? As a musician I've just noticed that over the years the 'quality' has improved with technology (no hisses or crackles, better microphone positioning and balancing etc.), but the issue remains that digital is a synthesis of analogue, and I'm not sure whether you'd ever get there no matter how high you get the resolution. I wonder if photographers feel the same about digital photography?
Love learning about diverse computing methods. Between analog, digital, and quantum computers, we'll be able to solve problems so much better than ever before!
you know, in 1974, researchers had developed an analog matrix method that could recognise the alphabetic and numeric characters, individually, with a very high success rate, and i remember being surprised at thesmall amount of analog hardware to perform the job, once learned/programmed...and, even then, the analog methods were very fast and very low power, per "recognition"
The first scanner for character recognition was used in the 1800’s for sending morse code on the telegraph. I had never heard of this until I started working for Xerox in 2007.
I just realized that analog computers with a digital nanny to adjust for drift or runaway errors is how the fuel control unit in many modern jet engines operates. They use an analog hydromechanical computers to calculate and dispense the fuel as required, and a digital computer to adjust things on the fly to keep everything optimized for the best balance of fuel economy, longevity, power, and emissions. Many small aircraft still use what is essentially a purely analog hydromechanical computer to control the fuel flow to the engine. The fuel injection on these piston engines uses inputs such as throttle position, fuel lever position, airflow, RPM, and air pressure to compute and meter out the fuel necessary for those parameters. Every annual or 100 hour inspection includes checking the fuel mixture at high and low power settings and adjusting as needed to keep the fuel system output correct.
I had a little trade school on automatic transmissions of the 1980s and before. I always asserted that the valve body in one of those transmissions was essentially an analog computer that operated hydraulically instead of electronically. I still assert that.
I remember using mechanical computers in the Navy that measured speed of targets for our Fire Control systems. Those computers wear far faster than anything around at the time.
I like vinyl, but other analog particles get in the way, like dust, unfortunately and cause a crackling sound. I think I would prefer digital sound at 24-bit, but if there was a way to get clean sound from analog audio, I would definitely prefer this.
The whole point of this is that for AI training they don't need a clean signal. A somewhat noisy value is good enough, just like vinyl gives you a good enough analog of the music encoded into it.
Yeah, Digital was never truly Digital you know, it was always just some conventions about analogue voltage levels(TTL/CMOS), and in a usecase like this, storing neuron BIAS values "vertically", we're basically removing those conventions and unleashing the full potential that every "digital Bit" had in the 1st place. LOVED this video... it suited my old school analogue brain really well 🙂
digital still uses real world transistors that are not binary but the logic is. inaryl. analog can't run software as in binary code compiled from source, in an analog computer it's just a fast, efficient and approximate parallel computation. when computing power and efficiency matter more than precision such as executing cheaply a neural network that has already be trained. an analog neural network is just baked in "intelligence" that cant learn anything
@@rogerphelps9939 not arguing there, Digital wont ever go away, but specifically for AI model storage i see a huge upside if they manage to save hundredfold on power and space, because that also increases reading speed. The idea is to utilize the untapped capacity within each byte, surely there are costs and caveats that comes into that, and that's why you are correct in your "digital supremacy" statement !
Great content today and always. I am somewhat of a music snob and grew up in the analogue era. I embraced the digital movement in music as it was introduced and find digital very satisfying. My ears can only hear so much frequencies as many of us baby boomers grew up in the rock n roll era . Therefore as close as digital simulation is getting to analogue it is good enough for a large percentage of the population's ears. Thanks again and keep the great content coming
You're mixing up being "digital" with the ability to generate high or low frequency sounds. Low sample rate devices can still capture and reproduce those same frequencies, but the lower the sample rate, the more data is lost between samples. A lot of digital audio was recorded from analogue sources, but it's the sample rate (and any kind of "lossy" data compression) that affects the quality. The higher the sample rate, the greater the amount of data, and hence the requirement for faster hardware to play it back.
When I was in the Navy, the fire control systems where analog; they used syncros and servos to compute in real-time the solution to fire projectiles onto targets.
Great video. Nice to hear more on analog chips and how they are important. Being a shareholder in Spectra7 active copper cable running on 800G analog chips, this explains a great deal how it’s used. Keep it up
It's not going to replace digital but the integration and use as a tool with digital is great, just like your example for runaway variation. Analogue can do the heavy lifting and digital spot checks.
From what I've read, Mythic's tech isn't really quite analog. One source reported 256 descript gate voltages for each transistor storing a value. This is no doubt closer to analog than digital binary states are, but the values are still quantized. Still, it's an interesting use of flash technology. I guess it also means the data can still be backed up and transmitted digitally.
While you are correct that the floating gate transistors can only be controlled to some precision of discrete gate voltages (even sometimes 4-bit [16 values] in mythic's case), this is not where the "analog" comes from. In fact, these stored values can vary during operation by a significant amount from a variety of factors, so they are even less precise. Moreso the term "analog" comes from the method of not using a bistable device as the main method of computation. Rather, the physical dynamics are used as a mode of computation. From an energy perspective, this is not ideal (since digital to analog or analog to digital conversion takes a lot of energy,) but the main energy sink, the actual computation, is done very efficiently. Unlike "analog" computers of old, such as those seen in navy targeting computers, the modern (near term) search for analog computers is to find very efficient ways to compute with digital data. For this reason, the term "mixed-signal" is commonly used in academia to refer to this approach. There are certainly methods of computing that are all analog, mainly classic analog computers and biological brains, but these are either too difficult to scale (the former) or beyond our current technology (the latter). TLDR: Digital in, compute with analog, digital out.
I do believe that analogue will have a big role in future computing. Though I don’t think it will be a replacement for the core computing of algorithms, it will be excellent for sensory functions and identification. The world is fluid and that’s why analogue will be best to interface with it. Eventually the analogue technology will get better and the interface between these devices and a digital computer will become fluid. The struggle will be to identify the categories between two objects (this is the learning process). Technically everything is different so making an identification is highly complex but when we can create a sensors that identifies with enough points (like 20 for a fingerprint or retina scan OR 30 to identify a species) then we will have a powerful system. For example, if an analogue device can identify from kingdom to species then send that information to the digital system then we have made a huge increase in efficiency. The real benefit will come from “fingerprints”that the human mind currently doesn’t utilize. The analogue can process sounds we can’t hear, temperatures we can’t see, smells, tastes, vibrations, etc. Bottom line: If you build the sensor correctly then you can download indexed identification data that will make observation incredibly fast. It will also be able to identify new identities that were previously unknown. The downside is that a computer will be able to make an assessment about a person from data so rapidly that it can predict / anticipate behavior immediately. The only way to protect privacy will be through the creation of chaos and noise!
This explanation is exactly how electronic devices have worked for decades, the analogue circuits manage the interface to the real world, and the digital circuits do the actual repeatable, reliable work
I love vinyl records, since I started playing my dad's collection at the tender age of 7 or 8 in the early 80s. However, I fully embrace digital music as well. Being able to carry a lot of music around in my phone is wild.
You need about 128khz sampling rate to approach analog vinyl record sound quality, as well as equivalent DAC and inline digital chips... then it's only approaching analog sound quality that we had in the 1950's.
@@VenturaIT No, you don't. All vinyls of the last decades are based on the same digital material you get on CDs, just with what's effectively lossy compression artifacts. Full analog audio was so terrible in comparison that artists were insisting on digital processing for their new albums as soon as the tech for digital vinyls was introduced. I like the sound of vinyls too but in reality it's just a low-tech DAC.
@@SaHaRaSquadIt's more complicated than that. There have been studies done and they realized that to fool the human ear with digital it would need to be at least 128khz sampling rate... "artists" don't know much about these topics... even today many top recording studios use analog tape for the master because it's still superior. Digital as utilized currently is inferior to master quality analog in almost all aspects. "In recent years, digital masters have become usual, although analog masters-such as audio tapes-are still being used by the manufacturing industry, particularly by a few engineers who specialize in analog mastering. Magnetic tape was commonly used to create master copies." - Wikipedia "After two full decades in which digital formats have been arguably the primary music recording methodology, analog has made a comeback. Recent recordings from John Mellencamp, Elton John and Leon Russell, Taylor Swift, and the Secret Sisters were recorded completely on analog tape." - Grammy The trick if you buy records is to buy only the ones that are fully analog through the entire process... not digital that has been put onto vinyl.
The sound waves coming out of the speakers of the TV I'm watching you on are analog. Sound is analog because it inherently exists as continuous, varying pressure waves in a medium like air. These variations in pressure are not segmented into distinct units; instead, they form a seamless and unbroken flow of fluctuations. This means that sound can take on any value within its range, not just specific, predetermined levels. This continuous nature of sound contrasts with digital signals, where values are represented by discrete, separate steps. The analog characteristic of sound allows for an infinite gradation of tones and volumes, leading to its smooth and uninterrupted quality. This is why sound, in its natural form, is fundamentally analog. The light from the video is both analog and digital. Imagine light as being like a smooth, flowing wave (that's the analog part) and also like a bunch of tiny, indivisible packets (like digital bits in a computer game). These tiny packets are called photons. When we look at light as a wave, it seems to change smoothly, like the colors in a rainbow. But when we look closer, we see that light is actually made up of these photons, each like a tiny dot of light, which you can't break down into smaller parts. So, light is special because it can be both a smooth wave and a collection of these tiny, distinct dots at the same time. Voltages in the equation V=IR (Ohm's Law) are analog because they represent continuous values in an electrical circuit. In this context, 'V' stands for voltage, 'I' for current, and 'R' for resistance. The analog nature of voltage means that within a circuit, it can take on a continuous range of values, not just specific, set levels. For example, the voltage across a resistor can be any value within the limits of the circuit's design, smoothly varying as the current (I) or resistance (R) changes. This is different from a digital signal, which would have distinct, separate levels (like 'on' or 'off'). In essence, the relationship expressed by V=IR operates in a continuous, fluid manner, where adjustments to current or resistance lead to proportional and uninterrupted changes in voltage, characteristic of analog behavior.
Analog computing definitely has a place. I don't think it will change mainstream computing, but certainly for application specific uses as you described, it could be a game changer. Data storage seems to be another beneficial area since currently we would need to use 8 bytes per floating point number. With an analog solution a single analog data point could represent that number.
You can represent a floating point number using any number of bytes. The format used to represent the data is decided by the required range of numbers and the accuracy needed.
@@heliusuniverse7460 yeah it's all a matter of precision. The difference between analog and digital is infinite precision vs discrete values. Remember when CD's first came out, and all of the audiophiles refused to listen to them because they preferred their records? This is why, the transition to digital caused a loss of precision. To many it was good enough, but for some use cases it was insufficient. This will be the same. For most things digital is fine, but there are some use cases where infinite precision is a boon.
@@BruceWayne15325 It has always been myth that vinyl is superior to CDs in terms of perceived audio quality under controlled, blind conditions. The only reasonable explanation for it having some truth was that the old record players had an overall superior sound system than the early CD players. With an otherwise equal quality sound system, CDs will produce higher quality sound just from the fact that there's substantially less noise. I think you mean accuracy, and not precision. Analog can be more accurate than digital, but it can’t ever be more precise. Accuracy is about average closeness to the original, while precision is how reproducible the outcome is. Baring some hardware error or anomaly digital is 100% precise because it always gives the exact same result. But even if analog can potentially be more accurate, there is a point where enough digital representation surpasses the noise that's inherent in analog to the extent that digital is also more accurate.
