NVIDIA Chief Scientist Reaffirms Huang's Law

[Dr. Dro]

Nope CPU's have gone even higher, the top chip on desktops 10 years back was the 5960x(?) & this year(next?) we'll have a TR with probably up to 96 cores. And it's definitely more than 12x faster than Intel's best HEDT chips back then, even if you take the top server chips they now top out at 128c/256t for AMD. In fact you could argue that CPU's have progressed far more, in part of course due to the stagnation with *dozer & Intel deciding to milk quad cores for at least half a decade!

The top Ivy bridge Xeon chips topped out at 12 cores, so again vastly lower.

Threadripper Pro is not a desktop processor, Threadripper as a consumer grade CPU died with the 3990X.

But even if you account for the market niche and multi-die CPUs (which really are multiple CPUs in one package), I don't think IPC hasn't gone up a full 10x from Haswell to Raptor Cove (2013-2023). Operating frequencies increased greatly in the interim as well.

Core counts went from 18 (Haswell-EP) to basically around 128, so not a full 10x increase. IPC must have gone up around 6 times higher, and also an extra GHz on average, but I guess that's about it.

Might have if you compare Piledriver to Zen 4 but AMD CPUs were hopeless garbage until Ryzen came out. Could be worth looking at sometime with some real data, but we all remember how 1st gen Core i7 CPUs made sport of FX.

Still GPUs have easily outpaced this growth. GK110 to AD102 is one hell of a leap.

The naked narcissism inside nGreedia must be absolutely awful to have to navigate through.

Ah, yes, I'm sure "nGreedia" engineers are just jumping at the opportunity to work at better companies, such as AMD, perhaps?

For gaming, GPUs have gotten faster by about 10 times in the last 10 years. Ten years ago, the fastest GPUs were the 780 Ti and the 290X. The performance improvement from the 780 Ti to the 4090 at 4K is about 10 times. The table below uses TPU's reviews at 4K for the GTX 1060, 1080 Ti, RTX 3080, and RTX 4090 respectively.

GTX 780 Ti to GTX 970	GTX 970 to GTX 1080 Ti	GTX 1080 Ti to RTX 3080	RTX 3080 to RTX 4090
85/83	1/0.36	100/53	190/99

Multiplying all the speedups gives 10.3 which isn't too far off the multi-threaded performance increase for CPUs in that time. Anandtech's CPU bench can be used to compare the 4770k and the 7950X. There are common applications where the 7950X is as much as 9 times faster than the 4770K and these applications don't leverage any instructions unique to the newer processor such as AVX-512. I haven't used the 13900K because their database doesn't have numbers for any Intel CPUs faster than the 12900K.
View attachment 316005
View attachment 316006
Rather than blaming CPU designers, you should be asking game engine developers why their engines are unable to utilize these CPUs efficiently.

I'm saddened that Bill Dally is misrepresenting TSMC's contribution to these gains. The 28nm to 5 nm transition isn't worth only a 2.5 times increase in GPU resources. From the Titan X to AD102, clock speeds have increased by nearly 2.5 times and the GPU has 6 times more FP32 flops per clock. That is a 15 fold increase in compute solely related to the process. We shouldn't ignore the work done by Nvidia's engineers, but if we take his claim at face value, then a 28 nm 4090 would be only 2.5 times slower than the actual 4090 which is patently ridiculous.

You're also accounting shipping products (and at a relatively low weight class) to normalize for performance, the comparison in progress should IMHO be done comparing fully enabled and endowed processors that are configured for their fullest performance, perhaps normalized for frequency to accurately measure improvements at an architectural level. We don't even have such a product available to the public for Ada Lovelace yet.

 

[Wirko]

Nope CPU's have gone even higher, the top chip on desktops 10 years back was the 5960x(?) & this year(next?) we'll have a TR with probably up to 96 cores. And it's definitely more than 12x faster than Intel's best HEDT chips back then, even if you take the top server chips they now top out at 128c/256t for AMD. In fact you could argue that CPU's have progressed far more, in part of course due to the stagnation with *dozer & Intel deciding to milk quad cores for at least half a decade!

