I'm a mechanical engineer (independent contractor, so I have to pay out of pocket to rent software). I rented a stand-alone seat of ANSYS (structural + thermal + SolidWorks Plug-in) for a year and a half for a project, and it cost more than 2X as much as my car. It's powerful, but *man*, it's expensive.
There's free Finite Element Analysis software that's maintained by the Finnish Government called Elmer FEM. It's a bit wonky to get everything integrated into a workflow but it does all the things, and is free.
Ansys also owns Redhawk which is the industry-leading IR/EM analysis tool. This was a hole in their backend design flow that I'm sure Synopsys wanted to fill. Cadence has Voltus for this type of verification and now Synopsys has responded.
They also bought AGI (formerly: Analytical Graphics Inc, now: Ansys Government Initiatives) a year or two ago, who track satellites and do geospatial simulations, and a few years back they created and spun off cesium (3D maps in web browsers).
@@LostieTrekieTechie - just to clarify, ANSYS did not spin off Cesium, AGI did the spin off as part of the (many) events involved in ANSYS acquiring AGI. Former AGI employee number 37 posting here.
Hopefully Ansys start to a make workable programs. The programs are too famoust to corrupt files and delete simulation results. I have even wrote a parady ”the project must go on” from the ”the show must on ”because of Ansys
As a Mech. Engineer who has mostly interacted with Ansys as a general Mechanical FEM software (Stresses, Eigenfrequencies) its a nice suprise to see software from the CAE-Sector show up on your channel. A big Advantage of ANsys as a simulation software I have seen a lot of people proclaim over other CFD/FEM/Thermal programs is that it has a lot of different modules, and is steadily connecting them (e.g.: Heat Transfer + the thermal impact on material properties under heat). Have you considered covering general CAD (i.e. Modelling) Software? Might not be highly relevant to your main topic of semiconductors, but I think it would be nice to see a coverage of the 2D->2.5D->3D evolution of Software, and the modern trends (integrated FEM (aka what you tend to call CAE in this video), CAD-CAM...).
I'm a fan of silicon graphics computers and this is the perfect topic to bring up that company with those types of machines, those machines are so cool.😊
If everyone is paid fatly, that’s some nice wealth redistribution towards employees then. I often think about it with my company: Every buck I save on employees produces 10 bucks on valuation of my shares. (Real story: I try to differentiate employees who f me over from those who are genuine) (and now we have an absolutely awesome team).
Its more than that. Their softwares are used for basically every single product and most steps of development. HFSS is used to simulate EM behaviour of devices for everything from small RF chips to wifi enabled coffee machines. They have SPEOS and Lumerical to simulate photonic chips up to how light behaves inside a newly designed car, the list ist pretty long.
I find it very inspiring to watch any of your semiconductor videos as I am currently studying electrical engineering and we have a course that is related to circuit design (mostly analog) on a silicon level. It's really great to have a peak into cutting edge from time to time! It's amazing, your videos are both entertaining and motivate me to study... Insane combo!
I was just running a CFD simulation on ansys and you uploaded this video, so nice to see an youtuber that I follow talking about something that I use everyday.
You had me at "In the 1960s, John Swanson was working at the Westinghouse Astronuclear Lab in Pittsburgh.He went there to work on nuclear-fueled rockets with the goal of going to Mars - the NERVA " What an incredible series of words, I'd love to see what Vonnegut/Clarke/Asimov/Gibson/Haldeman/Heinlein would do with that as the first line of a short story.
I have to thank you. Not sure why but you caused me to search for a similar words pronounciation. I googled it on the phone and discovered google not only has pronunciation (which I expected) but it actually will listen to you say the word and write out your mistakes. Thanks!
another great video. If only tech review youtubers would realize how tough it is to design and build chips, maybe they would shut up about performance benchmarks. fact that AMD and Apple can keep increasing performance every year is an amazing achievement that people take fore granted
This critical misconception has permeated everything in physics and engineering and is so damaging: "the complex non-linear equations that _govern_ fluid dynamics". no. just no. the flavor of algebra, and equations we express in that specific algebra, that we've invented, are just imprecise _models_ of fluid dynamics. This isn't a subtle difference in language, it's a fundamental and enormously damaging misconception that holds back human progress.
Not only that, but the underlying mathematical differences between solid modeling and fluid modeling mean that (in general) any model that includes fluid-structure interaction cannot be arbitrarily scaled. That's why wind tunnels are still being built. Also, I find it somewhat ironic that ANSYS (the structural FEA application) is one of those applications that does not easily take advantage of parallel computing, given that the company was purchased with the goal of aiding the chiplet design process.
There are software like COMSOL that operate on the constitutive relationships: solving the first-principles equations inside mesh cells and the boundary relationships between them. Monte Carlo simulations (more common in research, but do exist in some industries) also operate similarly. Confusing these with the sort of CFD simulation that ANSYS does in a genera-audience video like this does not hold back human progress. The humans making progress use all sorts of different tools for their simulation. There is even a third archetype that gets some use, and that's before we get into the FFT/ IFT versions of all the above.
All it takes to accelerate human progress…. Is grab the bull by the horns, and use your calculus skills… More time spent complaining about what another person has done, is one more day wasted not getting to the place you want to be… You can be a billionaire… get on your horse and ride! 😃
Aye. People seem to think the imprecise models they use actually drive reality. And then complain when reality disagrees. Physics approximates reality, nothing "governs" nature.
man do I not miss the math of my fluid dynamics class. never take 2 programming classes and 2 physics classes in 1 quarter unless you are genuinely obsessed with both 😂
I'm a Chemical engineer who has worked with ANSYS fluent & built some structural components which we later tested out. It is an awesome software which works great & I am really amazed to see the fluid simulations. We had Computational Fluid Dynamics as a subject which deals with various parts as also explained in the video.
Somewhere in the bowels of each automotive company, someone has the data on how they know all those plastic parts on your car's engine will turn brittle and fail. All modelled in Ansys I'm sure.
Another note on anti-monopoly issues: given how crucial thermal simulation is for advanced packaging, Synopsis could shut off this capability for competitors such as Cadence I don't know if there are competitors to Ansys that could fill this role as a complementary solution to other EDA vendors
I don't think that would fly in most of the world with respect to anti competitive practices. The only thing that comes close is that I've heard (but not myself seen) that Cadence, for example, gives discounts for EMX if cadence is always your first supplier for any tool they have in their portfolio. In other words, if you also want to use Keysight RFPro, you pay more for EMX.