I listen to plenty of digital music all the time: while Im working, driving, doing stuff around the house or gaming. However, when I want to sit down and dedicate my time to listening to an album I much prefer the experience of playing vinyl records. The entire ritual of removing the record from the sleeve, placing it on the platter, brushing it and dropping the needle then settling in to look over the big, beautiful presentation of the album art and liner notes while I listen and enjoy a coffee or whiskey is extremely satisfying and relaxing, almost therapeutic in a meditative way. While there are arguments about sound quality that can be made to support either digital or vinyl, at the end of the day it's the unique experience of engaging with vinyl as a medium that made me, and has kept me, a fan of vinyl records.
The Roborock is amazing when it works. Dealing with their customer service is a whole different story. Some of the accounts in the Amazon reviews are really something, and definitely a must read before you buy.
Analog computers could make a come back but I think the most likely analog system to do it would be an optical one using nonlinear optics. The thing about optics is that a step like doing an FFT and be done instantly. An image can be filtered or processed to detect some object in practically zero time.
Analog computers especially for AI weight algorithms storage where absolute accuracy is not critical, it could be a useful addition, aka a niche application. However, digital has a great strength, accuracy and repeatability.
Sounds like a very efficient & energy efficient system. My curiosity about alternative computing systems is based on the need for resilience vs threats to digital systems such as: cyber attacks, EMP attacks, solar flares, etc. So in line with this: - Do you see analog computing being less vulnerable to the aforementioned threats? - Raising it a notch higher, do you see analog computing being further developed to be a good system to have as a back-up in case of any of the 3 threats (or other threats of similar nature) takes place?
At Oregon State University in the 60s we started by wiring logic boards to do tasks to a deck of punched cards. We later moved to an IBM1620 that was a mix of analog (had dials to set for many engineering functions) and digital (could run an early version of Fortran II). Used tons of power compared to the CDC 3300 that replaced it. Now that digital power requirements are dropping for both CPUs and memory I can see Analog Computers used for a specific tasks but not for General Purpose Computing. One of the issues with Analog Computing is numbers like Pi or e out to multiple decimal points required for many scientific calculations and transforms like Fourier and LaPlas.
No. Just not true. The IBM 1620 was 100% digital. Yes, it contained may analog components. When you get down to the individual transistors, they are all analog, even in fully digital computers. But the circuits did not operate on analog values. The distinction between analog and digital computers is in how numbers are represented. If numbers are represented as specific states, whether there are two states in a binary digital computer, or four states or more in a modern flash memory cell, they are still digital.
Are you sure that the 1620 wasn't a 1710--the process control version of the hardware? Had DAC and ADC peripherals. Still instruction-compatible; just a few added features and instructions. In the 60's, the notion of a "hybrid" computer was all the rage.
@@sonickrnd My kitchen induction stovetop has rotary controls for individual element heat. However, what's used are simple rotary encoders, so that a dial position doesn't correspond to a setting--you have to look at the LED readout to see what heat level you're setting. Which, in my opinion, defeats the whole purpose of rotary controls. I have a blind sister-in-law who is absolutely useless with this kind of control--she has no idea what setting is being established. "Touch" controls are similarly useless to her without some sort of feedback. It seems that human engineering has taken a backward step with this sort of thing.
In WW2, the navy had analog computers that used gears to set range and speed of targets and how to aim the guns. They were quite accurate when used right.
Digital audio is better than those graphs suggest. Any continuous signal can be decomposed into a sum of sine waves of different frequencies and amplitudes using Fourier analysis. According to the Nyquist-Shannon sampling theorem, any given bandwidth of frequencies can be perfectly represented by sampling at double the frequency bandwidth. So, yes, uncompressed digital audio can be equally good as the best vinyl vinyl (arguably better in the real world since vinyl cannot be made perfect). Thus lossless digital can really be lossless.
Laplace, Fourier and other transforms are used to convert between different domains. A Fourier transform will convert even a square wave or square pulse into multiple sine wave components. In theory, if you could isolate those components and feed them all back through an analog summer, you would recreate the original waveform.
People who say that vinyl is a higher fidelity format than CDs or SACDs are extremely delusional. It's okay if you like noise added to your music but that's what's added, noise!
Another pathway to solve the energy efficiency issue would be classical reversible computing. The main tricks used by the designs explored in this video have to do with the fact that you need less bits of informations to describe a physical system when accuracy isn't critical AND you spend much less energy when this less accurate memory is closer to the computing units (in proper neuromorphic computing, those tend to be the same, even). Reversible computing directly goes after the fundamental cause of energy dissipation in digital systems (which, you have to remember, are necessary for faithful information transmission and general error-correction) aside from intrinsic factors such as friction and electrical resistance (that analog computers don't address de facto either): information erasure. It turns out, from *Landauer's principle* that there is an irreducible energy cost associated with information erasure at the bit level and since the logic gates used in conventional CPUs aren't reversible/injective (simply for cardinality reasons), these costs add up... before you even start taking intrinsic resistance into account. Even a perfectly superconducting, ultrastable conventional CPU would need to dissipate tens of Kilowatts of heat to perform a ZettaFLOPS of compute. In contrast, there are reversible nanomechanical "chips" designs that could theoretically perform at the Zettascale while barely dissipating a single Watt of heat... less energy than a typical smartwatch ! More than analog computing, it's the field of reversible computing that we should keep an eye on.
There's one very, very important word in your post that sums up everything you've said. That word is "theoretically". In theory lots of things can happen. However, reality is somewhat different.
I have recently built an analogue computer using nothing more than some MOSFETs OP-Amps resistors and capacitors. The FET's were only used for multiplexing. Also there is an easy way of doing this. Something simple like a Z80 based computer with FRAM which is ram that does not need power to retain data. The Z80 writes to 8 latches that multiplex in the type of node needed EG sample and hold, multiplier, adder, subtractor, integrator, differentiator etc. That is just the first 3 bits. You then have 5 for weighting etc. That gives 32 possible values.
Im a music producer, and whats crazy is how weve moved way way back to trying to give our music an anlog feel. There are so many plugins nowadays designed to emulate the beautiful imperfections of analog. The more degraded the better, but the trick is, to still make it clean, while also immediately getting across that analog flavor.
The analog computer chip from mythic would have to be considered digital IMO if its converting the signals to digital and making determinations based on the digital signals. This is a digital chip. This is how regular digital chips already work. What you have here is an ASIC.
Not really. Of course it will have a digital interface since it needs to be programmed and operated through a connection to a regular definitely-not-analog computer, but the processing behind that interface is analog (at least based on what I've heard about them - I have no relation to or knowledge of their company so there's always the possibility that the sales pitches are leaving out some important details). Regular digital chips to not play around with arbitrary voltages or currents - that tends to make them not function (or worse, still function but not correctly). Basically its digital logic to prepare the system for a calculation, then analog to run the calculation, then digital again to read the result back. That's also how quantum computers work (except that middle part is qubit magic rather than analog circuitry). Its still an ASIC though I suppose, as that term doesn't really specify that the IC components on the must be strictly digital.
I can see these as low power processors that detect “possibly x” and wake up a a digital circuit to “verify it’s x” and “compute or act on x.” Similar to how our smartphones have specialty digital circuits that listen for speech and wake up higher functions when they detect attempts at vocal input.
That's not really how it works unfortunately. These things aren't detection machines - they're just do really cheap matrix multiplication, with the downside being that the result might not be entirely accurate. That's fine for AI because the matrices involved are representing neurons, and its often even beneficial for there to be some noise in the system as it functions as a source of non-linearity and "good" non-linearity is critical to making a NN be better than a "simple" calculation (mathematical simplicity that is - a completely linear NN reduces to a polynomial. Likely a very large polynomial that you still wouldn't want to compute by hand but in terms of its processing capabilities its no better than a relatively basic function. Introducing non-linearity fixes that problem and makes NNs far more powerful than any "simple" function could ever be). That said, (some) AIs absolutely are detection machines and if these can be used to make more efficient AIs then you indirectly get what you want somewhat cheaper.
I have always pondered the great audio debate myself, and although i am going off memories along I think i prefer vinyl. See when something is still analog it's existence seems to be in that of our physical world. When digital things feel too good or fake, being beyond the intention of what the human body was meant to absorb. Its the same difference between playing a rif on a guitar right in front of you (or playing it yourself), versus just listening to a cd of the sounds. When listening to a digital recording everything is on one wavelength it feels. The highs and lows just don't seem to truly reach out and touch our senses or our bodies in the same way that physically making the sound does. This is why I choose vinyl (Analog) over the Digital revolution. Maybe I am right, maybe I am wrong, but based on what I feel is what I wrote.
As much as I love analogue audio, in every single revolution of the Vinly, you lose the original quality of the recording. There's just no physical way to avoid sound degradation on a Vinyl. Digital recordings, in principle, can last forever, unchanged and in original quality.
@@GoldenEDM_2018 except its digital sounds, not physically real sounds, or so it feels. with non digi vinyls there may be slight degradation, but your body feels every frequency, every wave, rising and lowering with every beat or rhythm. Whereas with digital every sound hits your body one and the same, with no physical difference in the waves because they are programmed to all come out at the same level, whereas vinyls levels raise and lower.
@@YouGoneMad come on, vinyl has a lower resolution than CD, and like most people doesn't need even these fine details in audio, as all use lossy audio now, like MP3 and so on. Vinyl is a placebo, 95% blind test will fail to detect a difference.
@YouGoneMad if you're talking about the volume being the same thats just bad sound mixing, the only advantage that vinyl has is any sentimentality you have for it
A lot of the misconceptions about digital vs analog where vinyl is concerned, have largely been debunked by numerous audio experts. Also worth considering with the resurgence of Vinyl as "fashionable", that you are actually buying a digitally produced and mastered product with no analog in the chain at all.
There's analogue in the chain when the cutter is dragged across the lacquer master disc and at the other end when the stylus is dragged across the vinyl!
There are just misconceptions about analog and digital computers, too. It is analog at what I like to call the electrical engineering level. Some people think that they are completely different things.
@@Allyh1015 Good luck making that catch on! For 90% of the people out there, talking to an advanced LLM is indistinguishable from chatting with a friend, a really smart friend, mind you. So, the experience of interacting with AI feels a lot like interacting with a human. Not calling it AI is nothing more than a mere technicality for most people.
@@Allyh1015 Can we please agree to disagree? :-) AI is a rather big field, and its definition is still undecided, as it is growing rapidly. I would say statistics play a role in it, but once again, that’s why I want the creators of this video / channel to explain AI, as so much of this phrase is just sprinkled all over the many fields of life nowadays. Furthermore, I would like to add, that when talking about „AI“, I also often think that it’s statistics :-) so we can agree on that. I also looked up the short definition from IBM‘s website: Artificial intelligence leverages computers and machines to mimic the problem-solving and decision-making capabilities of the human mind. So, yeah, focus lies on mimicking human thinking/problem solving. Also, predictive analytics is a field of AI. Like I started the comment, AI is a broad field. And there are many areas to it, and some AI tools analyse and predict, while others simply might follow some specific rules. Don’t forget the „if-then“ clauses of Expert Systems“ :-)
My hearing was better when I was younger, so my memories of LPs as having better sound reproduction is biased by that. I remember the first CDs sounding better mainly because of no pop, hiss, scratch sounds and for the same reason: roll off of my hearing at higher frequencies. Analog "computers" have benefits but analog electronics are more subject to temperature variances and noise (or inability to identify and suppress noise) but they run fast and are reliable. In 1984, I was sitting in controls class and the professor rolled in an analog computer and told us how obsolete they were and we would never fool with them. But the majority of the control loops on the planet were still being controlled by analog (electronic and pneumatic) PID controllers, simple analog computers.