The top Ivy bridge Xeon chips topped out at 12 cores, so again vastly lower.

Also think about memory bandwidth, it's been lagging all along. The 5960X has four channels of DDR3-2133, now we are at 3x the transfer rate and 4 or 8 or 12 channels, depending on the segment.

 

[TheoneandonlyMrK]

Here’s some fodder for all of you

Huang's law - Wikipedia

en.m.wikipedia.org

Genius grade shit

Do these fools not know what the first Graphics processor was, as I said on the shoulders of giants shouting I'm biggest, Noice.

Huang's ego knows no bounds him musk and that other 7#@7 should have a cage match, last man alive takes all with gates as referee hopefully then suicide bombed out of my world.

 

[Dr. Dro]

Genius grade shit

Do these fools not know what the first Graphics processor was, as I said on the shoulders of giants shouting I'm biggest, Noice.

Huang's ego knows no bounds him musk and that other 7#@7 should have a cage match, last man alive takes all with gates as referee hopefully then suicide bombed out of my world.

The first fully programmable and featured GPU... With no reliance on 2D accelerator cards, full 3D support, transform and lighting and some form of rudimentary compute... Hmmm couldn't be the GeForce 256

 

[TheoneandonlyMrK]

The first fully programmable and featured GPU... With no reliance on 2D accelerator cards, full 3D support, transform and lighting and some form of rudimentary compute... Hmmm couldn't be the GeForce 256

The first Graphics output processor was the CPU, the first Dgpu was Intel, the 256 wasn't the first dgpu just the first with Hardware transform and lighting.

So you're random fact is irrelevant.

1986, Intel introduced the 82786 chip, the first discrete graphics coprocessor. The 82786 integrates a graphics processor, a display processor with a CRT controller, a bus interface unit (BIU), and its DRAM/VRAM controller.

 

[R0H1T]

Threadripper Pro is not a desktop processor, Threadripper as a consumer grade CPU died with the 3990X.

It is a desktop processor, just that AMD introduced a super tier HEDT 2 years back. Intel had Xeon E3(?) chips doing something similar, & if we're sticking to that definition then 5960x was also HEDT, Intel offered MSDT quad cores till Zen debuted ~ heck they only had "dual core" i7 ULV till the 8th gen chips were released

So if we're talking MSDT then you have quad cores (2700k/3770k) vs 13900k/14900k on the Intel side & for AMD 7950x ~ even then they're at least 10-15x faster in lots of tasks. Talking about MT performance of course.

 

[Dr. Dro]

It is a desktop processor, just that AMD introduced a super tier HEDT 2 years back. Intel had Xeon E3(?) chips doing something similar, & if we're sticking to that definition then 5960x was also HEDT, Intel offered MSDT quad cores till Zen debuted ~ heck they only had "dual core" i7 ULV till the 8th gen chips were released

So if we're talking MSDT then you have quad cores (2700k/3770k) vs 13900k/14900k on the Intel side & for AMD 7950x ~ even then they're at least 10-15x faster in lots of tasks. Talking about MT performance of course.

They're not even available in the DIY channel afaik, TR Pros are quite literally workstation processors and are marketed as such. You're probably thinking of the Xeon E5-2687W which serviced a similar niche: server platform with boosted clocks aimed at high performance workstations. The E3 were Xeon branded counterparts of same regular desktop processors.

Sandy Bridge is older than 10 years old by now. It released in early 2011, so you'd have to stretch it even further, not to mention you're also compounding all frequency increases and node improvements since that.

To calculate 10 years from now you'd have to take Haswell as a base and we already had 18 core Haswell chips. Their frequency were in the low 2 GHz's range as well, so compound the frequency gains the IPC and all, 8 to 10x seems about as reasonable as we get in MT, and that's bringing all of the improvements we've had since into account: from the weakest link to the strongest one.