@@mahussain1Comsol probably works great for truly coupled problem like RF -> heat -> deformation -> RF but it doesn’t target any specific problem well.
This can be summarized as virtually prototyping a design from “silicon to system.” Imagine a design flow that starts at a chip and goes all the way through package to pcb to automobile.
I actually did my PhD on optimizing MCM packages. Balancing heat density and inter-chip latency due to distance was a huge tradeoff I had to consider. I think a great example of this now is looking at Intel's Sapphire and Emerald Rapids CPUs with the cores spread out away from the EMIB interconnects is a great example of this need.
One thing missing from academia teaching of tech history is biography, which you cover. It is amazing how important agency and human decision is for change.
Not even a mention? The core design of eUV and every laser, imaging, illumination, etc. optical application is designed and analyzed on Code-V (Synopsys) and Zemax (Ansys). Now, Synopsys and Synopsys. As a Zemax user, I am concerned what will happen to Zemax.
Lumerical and Rsoft will also become one company. The entire optics simulation industry will become highly concentrated. I am afraid they will stop rolling out features because the market is locked down. Of course, price hikes are a given.
I only clicked on this because I build powerful servers on a daily basis and CFD is one of the more useful ways of really leveraging all that computing power. Simulating thermodynamics in processors using processors is wonderful, we went full circle ;)
"In any system of energy, Control is what consumes energy the most. No energy store holds enough energy to extract an amount of energy equal to the total energy it stores. No system of energy can deliver sum useful energy in excess of the total energy put into constructing it. This universal truth applies to all systems. Energy, like time, flows from past to future"(2017).
This package design is definitely the intermediate future thing… The main problems will definitely be focused around cooling. Reminds me of the dawn of high rises in cities, we built up for specialization and climate control as we built upwards it was a learning process. The typical heat sink and heat dissipator/IHS isn’t going to work as well as it used to anymore..
I remember implementing Finite Elephant Method in school - such a beautiful algorithm. I wanted to compliment you on your quality subtitles, we can really tell the care that went into getting them right.
After a 20 year love affair with Altium, our team is moving on. Kicad is pretty good these days. I'd rather stick with Altium but eventually I will be forced to get with the program.
Awesome that you mention Welch! He's the father of staggered grids. Staggered grids are the future of CFD. If only Perot will figure out his 3D mimetic interpolation for polyhedrals.
Nice video...I have always enjoyed your videos....Ansys also has Lumerical FDTD solutions...which are used for optical interconnects, etc. Eventually, all interconnects on these chips might be replaced by light based interconnects to reduced heat...
Just to put that in context, a few years back, IBM bought Redhat for US $37 bil. And Apple is worth (I googled this before typing it) US $2.4 Trillion. We are talking about a substantial purchase, but not insane by any means.
I will say that maybe $35 billion is too much, it represents 17 years of Ansys revenue, not earnings. They are growing pretty fast: 8.33% but the net Income is less than $500M. This will kind of problematic numbers, normally leading to employee reduction and cost-cutting as the C-level executives try to justify the purchase. Clearly letting go of talent and cut R&D and workers' satisfaction is not going to increase the value of ANSY, but it could get the C-Lebeks bonuses. The other way around it, normally, is trying to get a monopoly and increase the prices, a lot. We need to change the legal framework to allow both the stockholders and the courts to make the C-Levels responsible for dealing with it, and force them to show that they make financial sense under penalty. $35 billion is around 41% of Synopsys' market cap.
Your analysis, questioning the rationale for buying Ansys for $35B is astute.. There must be at least some semblance of a counter-argument which persuaded both the execs, as well as other financial stakeholders that claims to warrant the purchase price. I doubt that rationale would be to monopolize the product, as that surely would result in years of costly litigation, with an eventual loss. My WAG (Wild Ass Guess) is that Ansys will continue, but that the engineers may also have an ambitious 'skunkworks' project for components that uniquely serve the chip industry. For example, #1) With sizes shrinking to a couple nanometers and smaller, there may be quantum effects that are yet to be modeled. #2) IBM and others have begun experimenting with using light (nanometer sized LEDs and photosensors in a 3d matrix to fan-out signals. So, perhaps the package could emulate physics inclusive of optics, which at that speed and scale are nonlinear. #3) I've read of advantages to pulsing fluids instead of continuous flow (e.g. heartbeat). That may be applicable at nano scale. #4) I once did a little work on MEMS. While Ansys can already simulate MEMS, including multi-physics, such as electrical input/output of piezoelectric materials, there may be much more to this, especially at nanoscale. I may be overly optimistic, but my WAG is that there's some futuristic prospect to warrant such a huge gamble, something better than just upping the price for the product's use, or monopolizing it.
The lever isn’t cost here. I predict that they will jack up the cost licenses massively. Ansys is already so entrenched in many design processes, companies will pay up.
As if Ansys is the only company in the field CFD. I think that they were searching for other options and this boosted the price for buying Ansys, because the others were not for sale.
As a CAE engineer, I would argue that your perspective may be biased. The semiconductor industry is relatively small compared to the billions these companies earn by charging hundreds of thousands in licensing fees each year from universities and R&D labs. ANSYS, the world leader in CAE, has changed ownership multiple times and has acquired numerous CAE companies to gain access to their solvers, such as Fluent and ICEM for CFD, and LS-DYNA for automotive crash analysis. Now, they are introducing ADINA, which uses the same technology as LS-DYNA, potentially threatening its existence. Moreover, ANSYS faces stiff competition from Dassault's Abaqus, Altair, MSC, and COMSOL. Interestingly, in the semiconductor and laser R&D sectors, COMSOL is widely used, while ANSYS is predominantly used in mechanical systems and HVAC.
I don't think Synopsys must be doing too bad with revenue when half the projects I've been in have talks of adding a couple hundreds of licenses of VCS mid project to keep up with the regression eating up all of them.
Micron had kgd (known good die)process where we would take good die and place in temp package and do full burn in testing, then removed the die from package and sold it for early ceramic packaged MCM, multi chip modules. Good money in KGD back then.
Ansys, whilst powerful, reminds me of using a piece of software from 2 decades ago. They acquired lots of companies and simply bolted it all together, and unfrotunately the user experience reflects this. I wish it had a UI/UX like COMSOL. Unfortuantely for COMSOL it falls over when trying to do anythign remotely complex. As a Mech E with lots of experience in this field, it was nice to watch a video on something closer to home!