Must be like 15ish years ago now, I converted an old guy's records to digital, then burned them to CDs, pops and static and all, he frickin loved it in his car
I'm sure they played better in his car than records would have. But I'll bet some of them have gone bad by now, while the records from which they were recorded probably still sound fine.
@@pcno2832 CDs aren't that short lived, besides, you can just do it again, or cut out the media and record MP3 or WMA or something from the vinyl player direct.
Very interesting update on analog computing which has been a struggle for decades. Thanks! However, I would strongly suspect that any method such as the one in this video that depends on ohm's law will be highly sensitive to temperature variations causing huge errors and unrepeatability much greater than the 5% they claim. That is one of the huge benefits of binary it gets rid of all nonlinearities and parasitic factors by slamming the node to the maximum minimum and thresholding everything in between.
What's interesting to me about digital moving forward is that qbits kind of break that mold, at least before the output stage. Sort of a weird mix between analog and digital in a metaphorical sense.
@@ezpk- Qubits break the mold but they don't have any zero or one state at any given time, and I don't think anyone can even imagine how to make a Turing machine out of them. I suspect they'll be at best a co-processor, like a GPU or an ALU, to a classical Turing machine, for many decades to come.
78s have a nice, long track for every second of sound. It spins so fast that many scratches are rendered out that would have been very loud at 33.3 rpm. They are also thicker than a 33.3 but are more brittle. If you ever do break a 78, crush it into small chunks and put them in a jar and then cover them with turpentine or acetone. It makes an amazing black paint that shines like a mirror.
I think you just scratched the surface of non-digital computing architectures. You could try looking for memristive computing, stochastic computing, photonic computing, magnetic junction computing, etc.
Digital can approach absolute accuracy without limit; increase the bits/sample ad infinitum. The limiting factor is noise and that applies to inputs to both a digital and analog system which has to process an analog signal. So minimizing noise in the low-level input signal as it approaches zero is the key and in the case of LIGO, quantum squeezing of light allowed it to get that much closer to the ideal. An important note: when using resistance-controlled currents or voltages, it should be remembered that resistors/resistance is inherently noisy, no voltage applied. The tradeoff is that the lower the resistance used, the higher the power consumption but the higher the resistances used, the more analog noise is generated. Computing is really a horses for courses game; one uses the methodology most suited to the problem being solved.
same principle as optical matrix computers, the massively parallel ones, not the logical gate ones, both work anyways and are fast, the best optical logic gate computers use waveguide with x-ray photons in the gated pipe system.
Back in the early 90's, I would copy records to my computer in 46kHz MIDI format. Each song was huge. I had to have a server to store my small collection of 20 to 30 records.
You lost me in parts of this but the last bit about use cases made it all clear. Yep, I think like everything else, there needs to be a balance -- between food & exercise, wind & solar, analog & digital.😊
It's a common mistake to say that digital audio is an approximation of the analogue sound waves. Anything recorded above the Nyquist value is essentially perfect. Digital audio is not little squares and rectangles. The poorer sound of some early digital equipment was due to bad DACs and the pre-amp needed to get the signal to line level. Some people prefer vinyl because the compression makes the mid-range stand out.
I see a way analog computers can compliment digital computers, game consule AI. Basic game logic could be stored and use digital computing, while enemy and NPC logic could use analog logic to evolve the game over time.
Vinyl has its tradeoffs, first, the waveform has to be compressed, as the full dynamic range cannot fit into the grooves, so while you might be getting (in some cases) a more accurate representation of the original sound wave on a vinyl, if it was recorded right and the record is in perfect condition, the dynamic range is reduced, so you might not be getting what the original recording sounded like even then... With digital, it's possible to get to such high bitrates and sample rates that our ears and brains simply cannot begin to tell the difference from the original sound wave.
... though the irony is that when we moved to digital we also moved to greater dynamic compression than was used for analogue recordings and vinyl pressings (for reasons of musical taste/fashion/style). So yes, the digital formats can potentially represent the higher dynamic range of a musical performance than vinyl but in practice they use less of that potential in most commercial music recordings.
Exact computations are not necessary for most applications. Our eyesight & hearing is not exact, but speed to process massive quantities of divergent information is more important. Reducing power to perform high speed computations in near exact results is good enough for smart bombs 💣 , smart rockets 🚀, smart artillery, smart bullets. Similar needs for autonomic driving systems could be an excellent application… conserving more power for expanding distance of the vehicle. Honestly, this can change a lot in military & civilian applications.
I think these analog chips will make their way in devices. Regarding the question between vinyl and digital, definitely vinyl for me. The experience is so much more than just the music. It’s the decision process on what to listen next, the album art, and as you have to be careful with it, it feels valuable
"and as you have to be careful with it, it feels valuable." This kind of nonsense just makes me want to go on a rampage, because there is absolutely nothing about music recorded on vinyl that comes close to the repeatability you get with digital. If it feels more valuable because every time you drag a diamond through that vinyl, it loses quality, and you're happy with that, then by all means, DON'T listen to your vinyl records because of how careful you have to be with them.
@@BrightBlueJim You're exactly right. Back in the day, I found it exasperating to put on a brand new factory-sealed album and hear scratches *the very first time I played it*. At the time, it was explained to me that this was because North American record plants only pressed the albums for 1 second while those in Japan would press them for 30 seconds. This was supposed to account for the great superiority of Japanese pressings. Give me a CD any time!
its like mechanical watches, its vastly inferior but it feels more human because its not digital perfection and you can see and touch it@@BrightBlueJim
My position on digital versus analog recordings is that they are just that, recordings. I prefer to listen to live music even if it’s by amateurs. A recording will always be an estimation of the live performance. Analog is still an estimation.
During a electric/electronic class, I was asked about high impedance with transistors use. I said a three answer computer would be much more useful than a two answer. The teacher said it would require a whole redo of programing and not be cost effective. Both were correct.
FPAAs (Field Programmable Analog Arrays - Analog FPGAs more or less) are reconfigurable analog logic devices that can be used for prototyping, changing resistors won't be too complicated. - haven't read much on how these work though 🤔 This makes me even more excited 🤩 for starting my degree in electronics engineering!
Vinyl isn't really analog anymore. It's recorded in a digital based media, then translated back to analog. Not the same as storing the data on tapes, then reproducing to vinyl.
Vinyl is a whole can of compromises to have an acceptable reproduction of a recording on media. You then boosted bass to compensate when listening for example. Past a point, bass frequencies would make the needle hop in some bad way I understand. I lived in the day when you always boosted bass :) You might like hearing the result of those compromises. So yeah, not a good example. :)
Has been this way since the 1980's, some people were doing it in the 70's, all stuff is recorded on digital audio tapes or other digital storage today.
Vinyl has a digital master track encoded into it but the medium and the way music is produced from it is definitely analogue. It's true it wasn't made with a 100% analog signal chain from start to finish but who gives a shit?
Analog is “random or imprecise digital” just that. Forget the myth. The granularity of an analog picture is their “pixel” only randomized. The same about the fidelity and noise to signal ratio, on an analog recording (not to mention mechanical artifacts). With the added problem that each copy adds another layer of imprecision or “pixelation”. Edit: Remember analog computing is only a cool name for more bit states than 0 or 1 determined (usually) by voltage levels.
Yes, analog is just lossy compression. Though floating-point calculations are already a bit imprecise even on digital computers, and GPUs can be even less accurate because in real-time graphics tiny errors don't really matter and AI doesn't need perfect precision either (8-bit quantization in AI models is kind of the norm because it reduces memory usage by ~75% while only introducing around 1-2% deviation to the results)
Nope, they won't change anything. There is a reason why pretty much all computers are digital, and that reason is noise. This is pretty much the first thing you learn when you start with digital electronics. Digital computers are just orders of magnitude easier to build for our noisy environment. Or like analog IC designer Bob Widlar said, "every idiot can count to one". For example, quantum computers are full analog and they are so incredibly fiddly, that we haven't seen a useful one yet. According to Wikipedia the largest factored number on a quantum computer using Shor's algorithm is still 21.
Analog is a better representation of the real world. We expect the world to be a competition between sources. An environment that contains no sources is unnatural. It may be easier to conduct science by filtering out the distractions but it does not contribute to the understanding of the impact of the interrelationships.
Not all analog computers are loud. I saw a video showing how you can use electricity inside an analog computer that is using the voltage as the value and there they showed how simple multiplication was which is a large part of AI computation so while not big in usual use cases they are good for AI
Many great points, but some applications are tolerant of noise and the speed and energy savings in analog computing could far outweigh dealing with noise. Computer vision prossesing for starters. Speech to text could do a significant amount of processing before digitizing is nessesary. Analog computing would not change much, except in a few applications superseding digital processing, but it will enabling more rapid advancement in those areas. In nural networks with weights that are currently digitized, switching to analog weights is far far far far more energy efficent and far far far far far more speedy. Memresistors may have significant aplication in analog memory with far greater storage capacity and speed in neural networks. In a neutral network, a weight with a noise between 0.998 and 1.002 can be insignificant enough to justify analog computing benifits in an aplication.
WW-II naval ships, used well into the the1970s, used mechanical analog computers for navigation. All automatic transmission, prior to computer controls, used fluidic analog computers (the valve body) to upshift and downshift.
The inaccuracy can potentially be a good thing for general AI. With sufficient checks and balances via a few extra neural networks, errors can be seen by the computer, that's how it works with people. We gain experience and we learn how to spot our errors. Plus, error creates variation, variation is creativity... the foundation of art and culture.
@@evertchin Think about image recognition. Training the AI with a couple % error rate could improve the ability of the AI to detect noisy or difficult images, as it was trained from the start with error-ridden data.
@@kindlin That's how AI has been used for decades. All trained AI models have intentional inaccuracy to prevent an issue called overfitting. It's one of the basic things taught in every AI course because if you don't do it the AI model is near unusable. But you don't need analog computers for that, all that's necessary in most cases is to not run the training for too long.
I'm 70 and grew up with analogue vinyl and magnetic audio. I loved them, of course...they were what existed, and high end equipment sounds fabulous!! At one point I owned several thousand vinyl LPs and hundreds of cassettes. They had their weak points, surface noise, tape hiss, and a far lower dynamic range. I was easily persuaded to the dark side (digital) four decades ago, when my sense of hearing was far better than I can claim today. Even then, around 30 years of age, I could not tell the difference, aside from the advantages of digital. EDIT 2 weeks later: Once beyond the discussion of analogue vs. digital audio, things required a few more watts to take in. There's a lot I didn't absorb in the first run.
The only advantage CDs had over vinyl and tape was the lack of scratch/pop/click noise or tape hiss. As for the sound, at 44.1kHz / 16 bits it simply lacked resolution. Today's high resolution audio is much better sounding and is close enough for most people to not be able to tell digital from analog in listening tests.
Seems like someone is trying to leverage the hype of AI training with Analog Computing(which has nothing to do with AI training to start with🤔) for a reason, the usual suspects😂. Things will fall apart, once the fad winds down, or takes on a different route? that could use digital computing effectively!.❤👍
oops... One meter does not equal one foot, or even similar to one foot. 12:39 One yard, yes... but one foot? Nope... Not sure if that was intentional, but sounds like a huge error.
