Видео

Hey, I hope this message finds you well! 🙂 I recently discovered your RUclips channel while scrolling homepage and found your content to be amazing and incredibly helpful. I have a cooperation proposal for you and would love to discuss it further. It could improve your channel. Please let me know if you're interested! Email address will be great :D

I find it sad that even a professor at Cambridge like the big names in the industry doesn't distinguish between the letter x and a cross-product

But why did it shut down? The final image ever taken makes you feel sad.

Perfectly amazing

AI should never be allowed to alter the composition of living things.

Is Sonic Pi still in active development? The website seems to not have been updated for a long time.

Is the recording a bit glitchy? Great talk, this is helpful and informative

Amazing lecture dear Ms.Willson❤️🤗

FYI : I don't like long-winded people.

i Can't speak english well. Help me in the class plese tha i can Understand what you speech. If I need translator, are there?

I lived in Indonesia. Suradita ilIndah Blok H2/12 suradita Indah KORPRI province Banten Help me in the class my English. Why I like computer science? Because ... hope yo know.

My name is Arif Rachman. I'm attended in BINUS Unevirsity. Hope you understand my English. I wan to registration in computer science. But help me to requirement registration in the CAMBRIDGE UNIVERSITY. but my English not well. Hope others understand my English letter. Thank you

I think I pointed this out on another version of this lecture, but Roger's cycle count for the 32-bit add on the 6502 is way way way off! - well, just a bit anyway - the actual cycle count is 38 minimum - the CLC takes 2 clks and all the rest take 3 (assuming they're using page zero addressing)

A brilliant and handsome woman.

Promo-SM 🌷

I think Intel is holding back because the cost v. reward ratio for going smaller on conventional silicon lithography isn't going to be great. Essentially it will be easier to print individual atoms onto a substrate than it will be to create and develop all of the masks required to print with conventional lithography onto silicon. They're essentially at their limit with the use of actual LIGHT frequencies. Once the architecture is less than 1/4 of the light wavelength you run into the problems, like propagation, that make you require all of the extra masks to make the lithography actually work so this is a path of diminishing returns, just like tooling up for 450mm silicon would have been, and it is obvious that Intel is well aware of the fact. I wonder if the execs who green-lit 450mm silicon projects are still working for the same companies.

So, essentially exactly the same content as 4 years prior with only a passing comment on quantum computing only after being asked a question in the last 5 minutes. Hmm. The advancements in GPU's happened because they are used primarily for mining bitcoin... LOL, I bet that slows down markedly on next year's graph.

(1) Is supercomputing now exclusively done by parallelism? Is there any research that could yield a much faster (TeraHertz +) CPU even at great cost? (2) 1:00:50 Since masks are expensive, has there been research for alternatives? Perhaps some kind of dynamic mask where elements can be polarised like liquid crystals, or reprogrammed by a laser?

These short talks are so interesting! It would be great to see more about students' projects.

Great idea, I hope it really helps to inspire young people and help them to develop the skills needed. I'm also jealous of the exciting future that will open up to them.

Hi The Processors of a processor I'm designing can be used in virtual environment for speed, energy efficiency, etc. I'm looking for free software if possible that tests things. Do you know any software or method? Even if you don't know, do you know an address that is likely to know?

Damn, this is a good lecture.

What you have done is so inspiring. Don't stop until the present day morass of tools is only an unpleasant memory.

was this made using an AI or something?

Wow, I love the second guys sense of humour

1:24:24 Zen 2 - Intel compatible instruction set... well, x64 was designed by AMD.

Wow I now have a goal, thanks for the video.

1:25:45 gotcha, won’t squeegee my pentium for extra gigahertz

We should figure out ways to decentralize manufacturing of microprocessors so DIYers can come up with much more powerful ones. There is too much centralization. Let's also bring back Made in America.

Processor speed progress walls. We need more parallel software? Say you had it. Processors still need data to process, right? Well with infinite power comes infinite data starvation. What about memory latency? Total bandwidth can increase as long as we have buses that we can widen but latency? I think we need a software revolution that makes good use of scratchpad memory.