As an EE who is just starting to use COMSOL, I wish their UI were more intuitive, all the options for your solution are buried 5 layers deep. I agree that Ansys' bolting stuff together isn't great either: taking a mechanical simulation and then doing an RF simulation on it can get annoying. But I found that using a single tool from them wasn't too bad to set up and do optometric sweeps.
I think the caveat with comsol is that it is so flexible it lets you setup simulation in inefficient ways that don't lend themselves to complex problems; More planning is required when setting up boundary conditions and even solving with comsol over ansys, but on the flip side comsol can simulate things ansys can't.
Without having used abaqus a lot the impression I get is they have stuck to their core product and refined it. Whereas ansys has been shopping companies around like crazy to have a piece of software for every application without any intention to refine and perfect anything (for example I fear for LS-Dyna development)
Ansys is likely up there close to the ranks of ASML, some company not in the common man's consciousness but responsible for a huge portion of everything we have and enjoy today. It's stupid powerful and absolutely essential for some black magic engineering (e.g. RF)
Ansys has to face stiffer competition than ASML. They don't have any revolutionary technology unknown to anyone else. They have a history of buying all kind of businesses in the simulation world and integrating them into their portfolio, whereas their competitors usually stay with their core product/business. So ansys has very few if any big competitors with such a range of simulation products but there are plenty within each simulation niche (mechanical, CFD, electromagnetics...)
Let's not forget tha SNPS sells tools and services ( in my experience expensive rates but down to business consultants, which is rare enough to point out ) Ansys is an opportunity to expend their business on both front and get to merge the contact list from sales. Also a way to not be behind Siemens that has a whole lot of tools in both physical simulation and EDA after the mentor acquisition. Probably one of the most impactful merge in EE field in ages.
Some good points mentioned in the video but the main reason that enabled Synopsys to buy ANSYS for such a huge amount is the stock price of SNPS. The semiconductor industry has hit the peak and that can be seen in the growth of market cap of both SNPS and Cadence. The end users of the tools created by SNPS and CDNS also use CAE tools to do system level simulations like electronics cooling, aerothermal analysis of automotives and many more. So, both CDNS and SNPS saw this as an opportunity to use their stock price advantage to buy CAE companies. Infact CDNS approached ANSYS for acquisition which ANSYS then used as a bargaining chip to go to SNPS and get higher bid. It all worked out well for ANSYS in the end but I guess we'll see the effect of that huge price tag in the form of layoffs and license cost increases over the next year or two.
If not the most powerful engineering skill that can be leveraged through Ansys [Workbench] is the coupling of multiple physics analysis tools. For example: Electromagnetics, Mechanics and Fluid mechanics being coupled in a direct fluid structure magnetism coupling is huge! Ofc one has to get the time steps right, have equally good knowledge of alle the tools and physics behind it & still have the vision for the design. But I think coupling + design automation to a certain point alone with the models and tools we have atm will increase the performance of computer chips a lot in the next years. What a time to be alife!
I think the best way to do this is to combine photonics with electronics with the electronic layer on top and photonic feed mechanisms around the outside. Photonics while very sensitive in individual wavelengths to structural deformation is a problem, the interactions are exactly predictable based on temperature and so can be adapted to with a broader band signal. Then you get no heat produced by the underlying compute layer and all the heat readily offloaded in the electronic layer which serves as the I/O and memory. You can stack the photonics layers, well you have to at this point. Getting photonics small enough is the challenge and progress is being made.
One of my favorite advancements is the advancements of software that utilizes our best hardware to run very complex simulations. It's such a beneficial way to visualize and improve many different systems and aspects in a system. It's only limited to how accurate we can be with our measurements. Hense why I think we might be seeing the crisis in cosmology? Maybe we just don't have a good enough grasp on the immense scales throughout our universe and the way all of those immense scales interact with each other in the environment of space. Such as galactic filaments, entire galaxies, black holes, the list goes on and on. There's so many factors playing a part of very complex systems even in nebulas and so it only seems reasonable that we might not have a accurate enough grasp on these measurements to get accurate enough simulations when they do simulations of behaviors through out the universe.. Just a random thought I had
I used them both in my early engineering days - 2001. I saw license prices and was shocked. I personally hope though we will have a credible open source alternative in the future! With the fast progress in particle simulation for e.g. games, I think that should be possible.
Niche applications and a lot of man-hours to develop. I used salome-meca and it was a pain in the ass. Good if you want to use it for your PhD, but not for consulting/productivity/quick turnaround. Even when talking about parametric 3D CAD there are very few open source alternatives. In CFD there is open-foam, which is very powerful but it demands a lot from the user.
I was looking into using ansys for some software development. Sounded great until they wanted 45k$/year per license and were new to that industry :P At that price we developed our own software.
The simulation software is of questionable use, since thermodynamics relies on a significant statistical sample. Once you are down to micrometer size the laws of thermodynamics become "weird" and non-intuitive. I am not convinced that this , simulation needs, is the reason to buy ANSYS. But definitely a good purchase!
Makes you wonder if a new design could put a big old heat sink on BOTH sides of the CPUs ? A bit of a 'packaging' design change for the motherboard, an edge connnected 'Hole' in the board, with access to both sides of the chip to the Heat Sinks, fans, and fluids. Most amazing to me isn't the huge chips but the advancements in Low power processing. Less power, less heat. Redesigning chips to provide maximum computation with minimal heat would allow for more design options. Maybe even sell CPUs as a 'Block' instead of chip.
Right. Exposing top and bottom of chips to heat transfer has been done before, I've seen it most often for power components. Processors have a bit of an issue in that they need hundreds/thousands of pins, so the package size needs to increase a lot to push all of them to the perimeter, it can also hurt latency in some scenarios. Under a processor is also a great place for filtering capacitors, you needs to push those out too. Fortunately, most thermal constrained processors use the pins and bga balls as a heat sink - not all of them carry data lines, and the ones that carry power are split into many pins that are distributed across the package. Some have done it badly - see failure of bga graphics chips in laptops and gaming consoles around the 2010s. The thermal design for these has improved, some use gold plated vias between bga balls (like seen in the video 10:22) and active cooling with thin vapour chambers in mobile phones. If you can get the heat away fast enough, you can conduct trough the board sufficiently without resorting to more exotic packaging.
Ansys, 10K per license. Most of the studies done on Ansys are a huge waste of time and little more than a colour show. Sorry but after 35+ years as an engineer I really have little time for FEA like Ansys. It's a handy aid but really not worth the time & effort. 35 billion, wow, well done!