Exactly, "analog computers about to takeover", then proceeds to talk about what digital is, what NNs are, what ohms law is and finally why we use digital, Like where the fuck was the content described in the title?
What about wear though? Digital only needs to listen to on and off, but analog has a whole range. We already see similar in ssds adding more data by increasing the sensitivity to voltage. As the cell wears down, it becomes harder to measure the difference in qlc vs tlc or mlc.
It actually matters less in analog, as you don't need to measure differences - you're using the whole range. If you measure 4.83 volts then your answer is simply "4.83". You're not comparing 4.83v against +5v and -5v (or whatever values your circuit uses) and trying to match the closest as you would do if you were wanting to distinguish 0 from 1 in a digital circuit (or from a 2 or a 3 in multi-level cells).
There are kit robots with very simple analog circuits out there. I saw a video for one years ago that was built with paper clip type pieces of metal for legs the person filming demonstrated that you could damage a couple of the legs and the robot would "learn" how to walk without them. Seriously impressive emergent behavior coming from something that only had the equivalent of a few transistors worth of logic to work with.
Current hardware had an explosion in errors due to increasingly smaller hardware reducing error margins. There was a paper DDR4 increased error rates by 5.5x to DDR3 and then they slapped on die ecc on consumer ddr5 just to deal with it. Maybe they are saying "if we cant contain errors in digital and need checks anyway we might as well use analog"
Back in 1968 I used an analog computer to simulate a vehicle chassis. Position, acceleration, and jerk are related by differential equations, and the wheels are interconnected too. But a simple analog computer with maybe a dozen vacuum tubes could do differential calculus instantly, and you could change the spring constant or shock damping in real time to see how it changed the handling. To do one simulation on a computer I had to punch a whole stack of cards and wait for it to run on the mainframe ... overnight.
Any modern vinyl record would be mixed to have lower frequencies in the middle channel (not far apart in stereo spectrum) to prevent track skip. So... the FLAC and the vinyl copy of the same song will be significantly different anyway.
On the topic of digital vs vinyl, I think a property EQ’ed uncompressed 24 bit 96khz signal, heck probably even a little lower, is going to be physically indistinguishable from a vinyl recording. If you think about how speakers actually move, speaker motors are not infinitely fast. This means that the “jaggyness” that inherently exists within a digital signal compared to it’s analog counterpart is going to be inherently smoothed out by physics. Heck, I’m sure even the DAC in most systems, even the high end ones exhibit this kind of signal smoothing electricity themselves in the conversion from digital to analog. It’s not a digital signal that makes a sound inherently inferior, it’s the compression
What you describe as the inherent "jaggyness" of a digital signal is a very common misconception ... it seems like it must be the case when you look at some graphs and it fits with our intuitive expectation ... but it's wrong. If you do some research on the subject you will understand why. In short, the jaggy graph represents sample points but it does not represent the sound wave that is produced ... which is a perfectly smooth sine wave. Yes, it's hard to get your head around this at first (I was the same)! Now let's move onto compression ... are you talking about dynamic compression or data compression? If you are talking about data compression then the digital signal used for CD is uncompressed ... so no problem there. Digital recording doesn't use data compression either ... but ... streaming services generally do use data compression. They need to save data bandwidth because of costs for them as suppliers and for you receiving their product via your limited bandwidth data connection. If they use significant "lossy" data compression then the audio can be noticeably degraded. If they stick to "lossless" data compression then you end up with no degradation (but the data rate will need to be much higher and more expensive to them and to you). Quite separately we have the issue of dynamic compression ... it would be better if there was a completely different word for compression in this instance because it has nothing directly to do with data compression. In fact you can have a highly compressed (dynamically) track that is delivered to you via a totally uncompressed digital signal ... in fact that is very very common in modern music recordings! The element of a digital file that describes the potential for dynamic range is the bit rate (e.g. 16 bit 24 bit etc). It enables very quiet sounds and very loud sounds without you hearing any background noise. The advantage of 24 bit is mostly at the point of recording as it gives you the ability to match sound levels without bringing up the noise floor. Once the track has been mixed and mastered the advantage of the higher bit rates pretty much disappears ... because a final 16 bit copy is going to have plenty of available dynamic range such that it isn't going to be a restriction to your listener.
I saw the Veratisium vid when it came out and have been excited for the technology ever since. Digital computation could run up against significant limits and this looks like an interesting work around.
The fundamental issue with this is that what you're describing is just the analog version of ASICs, but even more narrowly focused and harder to design reliably. Given how rarely we already use ASICs even in dedicated high value applications, I see this having very limited practical use unfortunately, as it'll just never be able to reach economies of scale since the requirements vary too much in every application to reuse.
Just like with quantum computers, and in contrast to popular misconceptions, the goal is not to fully replace digital computers, but rather find more optimal means of solving problems. It's always exciting to see that there could be a different computer structure that does it better than classic digital computers, and a giant breakthrough in computing.
I started out my career in electronics design, led me a job with a DOD contractor working in a computer simulation labs, many many computers doing also hardware in teh loop. this required real time computing, taking several computers in a room, digital was the host computers to control other Analog hybrid computers, using voltages, op amps, multipliers, DAC/ADC converters. Analog at the time was the only way to compute real time as a voltage change was almost instantaneous. and input/outputs directly coupled to test hardware that could be switched in the simulation loop the hardware would think was operating in real time. Later came high speed digitial using ECL, which was actually digital circuits that operated in neither on or off state, but as amplifiers biased partially on and only change intermediate voltage levels to get the speed which normal digital circuits that were all off and all on created too much harmonics to get the speed needed, these were power hungry special designed beasts. Outside of any real commercial potential for quantum computing coming on line I would think a new hybrid approach to using analog in the computing mix again could break through current limitations on speed.
Check out the Roborock S8 Pro Ultra Robot Vacuum Today! amzn.to/3v5Yftf
I see you advertising that robotic vacuum cleaner…. My question is does this company sell my information about my house and floor layout with the information to collect from cleaning my house? Just a question if you haven’t looked into it, you’re probably should. What are the biggest markets today and growing is data?
Neural networks, please.
George Davis
When the roborock can pick up paper and put it in the recycling, pick up discarded socks and put them in the laundry, and do so without nagging me for untidiness, then I'll be interested.😊
there has to be some betrayal and revenge. people are involved. on analog vs digital, i saw a digital and analog paperclip-legged walker. the only change was bending one of the legs so it didn't touch the ground. the analog handled it much better than the digital.
analog can't take over everything, but it's underutilized.
what is wrong with the voice sir.
My first experience of digital computing was in 1962 when the U.S. Navy trained me for their first three ships at sea to be equipped with them. While those early computers were much smaller than an IBM mainframe of the time, they were the size of large double door refrigerators, and needed chilled water to cool the processors. Energy costs were high despite the transistors they used, and processing times were much slower than the simple digital devices we use today. The digital computers were used as an air traffic control system, for combat and it fed some of the information about potential targets for our missiles to the fire control technicians on a different set of display consoles.
Their computers and consoles were analog. At that time the speed of their analog resolvers and differentiators was superior to any digital simulation our computers could produce. But they were constantly having to fine tune the analog elements to their system to maintain accuracy. As you note, digital can be very precise and repeatable.
Hey, I was fire Control when I was in the Navy in the 1990’s. Your comment was really great and brought back some fun memories for me.
Some serious cool stuff 👁️👃👁️🙏
that's useful information.
The processing power of this little old Galaxy 9 cell phone is quite the Quantum Leap in technology isn't it ? I remember playing T.V Pong on a black and white T.V 40 years ago. Now we're colliding particles ? Come on. Men just stopped using bows and arrows to rule the world and now we have "Rods from God" ? Perhaps C.A.R.A.T [Commercial Applications for Recovered Alien Technology] isn't a mythical organization....after all.
When I toured the USS Misouri they told us that when they took the ship out of mothballs for the Iraq war the dig computers were not as accurate as the original annalog computers, so they re installed the annologs to aim her 16 inch guns. Really interesting.
I started my career working on analogue computers used in industrial and manufacturing environments. They were mostly used for controlling motors and used proportional, integration and differential calculations at very high speed. They worked well in environments that were not computer friendly and were very robust and reliable. That was 40 years ago and they lasted for years because they just worked. I think the challenge was that to program them you needed to understand electronics and mathematics and that isn’t so necessary with modern digital computers.
Some of those old controllers were pneumatic. PID controllers were used for a lot of valves.
Yep, when I was taking electronics at vocational school over 40 years, and our instructor talked about analog computers to include computers that used RC time constants to solve log calculations.
Now chat GPT is doing code and now people are gonna have to learn electronics.
I personally am happy with analog. It's like reality vs. artificiality with digital vs analog.
LOl we moved away from analogue for a very valid reason, moving parts because heat and wear eventually caused them to fail. Could someone explain what this guy is saying, as there is way too much filler.
@@andrewbrown6578The exact operation of the new analog CPUs is proprietary, there was a recent video about them. Hold on
Having grown up in the golden age of record players I fall on the side of digital. After a single play on a turntable a record will lose that frequency response curve. It really doesn't matter how expensive the turntable the wear and tear on the record is a function of the design. A physical needle riding grooves wears a bit of those grooves away with each play of the record. The quality of the digital recording is there forever.
A bigger issue with vinyl records is that high frequency response diminishes the closer you get to the center of the record, since the circumference (and hence, linear velocity) decreases. Essentially, you're trying to fit the same amount of information into an increasingly smaller area, resulting in decreased resolution. Inner groove distortion, as it's called, can be mitigated to a large extent with advanced (i.e. expensive) stylus profiles, such as Shibata or Microlinear, as opposed to the more conventional conical or elliptical styli found on most setups, where distortion will be noticeable, more or less. But even then, the fidelity comes nowhere close to that of even then cheapest DAC.
@@Hellcommander245 I just gave the simple explanation
A few decades ago my local newspaper ran a story about an audiophile. Because of the degradation you mentioned, he played his vinyl records once - to record to open reel tape. He listened to the open reel tape and archived the vinyl.
When CDs were new, most were created from analog multitrack masters. There were stories of classic albums suddenly revealing studio air conditioning noise on CDs, not heard on vinyl.
He's talking about a specific use case where you don't need the accuracy of digital.
The data on digital medium degrades too.
I'm a trained singer. The must authentic and believable representation of a human voice I have ever heard was a recording of Gigli played on an old 78 record played on a wind up gramophone with a horn. You could actually feel the voice like you can at a live performance.
It is enriching and pleasing to know this.
I wonder if that's because a horn is so resonant? Speakers basically have zero resonance.
@@LagMasterSam Could be? As a musician I've just noticed that over the years the 'quality' has improved with technology (no hisses or crackles, better microphone positioning and balancing etc.), but the issue remains that digital is a synthesis of analogue, and I'm not sure whether you'd ever get there no matter how high you get the resolution. I wonder if photographers feel the same about digital photography?
Love learning about diverse computing methods. Between analog, digital, and quantum computers, we'll be able to solve problems so much better than ever before!