I still believe GPUs have outperformed CPUs in this interim, from an evolutionary perspective. But CPUs also rely on more external factors, such as memory, motherboard, exploit fixes and such.

 

[AnotherReader]

Threadripper Pro is not a desktop processor, Threadripper as a consumer grade CPU died with the 3990X.

But even if you account for the market niche and multi-die CPUs (which really are multiple CPUs in one package), I don't think IPC hasn't gone up a full 10x from Haswell to Raptor Cove (2013-2023). Operating frequencies increased greatly in the interim as well.

Core counts went from 18 (Haswell-EP) to basically around 128, so not a full 10x increase. IPC must have gone up around 6 times higher, and also an extra GHz on average, but I guess that's about it.

Might have if you compare Piledriver to Zen 4 but AMD CPUs were hopeless garbage until Ryzen came out. Could be worth looking at sometime with some real data, but we all remember how 1st gen Core i7 CPUs made sport of FX.

Still GPUs have easily outpaced this growth. GK110 to AD102 is one hell of a leap.



Ah, yes, I'm sure "nGreedia" engineers are just jumping at the opportunity to work at better companies, such as AMD, perhaps?



You're also accounting shipping products (and at a relatively low weight class) to normalize for performance, the comparison in progress should IMHO be done comparing fully enabled and endowed processors that are configured for their fullest performance, perhaps normalized for frequency to accurately measure improvements at an architectural level. We don't even have such a product available to the public for Ada Lovelace yet.

I can only use what we have. We don't have any fully enabled AD102 to compare for gaming performance. The products are comparable: GK110 has a 561 mm^2 die while the AD102 has a 609 mm^2 die. Besides, the 780 Ti had greater overclocking headroom than Ada so I gather that it would even out in the end. As far as the CPUs are concerned, the 4770K has a die size of 177 mm^2 while the 7950X has 132.6 mm^2 devoted to the CPU with another 117.8 mm^2 for the IO die. We could compare the 15 core Xeon E7-4890 v2 to the 128 core EPYC 9754 which has comparable clock speeds. Unfortunately, I can't find benchmarks that compare the two, but the closest equivalent is the Xeon E7-4850 v4: a 16 core Broadwell. That result is for an eight socket server so we would have to divide the result by 8 to get the result for a single socket: 55 vs 981 for the EPYC 9754. A fairer comparison would use an older CPU like Sandybridge or even Westmere because Ivy Bridge used a much better process than the 780 Ti and the CPU would look even better then. While normalizing for frequency is popular, it doesn't really make sense as it penalizes the faster product whose latencies for instructions and cache accesses are a consequence of the higher clock.

Making GPUs faster is simpler than making CPUs faster as each GPU core is unbelievably dumb. Even a 486 is smarter than an Ada SMX. GPUs rely on massive register files and a lot of independent simultaneous work to achieve their performance. The Ada equivalent for a CPU would be SMT-48. CPUs are designed to minimize the latency of individual instructions and as such, the challenges for a CPU designer are much harder. Predicting control flow, recovering from mispredictions, bypass networks, multi-ported register files, instruction reordering, and sophisticated prefetchers for both instructions and data have no equivalent on the GPU side. Don't forget that they have to try and do all of this securely too even though that is a never-ending race.

 

[R0H1T]

Sandy Bridge is older than 10 years old by now. It released in early 2011, so you'd have to stretch it even further, not to mention you're also compounding all frequency increases and node improvements since that.

Again wrong, you forgot that Xeons usually released around a year later than mainstream laptop/desktop chips? At least till the 10nm botch job.

And how does that not relate to GPU's ~ which are even more node(improvements) dependent?

To calculate 10 years from now you'd have to take Haswell as a base and we already had 18 core Haswell chips.

You mean Xeons? It was released in Q3 2014

https://ark.intel.com/content/www/us/en/ark/products/codename/42174/products-formerly-haswell.html#@Server

You're seriously making a mess of timelines!

 

[Dr. Dro]

Again wrong, you forgot that Xeons usually released around a year later than mainstream laptop/desktop chips? At least till the 10nm botch job.