@@adityakulkarni4549He probably means that no FEA program, no matter how sophisticated, can be a substitute for thinking and properly understanding the problem. While that's true, I (IMHO) think he's a little bit too harsh. With today's computing power you can solve amazing problems.
@@alexandershendi7428 Simulations produce designs that can still fail in the real world. There's no substitute for proper testing and validation. Simulations only help when you know very little about the problems, or starting designs from scratch - which is almost never for an experienced engineer.
@@PaulSpades I wholeheartedly agree that validation and testing are always necessary. However simulations can help to narrow the design space. Otherwise you are stuck in a build->test->scrap->design cycle.
I am sorry but your description of FEA is quite far off the mark. Discretization (breaking up a domain into smaller units) has nothing to do with prediction being easier on the smaller units. In fact predictions even in extremely small units near corners for example, can be much harder than larger units in the middle of a domain. Discretization in FEA is necessitated by the nature of function approximation, namely piecewise polynomials. These piecewise polynomials are non-zero only over a small region, hence small geometrical units are a natural corollary. Where prediction is easy or difficult depends on the underlying physics, the geometry of the domain and the suitability of the functions we have choosen to approximate the solution.
Someone asked me to learn about FEA saying there's a lot of jobs for this, but without an engineering degree, and at my age it would be a waste of time to get one, it's pointless. I haven't really got a clue how FEA works.
Nah, multi-chip packages will also see their way into smartphones. They're packaging the RAM into the SoC package already, which some may argue already makes it a multi-chip package.
I worked for a direct competitor of ANSYS in the ‘90’s (SDRC) and the company I work for now OEMs the technology that ANSYS’ Minerva application is based on. Lot of memories in those old photos. There is precedent for a huge industrial company owning a CAE company - Siemens has owned NASTRAN and Dassault has owned ABACUS for a while now. FYI - Euler is pronounced “Oiler”, solder is “sodder” . . . but you nailed von Neumann 😊
Totally amazing! Talk about a formula with multiple variables...am I mistaken, or would this benefit from machine learning, eg LLMs? (just as I wrote this, Jon mentioned it) Synopsis, wow, I'm impressed. Same thing's happening with molecular design in pharmaceutical development, and all sorts of other material designing fields.....thanks Jon, I missed this one when it came out....
Great video! When you mentioned that 3DIC is unlikely to be implemented in smartphones, is it due to thermal issues? Because I thought stacking them can further miniaturize the whole system and thats why a lot of wearable/portable devices are looking at SiP. Thank you!
"Why Synopsys Bought Ansys (For $35 Billion)" - because they are using Synopsys stock like cash to buy Ansys and the stock market is at all time highs? Just a thought.
Wow, sold the company for the price of only 70 multiphysics licenses.
@@ashifarman4813He is Sarcastically mentioning the very expensive Multiphysics Software from Ansys.
I'm a mechanical engineer (independent contractor, so I have to pay out of pocket to rent software). I rented a stand-alone seat of ANSYS (structural + thermal + SolidWorks Plug-in) for a year and a half for a project, and it cost more than 2X as much as my car. It's powerful, but *man*, it's expensive.
That's every commercial license for a CAD (to a smaller degree), FEM, or CFD SW unfortunately
Some analysis can be executed in open source software - OpenFOAM and CalculiX.
There's free Finite Element Analysis software that's maintained by the Finnish Government called Elmer FEM. It's a bit wonky to get everything integrated into a workflow but it does all the things, and is free.
@@salmiakki5638 hey when your engineers cost 200k/yr, 50k is nothing for software that can double their efficiency
@@EuphoricDanbased Finland.
Ansys also owns Redhawk which is the industry-leading IR/EM analysis tool. This was a hole in their backend design flow that I'm sure Synopsys wanted to fill. Cadence has Voltus for this type of verification and now Synopsys has responded.
interesting!
what is ir em analysis?
They also bought AGI (formerly: Analytical Graphics Inc, now: Ansys Government Initiatives) a year or two ago, who track satellites and do geospatial simulations, and a few years back they created and spun off cesium (3D maps in web browsers).
@@LostieTrekieTechie - just to clarify, ANSYS did not spin off Cesium, AGI did the spin off as part of the (many) events involved in ANSYS acquiring AGI. Former AGI employee number 37 posting here.
Hopefully Ansys start to a make workable programs. The programs are too famoust to corrupt files and delete simulation results.
I have even wrote a parady ”the project must go on” from the ”the show must on ”because of Ansys
As a Mech. Engineer who has mostly interacted with Ansys as a general Mechanical FEM software (Stresses, Eigenfrequencies) its a nice suprise to see software from the CAE-Sector show up on your channel. A big Advantage of ANsys as a simulation software I have seen a lot of people proclaim over other CFD/FEM/Thermal programs is that it has a lot of different modules, and is steadily connecting them (e.g.: Heat Transfer + the thermal impact on material properties under heat). Have you considered covering general CAD (i.e. Modelling) Software? Might not be highly relevant to your main topic of semiconductors, but I think it would be nice to see a coverage of the 2D->2.5D->3D evolution of Software, and the modern trends (integrated FEM (aka what you tend to call CAE in this video), CAD-CAM...).
I'm a fan of silicon graphics computers and this is the perfect topic to bring up that company with those types of machines, those machines are so cool.😊
@@zilog1 There are dozens of us! Something so incredibly fascinating about those old monoliths.
Ansys is segmented and the ui is cumbersome. Siemens got better integration
I put a bid in too, but I guess they didn’t want my 2000 dollars 😤
Not enough try again
@@mustaphaben29212001?
Skill issue
They were probably hesitant with giving you a license at that price.
they were too intimidated, you gotta send them chocolates to break the tension
Amazing growth for Ansys: from $35 million to $35 billion in 30 years.
Heat transfer is being taken very seriously.
If everyone is paid fatly, that’s some nice wealth redistribution towards employees then. I often think about it with my company: Every buck I save on employees produces 10 bucks on valuation of my shares.
(Real story: I try to differentiate employees who f me over from those who are genuine) (and now we have an absolutely awesome team).
not heat only. semi and electrical analysis brings more than half of rev
Its more than that. Their softwares are used for basically every single product and most steps of development. HFSS is used to simulate EM behaviour of devices for everything from small RF chips to wifi enabled coffee machines. They have SPEOS and Lumerical to simulate photonic chips up to how light behaves inside a newly designed car, the list ist pretty long.