New ways to cook the books. 😅😂😉
There's even more types of computing. Check out research into neuromorphic & organic computing
@@galopeian I compute with the roulette table
@@2148aa life's a gamble
you know, in 1974, researchers had developed an analog matrix method that could recognise the alphabetic and numeric characters, individually, with a very high success rate, and i remember being surprised at thesmall amount of analog hardware to perform the job, once learned/programmed...and, even then, the analog methods were very fast and very low power, per "recognition"
That’s fascinating. As someone born into the digital age I feel I’ve missed out on a lot. Especially as a hobbyist electronics/pcb builder
@@phillies4eva there's no time like the present for getting into a new hobby. Anything that you missed can still be created by you today.
The first scanner for character recognition was used in the 1800’s for sending morse code on the telegraph.
I had never heard of this until I started working for Xerox in 2007.
@@jasonrubik exactly 💯
I just realized that analog computers with a digital nanny to adjust for drift or runaway errors is how the fuel control unit in many modern jet engines operates. They use an analog hydromechanical computers to calculate and dispense the fuel as required, and a digital computer to adjust things on the fly to keep everything optimized for the best balance of fuel economy, longevity, power, and emissions.
Many small aircraft still use what is essentially a purely analog hydromechanical computer to control the fuel flow to the engine. The fuel injection on these piston engines uses inputs such as throttle position, fuel lever position, airflow, RPM, and air pressure to compute and meter out the fuel necessary for those parameters. Every annual or 100 hour inspection includes checking the fuel mixture at high and low power settings and adjusting as needed to keep the fuel system output correct.
I had a little trade school on automatic transmissions of the 1980s and before. I always asserted that the valve body in one of those transmissions was essentially an analog computer that operated hydraulically instead of electronically. I still assert that.
I remember using mechanical computers in the Navy that measured speed of targets for our Fire Control systems. Those computers wear far faster than anything around at the time.
I like vinyl, but other analog particles get in the way, like dust, unfortunately and cause a crackling sound. I think I would prefer digital sound at 24-bit, but if there was a way to get clean sound from analog audio, I would definitely prefer this.
The whole point of this is that for AI training they don't need a clean signal. A somewhat noisy value is good enough, just like vinyl gives you a good enough analog of the music encoded into it.
Yeah, Digital was never truly Digital you know, it was always just some conventions about analogue voltage levels(TTL/CMOS), and in a usecase like this, storing neuron BIAS values "vertically", we're basically removing those conventions and unleashing the full potential that every "digital Bit" had in the 1st place. LOVED this video... it suited my old school analogue brain really well 🙂
digital still uses real world transistors that are not binary but the logic is. inaryl. analog can't run software as in binary code compiled from source, in an analog computer it's just a fast, efficient and approximate parallel computation. when computing power and efficiency matter more than precision such as executing cheaply a neural network that has already be trained. an analog neural network is just baked in "intelligence" that cant learn anything
Not going anywhere much., Digital reigns supreme.
@@rogerphelps9939 not arguing there, Digital wont ever go away, but specifically for AI model storage i see a huge upside if they manage to save hundredfold on power and space, because that also increases reading speed. The idea is to utilize the untapped capacity within each byte, surely there are costs and caveats that comes into that, and that's why you are correct in your "digital supremacy" statement !
Great content today and always. I am somewhat of a music snob and grew up in the analogue era. I embraced the digital movement in music as it was introduced and find digital very satisfying. My ears can only hear so much frequencies as many of us baby boomers grew up in the rock n roll era . Therefore as close as digital simulation is getting to analogue it is good enough for a large percentage of the population's ears. Thanks again and keep the great content coming
You're mixing up being "digital" with the ability to generate high or low frequency sounds. Low sample rate devices can still capture and reproduce those same frequencies, but the lower the sample rate, the more data is lost between samples. A lot of digital audio was recorded from analogue sources, but it's the sample rate (and any kind of "lossy" data compression) that affects the quality. The higher the sample rate, the greater the amount of data, and hence the requirement for faster hardware to play it back.
When I was in the Navy, the fire control systems where analog; they used syncros and servos to compute in real-time the solution to fire projectiles onto targets.
There's nothing like the clicks, crackles and pops on your favourite track on vinyl.
Great video. Nice to hear more on analog chips and how they are important. Being a shareholder in Spectra7 active copper cable running on 800G analog chips, this explains a great deal how it’s used. Keep it up
It's not going to replace digital but the integration and use as a tool with digital is great, just like your example for runaway variation. Analogue can do the heavy lifting and digital spot checks.
From what I've read, Mythic's tech isn't really quite analog. One source reported 256 descript gate voltages for each transistor storing a value. This is no doubt closer to analog than digital binary states are, but the values are still quantized. Still, it's an interesting use of flash technology. I guess it also means the data can still be backed up and transmitted digitally.
While you are correct that the floating gate transistors can only be controlled to some precision of discrete gate voltages (even sometimes 4-bit [16 values] in mythic's case), this is not where the "analog" comes from. In fact, these stored values can vary during operation by a significant amount from a variety of factors, so they are even less precise. Moreso the term "analog" comes from the method of not using a bistable device as the main method of computation. Rather, the physical dynamics are used as a mode of computation. From an energy perspective, this is not ideal (since digital to analog or analog to digital conversion takes a lot of energy,) but the main energy sink, the actual computation, is done very efficiently.
Unlike "analog" computers of old, such as those seen in navy targeting computers, the modern (near term) search for analog computers is to find very efficient ways to compute with digital data. For this reason, the term "mixed-signal" is commonly used in academia to refer to this approach. There are certainly methods of computing that are all analog, mainly classic analog computers and biological brains, but these are either too difficult to scale (the former) or beyond our current technology (the latter).
TLDR: Digital in, compute with analog, digital out.
0:18 The guy in the middle of the stock video clip was pretending to use an iPad while the screen was covered up with it's case.
I do believe that analogue will have a big role in future computing. Though I don’t think it will be a replacement for the core computing of algorithms, it will be excellent for sensory functions and identification. The world is fluid and that’s why analogue will be best to interface with it. Eventually the analogue technology will get better and the interface between these devices and a digital computer will become fluid.
The struggle will be to identify the categories between two objects (this is the learning process). Technically everything is different so making an identification is highly complex but when we can create a sensors that identifies with enough points (like 20 for a fingerprint or retina scan OR 30 to identify a species) then we will have a powerful system.
For example, if an analogue device can identify from kingdom to species then send that information to the digital system then we have made a huge increase in efficiency. The real benefit will come from “fingerprints”that the human mind currently doesn’t utilize. The analogue can process sounds we can’t hear, temperatures we can’t see, smells, tastes, vibrations, etc.
Bottom line: If you build the sensor correctly then you can download indexed identification data that will make observation incredibly fast. It will also be able to identify new identities that were previously unknown. The downside is that a computer will be able to make an assessment about a person from data so rapidly that it can predict / anticipate behavior immediately. The only way to protect privacy will be through the creation of chaos and noise!
This explanation is exactly how electronic devices have worked for decades, the analogue circuits manage the interface to the real world, and the digital circuits do the actual repeatable, reliable work
When I was in college low these many decades ago, we used individually interconnected analog boxes to solve differential equations…
New vinyl plates are an analogue representation of a digital recording, meaning they do not sound as analogue as people make them out to be.
I love vinyl records, since I started playing my dad's collection at the tender age of 7 or 8 in the early 80s.
However, I fully embrace digital music as well. Being able to carry a lot of music around in my phone is wild.
There is something missing when we can’t open an album & gaze at the large form factor artwork on the album covers while listening to music.
You need about 128khz sampling rate to approach analog vinyl record sound quality, as well as equivalent DAC and inline digital chips... then it's only approaching analog sound quality that we had in the 1950's.
@@VenturaIT No, you don't. All vinyls of the last decades are based on the same digital material you get on CDs, just with what's effectively lossy compression artifacts. Full analog audio was so terrible in comparison that artists were insisting on digital processing for their new albums as soon as the tech for digital vinyls was introduced.
I like the sound of vinyls too but in reality it's just a low-tech DAC.
@@SaHaRaSquadIt's more complicated than that. There have been studies done and they realized that to fool the human ear with digital it would need to be at least 128khz sampling rate... "artists" don't know much about these topics... even today many top recording studios use analog tape for the master because it's still superior. Digital as utilized currently is inferior to master quality analog in almost all aspects.
"In recent years, digital masters have become usual, although analog masters-such as audio tapes-are still being used by the manufacturing industry, particularly by a few engineers who specialize in analog mastering. Magnetic tape was commonly used to create master copies." - Wikipedia
"After two full decades in which digital formats have been arguably the primary music recording methodology, analog has made a comeback. Recent recordings from John Mellencamp, Elton John and Leon Russell, Taylor Swift, and the Secret Sisters were recorded completely on analog tape." - Grammy
The trick if you buy records is to buy only the ones that are fully analog through the entire process... not digital that has been put onto vinyl.
The sound waves coming out of the speakers of the TV I'm watching you on are analog. Sound is analog because it inherently exists as continuous, varying pressure waves in a medium like air. These variations in pressure are not segmented into distinct units; instead, they form a seamless and unbroken flow of fluctuations. This means that sound can take on any value within its range, not just specific, predetermined levels. This continuous nature of sound contrasts with digital signals, where values are represented by discrete, separate steps. The analog characteristic of sound allows for an infinite gradation of tones and volumes, leading to its smooth and uninterrupted quality. This is why sound, in its natural form, is fundamentally analog.
The light from the video is both analog and digital.
Imagine light as being like a smooth, flowing wave (that's the analog part) and also like a bunch of tiny, indivisible packets (like digital bits in a computer game). These tiny packets are called photons. When we look at light as a wave, it seems to change smoothly, like the colors in a rainbow. But when we look closer, we see that light is actually made up of these photons, each like a tiny dot of light, which you can't break down into smaller parts. So, light is special because it can be both a smooth wave and a collection of these tiny, distinct dots at the same time.
Voltages in the equation V=IR (Ohm's Law) are analog because they represent continuous values in an electrical circuit. In this context, 'V' stands for voltage, 'I' for current, and 'R' for resistance.
The analog nature of voltage means that within a circuit, it can take on a continuous range of values, not just specific, set levels. For example, the voltage across a resistor can be any value within the limits of the circuit's design, smoothly varying as the current (I) or resistance (R) changes. This is different from a digital signal, which would have distinct, separate levels (like 'on' or 'off').
In essence, the relationship expressed by V=IR operates in a continuous, fluid manner, where adjustments to current or resistance lead to proportional and uninterrupted changes in voltage, characteristic of analog behavior.
Most interesting comment in this thread has the least likes. Go figure. Thanks for this, especially agree about the sound part.
Analog computing definitely has a place. I don't think it will change mainstream computing, but certainly for application specific uses as you described, it could be a game changer. Data storage seems to be another beneficial area since currently we would need to use 8 bytes per floating point number. With an analog solution a single analog data point could represent that number.
You can represent a floating point number using any number of bytes. The format used to represent the data is decided by the required range of numbers and the accuracy needed.
you don't need an analog computer to read analog information. we already do it with microphones
@@heliusuniverse7460 yeah it's all a matter of precision. The difference between analog and digital is infinite precision vs discrete values. Remember when CD's first came out, and all of the audiophiles refused to listen to them because they preferred their records? This is why, the transition to digital caused a loss of precision. To many it was good enough, but for some use cases it was insufficient. This will be the same. For most things digital is fine, but there are some use cases where infinite precision is a boon.
@@BruceWayne15325 It has always been myth that vinyl is superior to CDs in terms of perceived audio quality under controlled, blind conditions. The only reasonable explanation for it having some truth was that the old record players had an overall superior sound system than the early CD players. With an otherwise equal quality sound system, CDs will produce higher quality sound just from the fact that there's substantially less noise.