And how does that not relate to GPU's ~ which are even more node(improvements) dependent?


You mean Xeons? It was released in Q3 2014

https://ark.intel.com/content/www/us/en/ark/products/codename/42174/products-formerly-haswell.html#@Server

You're seriously making a mess of timelines!

By 2013, the 2nd gen (Sandy Bridge) CPUs were already starting to enter discontinuance and market phase-out stage

Intel® Core™ i7-2600K Processor (8M Cache, up to 3.80 GHz) Product Specifications

Intel® Core™ i7-2600K Processor (8M Cache, up to 3.80 GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

Haswell also debut in 2013

Intel® Core™ i7-4770K Processor (8M Cache, up to 3.90 GHz) Product Specifications

Intel® Core™ i7-4770K Processor (8M Cache, up to 3.90 GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

Both of them now exceed a decade of age, HEDT and high-end server always follows a year later or so but they are very much part of the same architectural family and follow a similar support path, as you can see, servicing ended only 6 months after the client processors

Intel® Xeon® Processor E5-2699 v3 (45M Cache, 2.30 GHz) Product Specifications

Intel® Xeon® Processor E5-2699 v3 (45M Cache, 2.30 GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

I've got a 4669 v3 myself, 12115 on R23, 550 single

Compared to my i9-13900KS at 41430/2368

So that's the absolute highest end CPU that were available back then for the scalable server segment x the same for consumer client today, in 10 years comparing 18x Haswell cores at 2.4 GHz x 8x Raptor Cove + 16x Gracemont config, we have, at least on Cinebench: 3.42x the multicore clout and 4.30x the single-core performance, accounting for going from quad-channel DDR4-2133 to dual DDR5-6400

We'd be arguing semantics, I brought this as a somewhat reasonable/realistic benchmark, but the truth is that it will change wildly with the SIMD optimizations available to each processor and is extremely workload-dependant.

I find it reasonable to believe GPU growth has outpaced CPU growth, but idk, YMMV.

 

[AnotherReader]

By 2013, the 2nd gen (Sandy Bridge) CPUs were already starting to enter discontinuance and market phase-out stage

View attachment 316081

Intel® Core™ i7-2600K Processor (8M Cache, up to 3.80 GHz) Product Specifications

Intel® Core™ i7-2600K Processor (8M Cache, up to 3.80 GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

Haswell also debut in 2013

View attachment 316082

Intel® Core™ i7-4770K Processor (8M Cache, up to 3.90 GHz) Product Specifications

Intel® Core™ i7-4770K Processor (8M Cache, up to 3.90 GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

Both of them now exceed a decade of age, HEDT and high-end server always follows a year later or so but they are very much part of the same architectural family and follow a similar support path, as you can see, servicing ended only 6 months after the client processors

View attachment 316083

Intel® Xeon® Processor E5-2699 v3 (45M Cache, 2.30 GHz) Product Specifications

Intel® Xeon® Processor E5-2699 v3 (45M Cache, 2.30 GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

I've got a 4669 v3 myself, 12115 on R23, 550 single

View attachment 316084

Compared to my i9-13900KS at 41430/2368

View attachment 316086

So that's the absolute highest end CPU that were available back then for the scalable server segment x the same for consumer client today, in 10 years comparing 18x Haswell cores at 2.4 GHz x 8x Raptor Cove + 16x Gracemont config, we have, at least on Cinebench: 3.42x the multicore clout and 4.30x the single-core performance, accounting for going from quad-channel DDR4-2133 to dual DDR5-6400

We'd be arguing semantics, I brought this as a somewhat reasonable/realistic benchmark, but the truth is that it will change wildly with the SIMD optimizations available to each processor and is extremely workload-dependant.

I find it reasonable to believe GPU growth has outpaced CPU growth, but idk, YMMV.