Wow, $35 billion for ANSYS? The per seat cost was already astronomical and I guess it will get even higher now...
Hey, it's specialized software which is extremely useful for a very broad number of problems, and that kind of utility costs!
We definitely need more videos of this caliber and expertise on RUclips! They are truly valuable and deserving of recognition. 👍❤️
"Fluids are not like solids"❤🤯
As somebody who works in a semiconductor company and uses Ansys tools daily, I have to say this video summarizes things 100% accurately.
I find it very inspiring to watch any of your semiconductor videos as I am currently studying electrical engineering and we have a course that is related to circuit design (mostly analog) on a silicon level. It's really great to have a peak into cutting edge from time to time! It's amazing, your videos are both entertaining and motivate me to study... Insane combo!
I was just running a CFD simulation on ansys and you uploaded this video, so nice to see an youtuber that I follow talking about something that I use everyday.
You had me at "In the 1960s, John Swanson was working at the Westinghouse Astronuclear Lab in Pittsburgh.He went there to work on nuclear-fueled rockets with the goal of going to Mars - the NERVA " What an incredible series of words, I'd love to see what Vonnegut/Clarke/Asimov/Gibson/Haldeman/Heinlein would do with that as the first line of a short story.
Nice observation from you, tech corporates history gets really engaging at times.
I'm impressed with the restraint of calling it NERVA, which I'm sure they did just so they could call the full working version MINERVA
Correct pronunciation of Euler is to read it as “Oiler”
Based on the word Eule or owl in German. Thanks for the video!
I will now pronounce it as "owler".
@@jpkotta °v°
I have to thank you. Not sure why but you caused me to search for a similar words pronounciation. I googled it on the phone and discovered google not only has pronunciation (which I expected) but it actually will listen to you say the word and write out your mistakes. Thanks!
Nice you can say it that way in German then. In English, it is pronounced you-ler.
I work on Ansys everyday but never knew the history
another great video. If only tech review youtubers would realize how tough it is to design and build chips, maybe they would shut up about performance benchmarks. fact that AMD and Apple can keep increasing performance every year is an amazing achievement that people take fore granted
This critical misconception has permeated everything in physics and engineering and is so damaging: "the complex non-linear equations that _govern_ fluid dynamics". no. just no. the flavor of algebra, and equations we express in that specific algebra, that we've invented, are just imprecise _models_ of fluid dynamics.
This isn't a subtle difference in language, it's a fundamental and enormously damaging misconception that holds back human progress.
Not only that, but the underlying mathematical differences between solid modeling and fluid modeling mean that (in general) any model that includes fluid-structure interaction cannot be arbitrarily scaled. That's why wind tunnels are still being built.
Also, I find it somewhat ironic that ANSYS (the structural FEA application) is one of those applications that does not easily take advantage of parallel computing, given that the company was purchased with the goal of aiding the chiplet design process.
There are software like COMSOL that operate on the constitutive relationships: solving the first-principles equations inside mesh cells and the boundary relationships between them. Monte Carlo simulations (more common in research, but do exist in some industries) also operate similarly. Confusing these with the sort of CFD simulation that ANSYS does in a genera-audience video like this does not hold back human progress. The humans making progress use all sorts of different tools for their simulation. There is even a third archetype that gets some use, and that's before we get into the FFT/ IFT versions of all the above.
All it takes to accelerate human progress….
Is grab the bull by the horns, and use your calculus skills…
More time spent complaining about what another person has done, is one more day wasted not getting to the place you want to be…
You can be a billionaire… get on your horse and ride!
😃
Aye. People seem to think the imprecise models they use actually drive reality. And then complain when reality disagrees. Physics approximates reality, nothing "governs" nature.
In electromagnetics there is nothing imprecise.
How do you pump out quality videos so fast?!
Listen to how tired his voice sounds in this video. That's how. Our boi just trucks through the pain
@micro-organism-pv5gd I don't think the voice-over is the time-consuming part. The research is the impressive part
man do I not miss the math of my fluid dynamics class. never take 2 programming classes and 2 physics classes in 1 quarter unless you are genuinely obsessed with both 😂
Navier stokes equation and it's solution. 😅
It’s pronounced “Oiler” not “Euler”
He did that just to make sure we were listening.
If you want to be pedantic you might as well recommend the real pronunciation
So is Euclidean pronounced "oiclidean?"
@@codejunki567 Euclid, much to his shame, wasn't a German speaker. So his name isn't said with a German pronunciation.
@guaposneeze Cool, just curious, never heard that before.
I'm a Chemical engineer who has worked with ANSYS fluent & built some structural components which we later tested out. It is an awesome software which works great & I am really amazed to see the fluid simulations. We had Computational Fluid Dynamics as a subject which deals with various parts as also explained in the video.
Ansys, steep learning curve, every time. Excellent if you have the time and the crashes don't demoralize you too much. Great video.
Sounds like you have used Ansys Wb?
Somewhere in the bowels of each automotive company, someone has the data on how they know all those plastic parts on your car's engine will turn brittle and fail. All modelled in Ansys I'm sure.
Actually not. Dassault, MSC, Altair and Siemens are stronger in automotive
Thank you for the synopsis!
Another note on anti-monopoly issues: given how crucial thermal simulation is for advanced packaging, Synopsis could shut off this capability for competitors such as Cadence
I don't know if there are competitors to Ansys that could fill this role as a complementary solution to other EDA vendors
I don't think that would fly in most of the world with respect to anti competitive practices. The only thing that comes close is that I've heard (but not myself seen) that Cadence, for example, gives discounts for EMX if cadence is always your first supplier for any tool they have in their portfolio. In other words, if you also want to use Keysight RFPro, you pay more for EMX.
Siemens went the other way. They have a whole range of CAE tools including thermal and the acquired Mentor to add the EDA side of things
Cadence already has Celsius, and they bought AWR for the RFIC and EM tools.
Why no one is talking about Comsol?
@@mahussain1Comsol probably works great for truly coupled problem like RF -> heat -> deformation -> RF but it doesn’t target any specific problem well.
I had no idea ansys was born from NERVA project... wow!
"the poor hornet", this is the moment I realized Asianometry is unhinged. ;D
This can be summarized as virtually prototyping a design from “silicon to system.” Imagine a design flow that starts at a chip and goes all the way through package to pcb to automobile.
Thanks!