I think you mean accuracy, and not precision. Analog can be more accurate than digital, but it can’t ever be more precise. Accuracy is about average closeness to the original, while precision is how reproducible the outcome is. Baring some hardware error or anomaly digital is 100% precise because it always gives the exact same result. But even if analog can potentially be more accurate, there is a point where enough digital representation surpasses the noise that's inherent in analog to the extent that digital is also more accurate.
I think analog computers might be a little more common than we're thinking, depending on how common AI acceleration hardware becomes.
I listen to plenty of digital music all the time: while Im working, driving, doing stuff around the house or gaming. However, when I want to sit down and dedicate my time to listening to an album I much prefer the experience of playing vinyl records. The entire ritual of removing the record from the sleeve, placing it on the platter, brushing it and dropping the needle then settling in to look over the big, beautiful presentation of the album art and liner notes while I listen and enjoy a coffee or whiskey is extremely satisfying and relaxing, almost therapeutic in a meditative way. While there are arguments about sound quality that can be made to support either digital or vinyl, at the end of the day it's the unique experience of engaging with vinyl as a medium that made me, and has kept me, a fan of vinyl records.
The Roborock is amazing when it works. Dealing with their customer service is a whole different story. Some of the accounts in the Amazon reviews are really something, and definitely a must read before you buy.
Analog computers could make a come back but I think the most likely analog system to do it would be an optical one using nonlinear optics. The thing about optics is that a step like doing an FFT and be done instantly. An image can be filtered or processed to detect some object in practically zero time.
Fast by nature, Fast by name.
I'll get me coat.
Analog computers especially for AI weight algorithms storage where absolute accuracy is not critical, it could be a useful addition, aka a niche application. However, digital has a great strength, accuracy and repeatability.
So what you're saying is that the Orwellian, dystopic surveillance society is going to be cheaper than ever for massive corporations to deploy.
No. Your ability to fight back won’t be contingent on access to massive power grids or capital intensive silicon chips.
@@williamcaseywells604
Oh. Good to know I can fight military drones with sticks and stones.
Sounds like a very efficient & energy efficient system.
My curiosity about alternative computing systems is based on the need for resilience vs threats to digital systems such as: cyber attacks, EMP attacks, solar flares, etc.
So in line with this:
- Do you see analog computing being less vulnerable to the aforementioned threats?
- Raising it a notch higher, do you see analog computing being further developed to be a good system to have as a back-up in case of any of the 3 threats (or other threats of similar nature) takes place?
It's a chip.
If it's unhardened, like most chips, it's just as susceptible to Solar activity, supply spikes, &etc. as any other unhardened chip.
@@DickHolman Thank you
At Oregon State University in the 60s we started by wiring logic boards to do tasks to a deck of punched cards. We later moved to an IBM1620 that was a mix of analog (had dials to set for many engineering functions) and digital (could run an early version of Fortran II). Used tons of power compared to the CDC 3300 that replaced it. Now that digital power requirements are dropping for both CPUs and memory I can see Analog Computers used for a specific tasks but not for General Purpose Computing. One of the issues with Analog Computing is numbers like Pi or e out to multiple decimal points required for many scientific calculations and transforms like Fourier and LaPlas.
No. Just not true. The IBM 1620 was 100% digital. Yes, it contained may analog components. When you get down to the individual transistors, they are all analog, even in fully digital computers. But the circuits did not operate on analog values. The distinction between analog and digital computers is in how numbers are represented. If numbers are represented as specific states, whether there are two states in a binary digital computer, or four states or more in a modern flash memory cell, they are still digital.
Are you sure that the 1620 wasn't a 1710--the process control version of the hardware? Had DAC and ADC peripherals. Still instruction-compatible; just a few added features and instructions. In the 60's, the notion of a "hybrid" computer was all the rage.
Dials have same digital nature as selectors on washing machines, I suppose. It can be replaced with a line of push buttons, just was more convenient.
@@sonickrnd My kitchen induction stovetop has rotary controls for individual element heat. However, what's used are simple rotary encoders, so that a dial position doesn't correspond to a setting--you have to look at the LED readout to see what heat level you're setting. Which, in my opinion, defeats the whole purpose of rotary controls. I have a blind sister-in-law who is absolutely useless with this kind of control--she has no idea what setting is being established. "Touch" controls are similarly useless to her without some sort of feedback. It seems that human engineering has taken a backward step with this sort of thing.
In WW2, the navy had analog computers that used gears to set range and speed of targets and how to aim the guns. They were quite accurate when used right.
Digital audio is better than those graphs suggest. Any continuous signal can be decomposed into a sum of sine waves of different frequencies and amplitudes using Fourier analysis. According to the Nyquist-Shannon sampling theorem, any given bandwidth of frequencies can be perfectly represented by sampling at double the frequency bandwidth. So, yes, uncompressed digital audio can be equally good as the best vinyl vinyl (arguably better in the real world since vinyl cannot be made perfect). Thus lossless digital can really be lossless.
Laplace, Fourier and other transforms are used to convert between different domains.
A Fourier transform will convert even a square wave or square pulse into multiple sine wave components. In theory, if you could isolate those components and feed them all back through an analog summer, you would recreate the original waveform.
Yes, the representation of how digital audio works was terrible. Watch videos by Monty Montgomery for the truth.
People who say that vinyl is a higher fidelity format than CDs or SACDs are extremely delusional. It's okay if you like noise added to your music but that's what's added, noise!
Another pathway to solve the energy efficiency issue would be classical reversible computing.
The main tricks used by the designs explored in this video have to do with the fact that you need less bits of informations to describe a physical system when accuracy isn't critical AND you spend much less energy when this less accurate memory is closer to the computing units (in proper neuromorphic computing, those tend to be the same, even).
Reversible computing directly goes after the fundamental cause of energy dissipation in digital systems (which, you have to remember, are necessary for faithful information transmission and general error-correction) aside from intrinsic factors such as friction and electrical resistance (that analog computers don't address de facto either): information erasure.
It turns out, from *Landauer's principle* that there is an irreducible energy cost associated with information erasure at the bit level and since the logic gates used in conventional CPUs aren't reversible/injective (simply for cardinality reasons), these costs add up... before you even start taking intrinsic resistance into account.
Even a perfectly superconducting, ultrastable conventional CPU would need to dissipate tens of Kilowatts of heat to perform a ZettaFLOPS of compute. In contrast, there are reversible nanomechanical "chips" designs that could theoretically perform at the Zettascale while barely dissipating a single Watt of heat... less energy than a typical smartwatch ! More than analog computing, it's the field of reversible computing that we should keep an eye on.
There's one very, very important word in your post that sums up everything you've said. That word is "theoretically". In theory lots of things can happen. However, reality is somewhat different.
I have recently built an analogue computer using nothing more than some MOSFETs OP-Amps resistors and capacitors. The FET's were only used for multiplexing. Also there is an easy way of doing this. Something simple like a Z80 based computer with FRAM which is ram that does not need power to retain data. The Z80 writes to 8 latches that multiplex in the type of node needed EG sample and hold, multiplier, adder, subtractor, integrator, differentiator etc. That is just the first 3 bits. You then have 5 for weighting etc. That gives 32 possible values.
That's impressive
Im a music producer, and whats crazy is how weve moved way way back to trying to give our music an anlog feel. There are so many plugins nowadays designed to emulate the beautiful imperfections of analog. The more degraded the better, but the trick is, to still make it clean, while also immediately getting across that analog flavor.
The analog computer chip from mythic would have to be considered digital IMO if its converting the signals to digital and making determinations based on the digital signals.
This is a digital chip. This is how regular digital chips already work.
What you have here is an ASIC.
Not really. Of course it will have a digital interface since it needs to be programmed and operated through a connection to a regular definitely-not-analog computer, but the processing behind that interface is analog (at least based on what I've heard about them - I have no relation to or knowledge of their company so there's always the possibility that the sales pitches are leaving out some important details). Regular digital chips to not play around with arbitrary voltages or currents - that tends to make them not function (or worse, still function but not correctly).
Basically its digital logic to prepare the system for a calculation, then analog to run the calculation, then digital again to read the result back. That's also how quantum computers work (except that middle part is qubit magic rather than analog circuitry).
Its still an ASIC though I suppose, as that term doesn't really specify that the IC components on the must be strictly digital.
I can see these as low power processors that detect “possibly x” and wake up a a digital circuit to “verify it’s x” and “compute or act on x.”
Similar to how our smartphones have specialty digital circuits that listen for speech and wake up higher functions when they detect attempts at vocal input.
That's not really how it works unfortunately. These things aren't detection machines - they're just do really cheap matrix multiplication, with the downside being that the result might not be entirely accurate.
That's fine for AI because the matrices involved are representing neurons, and its often even beneficial for there to be some noise in the system as it functions as a source of non-linearity and "good" non-linearity is critical to making a NN be better than a "simple" calculation (mathematical simplicity that is - a completely linear NN reduces to a polynomial. Likely a very large polynomial that you still wouldn't want to compute by hand but in terms of its processing capabilities its no better than a relatively basic function. Introducing non-linearity fixes that problem and makes NNs far more powerful than any "simple" function could ever be).
That said, (some) AIs absolutely are detection machines and if these can be used to make more efficient AIs then you indirectly get what you want somewhat cheaper.
I have always pondered the great audio debate myself, and although i am going off memories along I think i prefer vinyl. See when something is still analog it's existence seems to be in that of our physical world. When digital things feel too good or fake, being beyond the intention of what the human body was meant to absorb. Its the same difference between playing a rif on a guitar right in front of you (or playing it yourself), versus just listening to a cd of the sounds. When listening to a digital recording everything is on one wavelength it feels. The highs and lows just don't seem to truly reach out and touch our senses or our bodies in the same way that physically making the sound does. This is why I choose vinyl (Analog) over the Digital revolution. Maybe I am right, maybe I am wrong, but based on what I feel is what I wrote.
Most of the vinyls are digital
As much as I love analogue audio, in every single revolution of the Vinly, you lose the original quality of the recording. There's just no physical way to avoid sound degradation on a Vinyl.
Digital recordings, in principle, can last forever, unchanged and in original quality.
@@GoldenEDM_2018 except its digital sounds, not physically real sounds, or so it feels. with non digi vinyls there may be slight degradation, but your body feels every frequency, every wave, rising and lowering with every beat or rhythm. Whereas with digital every sound hits your body one and the same, with no physical difference in the waves because they are programmed to all come out at the same level, whereas vinyls levels raise and lower.
@@YouGoneMad come on, vinyl has a lower resolution than CD, and like most people doesn't need even these fine details in audio, as all use lossy audio now, like MP3 and so on. Vinyl is a placebo, 95% blind test will fail to detect a difference.
@YouGoneMad if you're talking about the volume being the same thats just bad sound mixing, the only advantage that vinyl has is any sentimentality you have for it
Digital was pretty much a term used for things that are Transistor based aka Solid State.
A lot of the misconceptions about digital vs analog where vinyl is concerned, have largely been debunked by numerous audio experts. Also worth considering with the resurgence of Vinyl as "fashionable", that you are actually buying a digitally produced and mastered product with no analog in the chain at all.
There's analogue in the chain when the cutter is dragged across the lacquer master disc and at the other end when the stylus is dragged across the vinyl!