No, it only appears that way because of two factors:

• GPUs enjoyed more node jumps along the way. The 2013 era GPUs were using nodes similar to those used for Sandy Bridge.
• game engines aren't utilizing more than six cores in an optimal manner and many games are still dominated by single threaded performance which, for obvious reasons, hasn't increased as much as multi threaded performance

 

[TheoneandonlyMrK]

By 2013, the 2nd gen (Sandy Bridge) CPUs were already starting to enter discontinuance and market phase-out stage

View attachment 316081

Intel® Core™ i7-2600K Processor (8M Cache, up to 3.80 GHz) Product Specifications

Intel® Core™ i7-2600K Processor (8M Cache, up to 3.80 GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

Haswell also debut in 2013

View attachment 316082

Intel® Core™ i7-4770K Processor (8M Cache, up to 3.90 GHz) Product Specifications

Intel® Core™ i7-4770K Processor (8M Cache, up to 3.90 GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

Both of them now exceed a decade of age, HEDT and high-end server always follows a year later or so but they are very much part of the same architectural family and follow a similar support path, as you can see, servicing ended only 6 months after the client processors

View attachment 316083

Intel® Xeon® Processor E5-2699 v3 (45M Cache, 2.30 GHz) Product Specifications

Intel® Xeon® Processor E5-2699 v3 (45M Cache, 2.30 GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

I've got a 4669 v3 myself, 12115 on R23, 550 single

View attachment 316084

Compared to my i9-13900KS at 41430/2368

View attachment 316086

So that's the absolute highest end CPU that were available back then for the scalable server segment x the same for consumer client today, in 10 years comparing 18x Haswell cores at 2.4 GHz x 8x Raptor Cove + 16x Gracemont config, we have, at least on Cinebench: 3.42x the multicore clout and 4.30x the single-core performance, accounting for going from quad-channel DDR4-2133 to dual DDR5-6400

We'd be arguing semantics, I brought this as a somewhat reasonable/realistic benchmark, but the truth is that it will change wildly with the SIMD optimizations available to each processor and is extremely workload-dependant.

I find it reasonable to believe GPU growth has outpaced CPU growth, but idk, YMMV.

It might have, but that's not the same as a consistent uptick in performance of 1000X, over a reciprocal period of ANY defined amount.

So horse shit.

I've had 1000000% less sick days this month then last, And, at least I mentioned a reciprocal period, see.

Also no constraints mentioned at all.

Second.


They're on about AI and machine learning performance, or were.
Games have not scaled anywhere near that rate even if you're drinking all the dlss cool ade you can while saluting AMD.

So gaming improvements are irrelevant.

You also limit the time span of viable CPU in the list,why.

In comparison terms year dot to now on CPU verses the first AI hardware Nvidia gpu to now is apt.

Yes GPU out scaled CPU but then WiFi 7 chips will exponentially out pace WiFi six last year.

I hope tenstorrent shows Huang his shit ain't all that like ASICS did to Huang's mining interest or did you all think GPU mining died on GPU just because?!?.

 

[Vayra86]

Most games today are entirely CPU-limited when paired with the NVIDIA AD102 (in its cutdown form present in the RTX 4090), even when you are talking about processors such as the i9-13900KS and the Ryzen 9 7950X3D

Games aren't CPU limited, start looking at the 1% minimums for a new story. Nvidia has some strides to make there especially in newer engines. Consider the recent Cyberpunk result on Linux with a 31% perf gain due to CPU constraints the game never actually had... Starfield... great example. There are many more. Truly optimized games don't need a lot of CPU and they never have, especially if you compare say, what a game like Starcraft 2 did and what games do now while we gained lots of CPU perf since then. There are no hard cutoffs below the 4090 in terms of GPU / average FPS charts. Especially at 4K.

You're saying GPU demand has stagnated, but really, it hasn't, look at new engine performance and how Ada struggles, and RDNA3 isn't killing it either. Dozens of fresh games out of the gate this year struggle with performance and even post bug fix I still can't say I understand why shit runs as it does, and that's definitely not because of CPU constraints.
And then there is the RT tax, or any non proprietary version of it. If Huang wants that implemented, he's gonna have to get a lot more serious than this paltry 4090.