I actually did my PhD on optimizing MCM packages. Balancing heat density and inter-chip latency due to distance was a huge tradeoff I had to consider. I think a great example of this now is looking at Intel's Sapphire and Emerald Rapids CPUs with the cores spread out away from the EMIB interconnects is a great example of this need.
One thing missing from academia teaching of tech history is biography, which you cover. It is amazing how important agency and human decision is for change.
Not even a mention? The core design of eUV and every laser, imaging, illumination, etc. optical application is designed and analyzed on Code-V (Synopsys) and Zemax (Ansys). Now, Synopsys and Synopsys. As a Zemax user, I am concerned what will happen to Zemax.
Zemax is trash tho
Lumerical and Rsoft will also become one company. The entire optics simulation industry will become highly concentrated. I am afraid they will stop rolling out features because the market is locked down. Of course, price hikes are a given.
Thank you and thanks to every asionmetrics Patreons.
A software company so useful and revolutionary as ANSYS was sold cheaper than Tweeter...
Inb4: advertising, data mining, etc. I know, but still...
Because twitter was bought by a rather low intelligence individual
Musk over paying for something doesn't really say much about ANSYS
Ansys workbench is quite useful: you can do stress, CFD and even EM given the write package was downloaded.
Also can corrupt your project easily and destroy everything. It is famous for this 😂
@@projecttitanium-slowishdriver ahh, something to be mindful of. Thanks for the info👍
Ansys also has Helic which a electro magnetic simulator useful for on die structures like large interconnects and inductors.
I only clicked on this because I build powerful servers on a daily basis and CFD is one of the more useful ways of really leveraging all that computing power. Simulating thermodynamics in processors using processors is wonderful, we went full circle ;)
I think they bought it cause the names both end in "sys"
"In any system of energy, Control is what consumes energy the most.
No energy store holds enough energy to extract an amount of energy equal to the total energy it stores.
No system of energy can deliver sum useful energy in excess of the total energy put into constructing it.
This universal truth applies to all systems.
Energy, like time, flows from past to future"(2017).
This package design is definitely the intermediate future thing… The main problems will definitely be focused around cooling. Reminds me of the dawn of high rises in cities, we built up for specialization and climate control as we built upwards it was a learning process. The typical heat sink and heat dissipator/IHS isn’t going to work as well as it used to anymore..
I remember implementing Finite Elephant Method in school - such a beautiful algorithm. I wanted to compliment you on your quality subtitles, we can really tell the care that went into getting them right.
You should get your money back
Finally a video that “takes place” in Pittsburgh! My home town. My friend works at Ansys and I work just down the road!
I was thrown off a bit yesterday when it was announced Renesas was buying Altium.
Strange place for me to get my news - YT comments. Thanks for the news. Hopefully Renesas doesn't screw up Altium.
Cant screw up something that is already f-d.
@@VEC7ORltFor us, it's been better than a sharp stick in the eye.
@@ChiefBridgeFuser Altium 24 has been mostly stable ish for me. Though I just got a runtime error 231.
After a 20 year love affair with Altium, our team is moving on. Kicad is pretty good these days. I'd rather stick with Altium but eventually I will be forced to get with the program.
Awesome that you mention Welch! He's the father of staggered grids. Staggered grids are the future of CFD. If only Perot will figure out his 3D mimetic interpolation for polyhedrals.
Ansys's HFSS software is also something of a modern marvel IMO - definitely one of my favorite suites out there.
Nice video...I have always enjoyed your videos....Ansys also has Lumerical FDTD solutions...which are used for optical interconnects, etc. Eventually, all interconnects on these chips might be replaced by light based interconnects to reduced heat...
Just to put that in context, a few years back, IBM bought Redhat for US $37 bil.
And Apple is worth (I googled this before typing it) US $2.4 Trillion.
We are talking about a substantial purchase, but not insane by any means.
That's a sweet thumbnail if i ever saw one
I will say that maybe $35 billion is too much, it represents 17 years of Ansys revenue, not earnings. They are growing pretty fast: 8.33% but the net Income is less than $500M. This will kind of problematic numbers, normally leading to employee reduction and cost-cutting as the C-level executives try to justify the purchase. Clearly letting go of talent and cut R&D and workers' satisfaction is not going to increase the value of ANSY, but it could get the C-Lebeks bonuses. The other way around it, normally, is trying to get a monopoly and increase the prices, a lot. We need to change the legal framework to allow both the stockholders and the courts to make the C-Levels responsible for dealing with it, and force them to show that they make financial sense under penalty. $35 billion is around 41% of Synopsys' market cap.
IA will do FEA and CFD in seconds without any calculations.
Your analysis, questioning the rationale for buying Ansys for $35B is astute.. There must be at least some semblance of a counter-argument which persuaded both the execs, as well as other financial stakeholders that claims to warrant the purchase price. I doubt that rationale would be to monopolize the product, as that surely would result in years of costly litigation, with an eventual loss.
My WAG (Wild Ass Guess) is that Ansys will continue, but that the engineers may also have an ambitious 'skunkworks' project for components that uniquely serve the chip industry. For example,
#1) With sizes shrinking to a couple nanometers and smaller, there may be quantum effects that are yet to be modeled.
#2) IBM and others have begun experimenting with using light (nanometer sized LEDs and photosensors in a 3d matrix to fan-out
signals. So, perhaps the package could emulate physics inclusive of optics, which at that speed and scale are nonlinear.
#3) I've read of advantages to pulsing fluids instead of continuous flow (e.g. heartbeat). That may be applicable at nano scale.
#4) I once did a little work on MEMS. While Ansys can already simulate MEMS, including multi-physics, such as electrical input/output
of piezoelectric materials, there may be much more to this, especially at nanoscale.
I may be overly optimistic, but my WAG is that there's some futuristic prospect to warrant such a huge gamble, something better than just upping the price for the product's use, or monopolizing it.
The lever isn’t cost here. I predict that they will jack up the cost licenses massively. Ansys is already so entrenched in many design processes, companies will pay up.
As if Ansys is the only company in the field CFD. I think that they were searching for other options and this boosted the price for buying Ansys, because the others were not for sale.
🎵You say soulder, I say sodder; let’s call the whole thing off! 🎵
As a CAE engineer, I would argue that your perspective may be biased. The semiconductor industry is relatively small compared to the billions these companies earn by charging hundreds of thousands in licensing fees each year from universities and R&D labs.