@@xanataph lol
There are just misconceptions about analog and digital computers, too. It is analog at what I like to call the electrical engineering level. Some people think that they are completely different things.
Analogue hasn't gone away or disappeared it's merely resting until it's advanced and improved enough again.
Yes, please. Would like to hear more about the AI aspect in a deep dive
@@Allyh1015 Good luck making that catch on! For 90% of the people out there, talking to an advanced LLM is indistinguishable from chatting with a friend, a really smart friend, mind you. So, the experience of interacting with AI feels a lot like interacting with a human.
Not calling it AI is nothing more than a mere technicality for most people.
I like the term stochastic parrots. ;*[}
But that's just one aspect ....
@@Allyh1015 Can we please agree to disagree? :-)
AI is a rather big field, and its definition is still undecided, as it is growing rapidly.
I would say statistics play a role in it, but once again, that’s why I want the creators of this video / channel to explain AI, as so much of this phrase is just sprinkled all over the many fields of life nowadays.
Furthermore, I would like to add, that when talking about „AI“, I also often think that it’s statistics :-) so we can agree on that.
I also looked up the short definition from IBM‘s website: Artificial intelligence leverages computers and machines to mimic the problem-solving and decision-making capabilities of the human mind. So, yeah, focus lies on mimicking human thinking/problem solving.
Also, predictive analytics is a field of AI. Like I started the comment, AI is a broad field. And there are many areas to it, and some AI tools analyse and predict, while others simply might follow some specific rules. Don’t forget the „if-then“ clauses of Expert Systems“ :-)
My hearing was better when I was younger, so my memories of LPs as having better sound reproduction is biased by that.
I remember the first CDs sounding better mainly because of no pop, hiss, scratch sounds and for the same reason: roll off of my hearing at higher frequencies.
Analog "computers" have benefits but analog electronics are more subject to temperature variances and noise (or inability to identify and suppress noise) but they run fast and are reliable.
In 1984, I was sitting in controls class and the professor rolled in an analog computer and told us how obsolete they were and we would never fool with them. But the majority of the control loops on the planet were still being controlled by analog (electronic and pneumatic) PID controllers, simple analog computers.
This could be the missing link we've needed to develop a true AI.
Must be like 15ish years ago now, I converted an old guy's records to digital, then burned them to CDs, pops and static and all, he frickin loved it in his car
I'm sure they played better in his car than records would have. But I'll bet some of them have gone bad by now, while the records from which they were recorded probably still sound fine.
@@pcno2832 CDs aren't that short lived, besides, you can just do it again, or cut out the media and record MP3 or WMA or something from the vinyl player direct.
Very interesting update on analog computing which has been a struggle for decades. Thanks! However, I would strongly suspect that any method such as the one in this video that depends on ohm's law will be highly sensitive to temperature variations causing huge errors and unrepeatability much greater than the 5% they claim. That is one of the huge benefits of binary it gets rid of all nonlinearities and parasitic factors by slamming the node to the maximum minimum and thresholding everything in between.
What's interesting to me about digital moving forward is that qbits kind of break that mold, at least before the output stage. Sort of a weird mix between analog and digital in a metaphorical sense.
@@ezpk- Qubits break the mold but they don't have any zero or one state at any given time, and I don't think anyone can even imagine how to make a Turing machine out of them. I suspect they'll be at best a co-processor, like a GPU or an ALU, to a classical Turing machine, for many decades to come.
I was first introduced to the classical music on 78 rpm records. The clarity of the music and the sound of the instruments was much clearer.
78s have a nice, long track for every second of sound. It spins so fast that many scratches are rendered out that would have been very loud at 33.3 rpm. They are also thicker than a 33.3 but are more brittle. If you ever do break a 78, crush it into small chunks and put them in a jar and then cover them with turpentine or acetone. It makes an amazing black paint that shines like a mirror.
I think you just scratched the surface of non-digital computing architectures. You could try looking for memristive computing, stochastic computing, photonic computing, magnetic junction computing, etc.
So true!!!
*YIKES, that went over my head i wouldn't want to try to prove you wrong, i learned from Foghorn Leghorn's mistakes vs Eggbert!*
this is super exciting :D Im so stoked to see analog computers finally have a second chance! this is also great news for security!
security?
@@kneesnap1041 yes, it means remot attacks on computer could be much more difficult to pull off, at least in theory.
@@samchickensandwich23 how?
I believe this went over most people’s heads 😅.
Digital can approach absolute accuracy without limit; increase the bits/sample ad infinitum. The limiting factor is noise and that applies to inputs to both a digital and analog system which has to process an analog signal. So minimizing noise in the low-level input signal as it approaches zero is the key and in the case of LIGO, quantum squeezing of light allowed it to get that much closer to the ideal.
An important note: when using resistance-controlled currents or voltages, it should be remembered that resistors/resistance is inherently noisy, no voltage applied. The tradeoff is that the lower the resistance used, the higher the power consumption but the higher the resistances used, the more analog noise is generated. Computing is really a horses for courses game; one uses the methodology most suited to the problem being solved.
same principle as optical matrix computers, the massively parallel ones, not the logical gate ones, both work anyways and are fast, the best optical logic gate computers use waveguide with x-ray photons in the gated pipe system.
Can I ask what kind of work or higher education has given you this knowledge. What subject even? 😮 Thanks
God knows alone@@JustAnotherThisDJ
on the usual suspect I have a semi-phd/master on semiconductor tech in UNI if it matters@@JustAnotherThisDJ
Back in the early 90's, I would copy records to my computer in 46kHz MIDI format. Each song was huge. I had to have a server to store my small collection of 20 to 30 records.
It would be great if you could make a dedicated episode on artificial neural networks. 🙏
Thank you so much it’s on our list!
You lost me in parts of this but the last bit about use cases made it all clear. Yep, I think like everything else, there needs to be a balance -- between food & exercise, wind & solar, analog & digital.😊
It's a common mistake to say that digital audio is an approximation of the analogue sound waves. Anything recorded above the Nyquist value is essentially perfect. Digital audio is not little squares and rectangles. The poorer sound of some early digital equipment was due to bad DACs and the pre-amp needed to get the signal to line level. Some people prefer vinyl because the compression makes the mid-range stand out.
Funny when most contemporary music already has the shit compressed out of it in production....
I see a way analog computers can compliment digital computers, game consule AI. Basic game logic could be stored and use digital computing, while enemy and NPC logic could use analog logic to evolve the game over time.
Vinyl has its tradeoffs, first, the waveform has to be compressed, as the full dynamic range cannot fit into the grooves, so while you might be getting (in some cases) a more accurate representation of the original sound wave on a vinyl, if it was recorded right and the record is in perfect condition, the dynamic range is reduced, so you might not be getting what the original recording sounded like even then... With digital, it's possible to get to such high bitrates and sample rates that our ears and brains simply cannot begin to tell the difference from the original sound wave.
... though the irony is that when we moved to digital we also moved to greater dynamic compression than was used for analogue recordings and vinyl pressings (for reasons of musical taste/fashion/style). So yes, the digital formats can potentially represent the higher dynamic range of a musical performance than vinyl but in practice they use less of that potential in most commercial music recordings.
@@MrAdopado Yeah it's only niche music that doesn't actually have the shit compressed out of it in contemporary digital
Exact computations are not necessary for most applications. Our eyesight & hearing is not exact, but speed to process massive quantities of divergent information is more important.
Reducing power to perform high speed computations in near exact results is good enough for smart bombs 💣 , smart rockets 🚀, smart artillery, smart bullets.
Similar needs for autonomic driving systems could be an excellent application… conserving more power for expanding distance of the vehicle.
Honestly, this can change a lot in military & civilian applications.
how did you move the hair roots from
your head to lower face half ?
Oh yes boi sure can’t wait to stall my sons 2045 gaming PC while he’s playing GTA IX 😂
I think these analog chips will make their way in devices. Regarding the question between vinyl and digital, definitely vinyl for me. The experience is so much more than just the music. It’s the decision process on what to listen next, the album art, and as you have to be careful with it, it feels valuable
I would rather have SACD, BR Audio or DVD Audio. I want the audio that matches the studio master, no more no less.
"and as you have to be careful with it, it feels valuable." This kind of nonsense just makes me want to go on a rampage, because there is absolutely nothing about music recorded on vinyl that comes close to the repeatability you get with digital. If it feels more valuable because every time you drag a diamond through that vinyl, it loses quality, and you're happy with that, then by all means, DON'T listen to your vinyl records because of how careful you have to be with them.
@@BrightBlueJim You're exactly right. Back in the day, I found it exasperating to put on a brand new factory-sealed album and hear scratches *the very first time I played it*. At the time, it was explained to me that this was because North American record plants only pressed the albums for 1 second while those in Japan would press them for 30 seconds. This was supposed to account for the great superiority of Japanese pressings. Give me a CD any time!
The right speakers really matter well.
its like mechanical watches, its vastly inferior but it feels more human because its not digital perfection and you can see and touch it@@BrightBlueJim
My position on digital versus analog recordings is that they are just that, recordings. I prefer to listen to live music even if it’s by amateurs. A recording will always be an estimation of the live performance. Analog is still an estimation.
Non engineer here, weeds got pretty deep in this episode.
At least for this old, analog brain.
But still I persist!
Sorry… was it at least a bit helpful?
i too was and still am that way. that's when i like to dive into the rabbit hole DIY guy here yet a photovoltaic / solar thermal technician
During a electric/electronic class, I was asked about high impedance with transistors use. I said a three answer computer would be much more useful than a two answer. The teacher said it would require a whole redo of programing and not be cost effective. Both were correct.
The first computer I ever used was an analogue computer. Implemented with op-amps.
@borregoayudando1481 Great question😅
@borregoayudando1481 🤣😅😂
FPAAs (Field Programmable Analog Arrays - Analog FPGAs more or less) are reconfigurable analog logic devices that can be used for prototyping, changing resistors won't be too complicated.
- haven't read much on how these work though 🤔
This makes me even more excited 🤩 for starting my degree in electronics engineering!
Vinyl isn't really analog anymore. It's recorded in a digital based media, then translated back to analog. Not the same as storing the data on tapes, then reproducing to vinyl.
Vinyl is a whole can of compromises to have an acceptable reproduction of a recording on media. You then boosted bass to compensate when listening for example. Past a point, bass frequencies would make the needle hop in some bad way I understand. I lived in the day when you always boosted bass :) You might like hearing the result of those compromises. So yeah, not a good example. :)
Has been this way since the 1980's, some people were doing it in the 70's, all stuff is recorded on digital audio tapes or other digital storage today.
Yeah right. High Fidelity original. Then you can fit it to the appropriate media tape or vinyl.@@edwardecl
Vinyl has a digital master track encoded into it but the medium and the way music is produced from it is definitely analogue. It's true it wasn't made with a 100% analog signal chain from start to finish but who gives a shit?
[13:03] Mmmmmm... that Ubuntu Unity desktop on the screen, even though it is slightly cut off, I love to see it!
Analog is “random or imprecise digital” just that. Forget the myth. The granularity of an analog picture is their “pixel” only randomized. The same about the fidelity and noise to signal ratio, on an analog recording (not to mention mechanical artifacts).
With the added problem that each copy adds another layer of imprecision or “pixelation”.