She also mentioned the EUV tech being made by an EU company. What many don't realize is, they don't actually own the tech themselves. The US does. They have a license to use it in manufacturing their machines as they took on the engineering challenge of taking the tech and producing a practical product with it.

She feels out of touch on the intel vs amd vs tsmc vs samsung situation. Doesn't invalidate much of what she says, she just has the reason wrong on why AMD has returned to competitiveness and how it has galvanized intel. You could also hear her clear partisanship for intel which is surprising giving her role history with ARM. The real reason many companies have had to drop out of the race for 10nm and below is because only a single company on the entire planet has successfully developed the tools for it, ASMC in the Netherlands. They can only build so many each year and every unit they build is being snapped up by TSMC, Samsung, and Intel.

so useful and informative for people who are truly interested in digging information about computers especially CPUs

59:14 the chart really needs some codename and CPU names correction, e.g: Nehalem

Schools in Australia had these. My Dad was a Maths 1, Maths 2 and Physics teacher, who spent his free time on electronics. As such, he was chosen to be the Computer Class teacher. He bought a BBC Model B to use at home. Somehow we ended up with 2 of them but I don't know why 😀 As a young kid I used to get BBC Micro User magazine. Each copy had the BASIC code for a game printed in them for you to then type in yourself (learn to code and get a game out of it). I learned BBC BASIC and Assembly Language through this period. Learning to code resulted in a lifelong interest / hobby, me pursuing an ICT degree, and a subsequent 20+ year career in ICT. So from my perspective they achieved their aim. I also have at least half a dozen Raspberry Pi computers performing functions around the house.

This is awesome! In 1983 when I was 6, my dad bought an Acorn Electron and that was my introduction to computers and programing in BASIC. We had room fulls of Acorn BBC Micro's and Acorn Archimedes machines at our senior school as well. I moved onto the Atari ST after using the Acorn for about 8 years, and then jumped into PC gaming at college. It seems like Dr Wilson's work might be the major reason behind my interest in tech.

The BBC micro was the machine that really taught me how to program. I'd dabbled with a Sinclair Z80 - with all it's limitations, but it was really the BBC Micro that really opened my eyes to what computers could do and gave me my entire career 😀. The Raspberry Pi gave me back the ability to experiment and have fun 🤩 so thank you both🙏🏻

We had BBC Micro's in school. I wrote my first lines of code on that aged 13 or so. Great to hear these stories!

Perfect, wish I had known of Sophie before now.

so inspiring!

I've seen Sophie give versions of this talk and am often surprised at how (unintentionally?) critical she is of the humble 6502. I urge people to watch the Computer History Museum's oral history with Chuck Peddle. In short, the 6502 design team had a transitor/die-size budget in 1974 and it is REMARKABLE what they achieved for the price. It is no surprise that with a further decade of computer science research into CPU design, Sophie and Steve were able to vastly improve their blank-sheet design. My point is, some industry and economic context would be useful with regards CPU design decisions.

Sir please inform me does Cambridge offer software engineering studies ??

Watching this in August 2021 I see that Sophie subtly previewed Tesla’s wafer scale AI processing architecture 😀

Wow. Leaned a-lot. Thank you so much Cambridge and of course, Dr. Wilson

Brilliant lecture, very interesting and Sophie was a great speaker, will definitely be checking out her other talks

At the 52 minute mark, a Danish politician named Karl Kristian is mentioned as the source of the saying "Predictions are hard to make, especially about the future". For those who are equally interested in politics and history as in microprocessors and history, I can inform you that K.K. is better known as K.K. Steincke.

i hope to become computer scientist one day at cambridge

very cool

TDS knows no career boundaries lol.

If I read 1:06:25 correctly, then the number of transistors per core has also plateaued out: If you subtract the number of logical cores from the transistors, the transistor scatter graph looks a lot more like the other three.