ANSYS, the world leader in CAE, has changed ownership multiple times and has acquired numerous CAE companies to gain access to their solvers, such as Fluent and ICEM for CFD, and LS-DYNA for automotive crash analysis. Now, they are introducing ADINA, which uses the same technology as LS-DYNA, potentially threatening its existence.
Moreover, ANSYS faces stiff competition from Dassault's Abaqus, Altair, MSC, and COMSOL. Interestingly, in the semiconductor and laser R&D sectors, COMSOL is widely used, while ANSYS is predominantly used in mechanical systems and HVAC.
Did you forget Simcenter Star CCM+ The industry standard
I don't think Synopsys must be doing too bad with revenue when half the projects I've been in have talks of adding a couple hundreds of licenses of VCS mid project to keep up with the regression eating up all of them.
"The poor hornet."
Ok, now you lost me.
Micron had kgd (known good die)process where we would take good die and place in temp package and do full burn in testing, then removed the die from package and sold it for early ceramic packaged MCM, multi chip modules. Good money in KGD back then.
Ansys, whilst powerful, reminds me of using a piece of software from 2 decades ago. They acquired lots of companies and simply bolted it all together, and unfrotunately the user experience reflects this. I wish it had a UI/UX like COMSOL. Unfortuantely for COMSOL it falls over when trying to do anythign remotely complex.
As a Mech E with lots of experience in this field, it was nice to watch a video on something closer to home!
As an EE who is just starting to use COMSOL, I wish their UI were more intuitive, all the options for your solution are buried 5 layers deep. I agree that Ansys' bolting stuff together isn't great either: taking a mechanical simulation and then doing an RF simulation on it can get annoying. But I found that using a single tool from them wasn't too bad to set up and do optometric sweeps.
I think the caveat with comsol is that it is so flexible it lets you setup simulation in inefficient ways that don't lend themselves to complex problems; More planning is required when setting up boundary conditions and even solving with comsol over ansys, but on the flip side comsol can simulate things ansys can't.
Without having used abaqus a lot the impression I get is they have stuck to their core product and refined it.
Whereas ansys has been shopping companies around like crazy to have a piece of software for every application without any intention to refine and perfect anything (for example I fear for LS-Dyna development)
Ansys is likely up there close to the ranks of ASML, some company not in the common man's consciousness but responsible for a huge portion of everything we have and enjoy today. It's stupid powerful and absolutely essential for some black magic engineering (e.g. RF)
Ansys has to face stiffer competition than ASML. They don't have any revolutionary technology unknown to anyone else. They have a history of buying all kind of businesses in the simulation world and integrating them into their portfolio, whereas their competitors usually stay with their core product/business. So ansys has very few if any big competitors with such a range of simulation products but there are plenty within each simulation niche (mechanical, CFD, electromagnetics...)
Let's not forget tha SNPS sells tools and services ( in my experience expensive rates but down to business consultants, which is rare enough to point out )
Ansys is an opportunity to expend their business on both front and get to merge the contact list from sales.
Also a way to not be behind Siemens that has a whole lot of tools in both physical simulation and EDA after the mentor acquisition.
Probably one of the most impactful merge in EE field in ages.
we are gonna need solid copper vertical fins going through the whole 3D chip
I used to work for Synopsys about 10 years ago, never imagined they would have $35 billion dollars lying around to make this kind of acquisition. Wow!
I was just down at the location where the Astronuclear lab was, Large PA. Company there does EMC testing, among other things.
Very nice video and great research. By the way, Euler is pronounced Oiler.
Some good points mentioned in the video but the main reason that enabled Synopsys to buy ANSYS for such a huge amount is the stock price of SNPS. The semiconductor industry has hit the peak and that can be seen in the growth of market cap of both SNPS and Cadence. The end users of the tools created by SNPS and CDNS also use CAE tools to do system level simulations like electronics cooling, aerothermal analysis of automotives and many more. So, both CDNS and SNPS saw this as an opportunity to use their stock price advantage to buy CAE companies. Infact CDNS approached ANSYS for acquisition which ANSYS then used as a bargaining chip to go to SNPS and get higher bid. It all worked out well for ANSYS in the end but I guess we'll see the effect of that huge price tag in the form of layoffs and license cost increases over the next year or two.
If not the most powerful engineering skill that can be leveraged through Ansys [Workbench] is the coupling of multiple physics analysis tools. For example: Electromagnetics, Mechanics and Fluid mechanics being coupled in a direct fluid structure magnetism coupling is huge! Ofc one has to get the time steps right, have equally good knowledge of alle the tools and physics behind it & still have the vision for the design. But I think coupling + design automation to a certain point alone with the models and tools we have atm will increase the performance of computer chips a lot in the next years. What a time to be alife!
i dont think ive ever seen someone sympathize with the hornet over bees before
As someone in the industry, I have always heard via pronounced ‘vee uh’, not ‘vy uh’ . Very good synopsis. Thanks
No mention of the Navier-Stokes existence and smoothness problem and its associated US$1 million prize? 🙂
The entire point of CFD is engineers giving up on that problem and resorting to numerical integration
@@dsdy1205 More like the mathematicians. The engineers need to solve problems in the real world and they use whatever they can to get results.
Ansys became a sponsor for the CAD club in our university a week ago
and Damn, I see Ansys everywhere in my feed nowadays
Waiting for a fab to create silicon-copper-water heat pipes / vapor chambers to stuff in between layers to remove heat.
I think the best way to do this is to combine photonics with electronics with the electronic layer on top and photonic feed mechanisms around the outside. Photonics while very sensitive in individual wavelengths to structural deformation is a problem, the interactions are exactly predictable based on temperature and so can be adapted to with a broader band signal. Then you get no heat produced by the underlying compute layer and all the heat readily offloaded in the electronic layer which serves as the I/O and memory. You can stack the photonics layers, well you have to at this point. Getting photonics small enough is the challenge and progress is being made.
One of my favorite advancements is the advancements of software that utilizes our best hardware to run very complex simulations. It's such a beneficial way to visualize and improve many different systems and aspects in a system. It's only limited to how accurate we can be with our measurements. Hense why I think we might be seeing the crisis in cosmology? Maybe we just don't have a good enough grasp on the immense scales throughout our universe and the way all of those immense scales interact with each other in the environment of space. Such as galactic filaments, entire galaxies, black holes, the list goes on and on. There's so many factors playing a part of very complex systems even in nebulas and so it only seems reasonable that we might not have a accurate enough grasp on these measurements to get accurate enough simulations when they do simulations of behaviors through out the universe.. Just a random thought I had
6:01 Leonard's last name is pronounced like 'oiler'; the Houston Oilers were named in honor of his work.