Edit: Remember analog computing is only a cool name for more bit states than 0 or 1 determined (usually) by voltage levels.
i'm gonna borrow that. i'm not blessed with that type of articulate grammar, i am though in a bit of gab. my hat off to you a nd *cerveza for you* 🍺
@@fvrrljr Thanks… I guess. Either way, if it is sarcasm.. or not… I am always open to improvement. Analog or digital 😂
@@frgv4060 i love sarcasm but no sarcasm here. i really was impressed with your comment 👍
Yes, analog is just lossy compression. Though floating-point calculations are already a bit imprecise even on digital computers, and GPUs can be even less accurate because in real-time graphics tiny errors don't really matter and AI doesn't need perfect precision either (8-bit quantization in AI models is kind of the norm because it reduces memory usage by ~75% while only introducing around 1-2% deviation to the results)
thanks for educating in simple and laymen terms!!
Nope, they won't change anything. There is a reason why pretty much all computers are digital, and that reason is noise. This is pretty much the first thing you learn when you start with digital electronics. Digital computers are just orders of magnitude easier to build for our noisy environment. Or like analog IC designer Bob Widlar said, "every idiot can count to one". For example, quantum computers are full analog and they are so incredibly fiddly, that we haven't seen a useful one yet. According to Wikipedia the largest factored number on a quantum computer using Shor's algorithm is still 21.
Analog is a better representation of the real world. We expect the world to be a competition between sources. An environment that contains no sources is unnatural. It may be easier to conduct science by filtering out the distractions but it does not contribute to the understanding of the impact of the interrelationships.
i wouldn't discount the possibility of a analog/digital being used to create a new technology using both. I am working on some things now.
Not all analog computers are loud. I saw a video showing how you can use electricity inside an analog computer that is using the voltage as the value and there they showed how simple multiplication was which is a large part of AI computation so while not big in usual use cases they are good for AI
@@patfrethey don’t mean noise as in sound
Many great points, but some applications are tolerant of noise and the speed and energy savings in analog computing could far outweigh dealing with noise. Computer vision prossesing for starters. Speech to text could do a significant amount of processing before digitizing is nessesary. Analog computing would not change much, except in a few applications superseding digital processing, but it will enabling more rapid advancement in those areas.
In nural networks with weights that are currently digitized, switching to analog weights is far far far far more energy efficent and far far far far far more speedy. Memresistors may have significant aplication in analog memory with far greater storage capacity and speed in neural networks. In a neutral network, a weight with a noise between 0.998 and 1.002 can be insignificant enough to justify analog computing benifits in an aplication.
WW-II naval ships, used well into the the1970s, used mechanical analog computers for navigation.
All automatic transmission, prior to computer controls, used fluidic analog computers (the valve body) to upshift and downshift.
The inaccuracy can potentially be a good thing for general AI. With sufficient checks and balances via a few extra neural networks, errors can be seen by the computer, that's how it works with people. We gain experience and we learn how to spot our errors. Plus, error creates variation, variation is creativity... the foundation of art and culture.
It depends on what kind of expert ais. For tasks that require a high level of certainty, it could be detrimental.
@@evertchin Think about image recognition. Training the AI with a couple % error rate could improve the ability of the AI to detect noisy or difficult images, as it was trained from the start with error-ridden data.
@@kindlin That's how AI has been used for decades. All trained AI models have intentional inaccuracy to prevent an issue called overfitting. It's one of the basic things taught in every AI course because if you don't do it the AI model is near unusable. But you don't need analog computers for that, all that's necessary in most cases is to not run the training for too long.
I'm 70 and grew up with analogue vinyl and magnetic audio. I loved them, of course...they were what existed, and high end equipment sounds fabulous!! At one point I owned several thousand vinyl LPs and hundreds of cassettes.
They had their weak points, surface noise, tape hiss, and a far lower dynamic range. I was easily persuaded to the dark side (digital) four decades ago, when my sense of hearing was far better than I can claim today. Even then, around 30 years of age, I could not tell the difference, aside from the advantages of digital.
EDIT 2 weeks later: Once beyond the discussion of analogue vs. digital audio, things required a few more watts to take in. There's a lot I didn't absorb in the first run.
Nah the future will be in mechanical computers!!! 😃😁
The only advantage CDs had over vinyl and tape was the lack of scratch/pop/click noise or tape hiss. As for the sound, at 44.1kHz / 16 bits it simply lacked resolution. Today's high resolution audio is much better sounding and is close enough for most people to not be able to tell digital from analog in listening tests.
Seems like someone is trying to leverage the hype of AI training with Analog Computing(which has nothing to do with AI training to start with🤔) for a reason, the usual suspects😂. Things will fall apart, once the fad winds down, or takes on a different route? that could use digital computing effectively!.❤👍
oops... One meter does not equal one foot, or even similar to one foot. 12:39 One yard, yes... but one foot? Nope...
Not sure if that was intentional, but sounds like a huge error.
Heres why this is clickbait crap.
Exactly, "analog computers about to takeover", then proceeds to talk about what digital is, what NNs are, what ohms law is and finally why we use digital,
Like where the fuck was the content described in the title?
Not always with mp3 files but with Wave yes also they have 320k 32-bit recording which is really good a lot more to work with
What about wear though? Digital only needs to listen to on and off, but analog has a whole range. We already see similar in ssds adding more data by increasing the sensitivity to voltage. As the cell wears down, it becomes harder to measure the difference in qlc vs tlc or mlc.
It actually matters less in analog, as you don't need to measure differences - you're using the whole range. If you measure 4.83 volts then your answer is simply "4.83". You're not comparing 4.83v against +5v and -5v (or whatever values your circuit uses) and trying to match the closest as you would do if you were wanting to distinguish 0 from 1 in a digital circuit (or from a 2 or a 3 in multi-level cells).
So much disturbing sound in the background we cannot concentrate on what you're saying. We have to unsubscribe from you.
We do?
@@dwang085Hivemind has spoken. We must do as we have instructed.
There are kit robots with very simple analog circuits out there. I saw a video for one years ago that was built with paper clip type pieces of metal for legs the person filming demonstrated that you could damage a couple of the legs and the robot would "learn" how to walk without them. Seriously impressive emergent behavior coming from something that only had the equivalent of a few transistors worth of logic to work with.
BEAM Robotics
Current hardware had an explosion in errors due to increasingly smaller hardware reducing error margins. There was a paper DDR4 increased error rates by 5.5x to DDR3 and then they slapped on die ecc on consumer ddr5 just to deal with it. Maybe they are saying "if we cant contain errors in digital and need checks anyway we might as well use analog"
well if you can make something 10x as fast and then need to use half of that for error correcting codes thats still a 5x improvement
Back in 1968 I used an analog computer to simulate a vehicle chassis. Position, acceleration, and jerk are related by differential equations, and the wheels are interconnected too. But a simple analog computer with maybe a dozen vacuum tubes could do differential calculus instantly, and you could change the spring constant or shock damping in real time to see how it changed the handling. To do one simulation on a computer I had to punch a whole stack of cards and wait for it to run on the mainframe ... overnight.
Any modern vinyl record would be mixed to have lower frequencies in the middle channel (not far apart in stereo spectrum) to prevent track skip. So... the FLAC and the vinyl copy of the same song will be significantly different anyway.
Please keep us up to date -Thanks
Excellent video 👍 Thank you 💜 and Merry Christmas 🎅🎄🎁
On the topic of digital vs vinyl, I think a property EQ’ed uncompressed 24 bit 96khz signal, heck probably even a little lower, is going to be physically indistinguishable from a vinyl recording. If you think about how speakers actually move, speaker motors are not infinitely fast. This means that the “jaggyness” that inherently exists within a digital signal compared to it’s analog counterpart is going to be inherently smoothed out by physics. Heck, I’m sure even the DAC in most systems, even the high end ones exhibit this kind of signal smoothing electricity themselves in the conversion from digital to analog. It’s not a digital signal that makes a sound inherently inferior, it’s the compression
What you describe as the inherent "jaggyness" of a digital signal is a very common misconception ... it seems like it must be the case when you look at some graphs and it fits with our intuitive expectation ... but it's wrong. If you do some research on the subject you will understand why. In short, the jaggy graph represents sample points but it does not represent the sound wave that is produced ... which is a perfectly smooth sine wave. Yes, it's hard to get your head around this at first (I was the same)!
Now let's move onto compression ... are you talking about dynamic compression or data compression? If you are talking about data compression then the digital signal used for CD is uncompressed ... so no problem there. Digital recording doesn't use data compression either ... but ... streaming services generally do use data compression. They need to save data bandwidth because of costs for them as suppliers and for you receiving their product via your limited bandwidth data connection. If they use significant "lossy" data compression then the audio can be noticeably degraded. If they stick to "lossless" data compression then you end up with no degradation (but the data rate will need to be much higher and more expensive to them and to you). Quite separately we have the issue of dynamic compression ... it would be better if there was a completely different word for compression in this instance because it has nothing directly to do with data compression. In fact you can have a highly compressed (dynamically) track that is delivered to you via a totally uncompressed digital signal ... in fact that is very very common in modern music recordings! The element of a digital file that describes the potential for dynamic range is the bit rate (e.g. 16 bit 24 bit etc). It enables very quiet sounds and very loud sounds without you hearing any background noise. The advantage of 24 bit is mostly at the point of recording as it gives you the ability to match sound levels without bringing up the noise floor. Once the track has been mixed and mastered the advantage of the higher bit rates pretty much disappears ... because a final 16 bit copy is going to have plenty of available dynamic range such that it isn't going to be a restriction to your listener.
This is exciting to learn about. I imagined this, but using the way solid state stores data in this way would have never occurred to me. Exciting.
I saw the Veratisium vid when it came out and have been excited for the technology ever since. Digital computation could run up against significant limits and this looks like an interesting work around.
The fundamental issue with this is that what you're describing is just the analog version of ASICs, but even more narrowly focused and harder to design reliably. Given how rarely we already use ASICs even in dedicated high value applications, I see this having very limited practical use unfortunately, as it'll just never be able to reach economies of scale since the requirements vary too much in every application to reuse.
That beat at 14 minutes is what they call fire.
We had analog computing in the 70s/80s when I was in the navy. Everything became digital. Hard to believe analog is making a comeback.
Just like with quantum computers, and in contrast to popular misconceptions, the goal is not to fully replace digital computers, but rather find more optimal means of solving problems. It's always exciting to see that there could be a different computer structure that does it better than classic digital computers, and a giant breakthrough in computing.
Well said! Yeah we have an all or nothing mindset. But really it’s about new tools for different tasks
I can't be the only person who watched this and immediately thought of Warhammer 40k's clockwork cyborgs and analog thinking machines.
I started out my career in electronics design, led me a job with a DOD contractor working in a computer simulation labs, many many computers doing also hardware in teh loop. this required real time computing, taking several computers in a room, digital was the host computers to control other Analog hybrid computers, using voltages, op amps, multipliers, DAC/ADC converters. Analog at the time was the only way to compute real time as a voltage change was almost instantaneous. and input/outputs directly coupled to test hardware that could be switched in the simulation loop the hardware would think was operating in real time. Later came high speed digitial using ECL, which was actually digital circuits that operated in neither on or off state, but as amplifiers biased partially on and only change intermediate voltage levels to get the speed which normal digital circuits that were all off and all on created too much harmonics to get the speed needed, these were power hungry special designed beasts. Outside of any real commercial potential for quantum computing coming on line I would think a new hybrid approach to using analog in the computing mix again could break through current limitations on speed.
Applied Dynamics International was a big player in this computing area