I used them both in my early engineering days - 2001. I saw license prices and was shocked.
I personally hope though we will have a credible open source alternative in the future! With the fast progress in particle simulation for e.g. games, I think that should be possible.
Niche applications and a lot of man-hours to develop. I used salome-meca and it was a pain in the ass. Good if you want to use it for your PhD, but not for consulting/productivity/quick turnaround.
Even when talking about parametric 3D CAD there are very few open source alternatives.
In CFD there is open-foam, which is very powerful but it demands a lot from the user.
I was looking into using ansys for some software development. Sounded great until they wanted 45k$/year per license and were new to that industry :P At that price we developed our own software.
At 6:08, Euler is pronounced “oiler” no “yueler”
I’m glad he didn’t say Bueller ;)
The simulation software is of questionable use, since thermodynamics relies on a significant statistical sample. Once you are down to micrometer size the laws of thermodynamics become "weird" and non-intuitive. I am not convinced that this , simulation needs, is the reason to buy ANSYS. But definitely a good purchase!
Everyone support this creator! That way we can get more awesome videos like this one!
Would live to see a video on the history of these companies. How they grew, and the people behind the scenes.
Makes you wonder if a new design could put a big old heat sink on BOTH sides of the CPUs ? A bit of a 'packaging' design change for the motherboard, an edge connnected 'Hole' in the board, with access to both sides of the chip to the Heat Sinks, fans, and fluids. Most amazing to me isn't the huge chips but the advancements in Low power processing. Less power, less heat. Redesigning chips to provide maximum computation with minimal heat would allow for more design options. Maybe even sell CPUs as a 'Block' instead of chip.
Right. Exposing top and bottom of chips to heat transfer has been done before, I've seen it most often for power components. Processors have a bit of an issue in that they need hundreds/thousands of pins, so the package size needs to increase a lot to push all of them to the perimeter, it can also hurt latency in some scenarios. Under a processor is also a great place for filtering capacitors, you needs to push those out too.
Fortunately, most thermal constrained processors use the pins and bga balls as a heat sink - not all of them carry data lines, and the ones that carry power are split into many pins that are distributed across the package. Some have done it badly - see failure of bga graphics chips in laptops and gaming consoles around the 2010s. The thermal design for these has improved, some use gold plated vias between bga balls (like seen in the video 10:22) and active cooling with thin vapour chambers in mobile phones. If you can get the heat away fast enough, you can conduct trough the board sufficiently without resorting to more exotic packaging.
I wonder if we can draw any parallels with Renesas buying Altium
Fea is used today to value engineer the newer cars so all value rests with manufacturer and the customer gets the shaft.
Ansys, 10K per license. Most of the studies done on Ansys are a huge waste of time and little more than a colour show. Sorry but after 35+ years as an engineer I really have little time for FEA like Ansys. It's a handy aid but really not worth the time & effort. 35 billion, wow, well done!
Can you elaborate?
@@adityakulkarni4549He probably means that no FEA program, no matter how sophisticated, can be a substitute for thinking and properly understanding the problem. While that's true, I (IMHO) think he's a little bit too harsh. With today's computing power you can solve amazing problems.
@@alexandershendi7428 Simulations produce designs that can still fail in the real world. There's no substitute for proper testing and validation. Simulations only help when you know very little about the problems, or starting designs from scratch - which is almost never for an experienced engineer.
@@PaulSpades I wholeheartedly agree that validation and testing are always necessary. However simulations can help to narrow the design space. Otherwise you are stuck in a build->test->scrap->design cycle.
Well it depends on the industry. It's not the same to design gym equipment than nuclear reactors.
I am sorry but your description of FEA is quite far off the mark. Discretization (breaking up a domain into smaller units) has nothing to do with prediction being easier on the smaller units. In fact predictions even in extremely small units near corners for example, can be much harder than larger units in the middle of a domain. Discretization in FEA is necessitated by the nature of function approximation, namely piecewise polynomials. These piecewise polynomials are non-zero only over a small region, hence small geometrical units are a natural corollary. Where prediction is easy or difficult depends on the underlying physics, the geometry of the domain and the suitability of the functions we have choosen to approximate the solution.
Someone asked me to learn about FEA saying there's a lot of jobs for this, but without an engineering degree, and at my age it would be a waste of time to get one, it's pointless. I haven't really got a clue how FEA works.
Nah, multi-chip packages will also see their way into smartphones. They're packaging the RAM into the SoC package already, which some may argue already makes it a multi-chip package.
Hi Jon, great video.
I am always wondering what the differences are between Synopsys and Cadences.
To me they seem to be doing identical things.
Great vids man
I worked for a direct competitor of ANSYS in the ‘90’s (SDRC) and the company I work for now OEMs the technology that ANSYS’ Minerva application is based on. Lot of memories in those old photos.
There is precedent for a huge industrial company owning a CAE company - Siemens has owned NASTRAN and Dassault has owned ABACUS for a while now.
FYI - Euler is pronounced “Oiler”, solder is “sodder” . . . but you nailed von Neumann 😊
Totally amazing! Talk about a formula with multiple variables...am I mistaken, or would this benefit from machine learning, eg LLMs? (just as I wrote this, Jon mentioned it)
Synopsis, wow, I'm impressed. Same thing's happening with molecular design in pharmaceutical development, and all sorts of other material designing fields.....thanks Jon, I missed this one when it came out....
love these shorter videos.
Great video! When you mentioned that 3DIC is unlikely to be implemented in smartphones, is it due to thermal issues? Because I thought stacking them can further miniaturize the whole system and thats why a lot of wearable/portable devices are looking at SiP. Thank you!
literally no one:
asianometry: poor hornet
11:12 the part that you came for
"Why Synopsys Bought Ansys (For $35 Billion)" - because they are using Synopsys stock like cash to buy Ansys and the stock market is at all time highs? Just a thought.
Risc architecture is gonna change everything!
Synopsys also bought Imperas, probably for their RISC-V stuff.
There, there. It did. In the '90s.
I firmly believe after seeing sora this is only a few years away from being worthless
Train an LLM on all those three letter design elements and you won't need Ansys.
Everyone talks about ANSYS but Synopsys also bought Imperas.
this might as well be an opportunity for Dassault Systemes Electromagnetics simulations to gain some market share
@6:05 references and resources go there
Thank you it was semi detailed ❤