I don't take 'most games doing a thing wrong' for granted as 'this is the new normal'. I just know shit code when I see it. And every once in a while, there is a game/a dev that proves everyone wrong, and proves me right. Blame commerce and money for all the bad things, but don't let it cloud your view of what's what. A lot of limitations and problems are self-imposed.

The real question is how much you are willing to invest as a consumer in brute forcing past all those bills passed to you from the developer side. The reality is, once a baseline of perf is achieved, everything above it is bonus. The application runs. It doesn't crash. Market that shit. And here we are, playing a half finished product and throwing 1500 bucks of GPU at it to get the frames we want GPUs just travel along with the demand in gaming, simple as. There is no Huang's Law. There is Huang's 80+% market share and the resulting ability to dictate nonsense that feels like a law - while stocks last.

 

[trsttte]

The reality is, once a baseline of perf is achieved, everything above it is bonus. The application runs. It doesn't crash. Market that shit

And that baseline is usually a console with much lower performance compared to newer pc hardware. Then put a bit of duct tape here and there to compensate for windows fuckery running in the background and there's the game on PC

 

[Dr. Dro]

It might have, but that's not the same as a consistent uptick in performance of 1000X, over a reciprocal period of ANY defined amount.

So horse shit.

I've had 1000000% less sick days this month then last, And, at least I mentioned a reciprocal period, see.

Also no constraints mentioned at all.

Second.


They're on about AI and machine learning performance, or were.
Games have not scaled anywhere near that rate even if you're drinking all the dlss cool ade you can while saluting AMD.

So gaming improvements are irrelevant.

You also limit the time span of viable CPU in the list,why.

In comparison terms year dot to now on CPU verses the first AI hardware Nvidia gpu to now is apt.

Yes GPU out scaled CPU but then WiFi 7 chips will exponentially out pace WiFi six last year.

I hope tenstorrent shows Huang his shit ain't all that like ASICS did to Huang's mining interest or did you all think GPU mining died on GPU just because?!?.

Their claims are 1000 times the AI inference performance since Kepler, not 1000 times the overall performance. And this is more than likely the same for AMD as well, if their gains aren't even higher. GCN was good at compute and as a result so is CDNA.

GPU mining died because Ethereum made good on their proof of stake transition. The other cryptocurrencies are too small to be profitable, so a lot fewer people kept on mining. To compound that, the lockdowns ended and people returned to work, with that a lot of the speculative potential of crypto disappeared overnight. ASICs will not push GPU out of the business... And I'm unsure of the relation between Wi-Fi (radio comms) and raw compute performance?

Games aren't CPU limited, start looking at the 1% minimums for a new story. Nvidia has some strides to make there especially in newer engines. Consider the recent Cyberpunk result on Linux with a 31% perf gain due to CPU constraints the game never actually had... Starfield... great example. There are many more. Truly optimized games don't need a lot of CPU and they never have, especially if you compare say, what a game like Starcraft 2 did and what games do now while we gained lots of CPU perf since then. There are no hard cutoffs below the 4090 in terms of GPU / average FPS charts. Especially at 4K.

You're saying GPU demand has stagnated, but really, it hasn't, look at new engine performance and how Ada struggles, and RDNA3 isn't killing it either. Dozens of fresh games out of the gate this year struggle with performance and even post bug fix I still can't say I understand why shit runs as it does, and that's definitely not because of CPU constraints.
And then there is the RT tax, or any non proprietary version of it. If Huang wants that implemented, he's gonna have to get a lot more serious than this paltry 4090.

I don't take 'most games doing a thing wrong' for granted as 'this is the new normal'. I just know shit code when I see it. And every once in a while, there is a game/a dev that proves everyone wrong, and proves me right. Blame commerce and money for all the bad things, but don't let it cloud your view of what's what. A lot of limitations and problems are self-imposed.

The real question is how much you are willing to invest as a consumer in brute forcing past all those bills passed to you from the developer side. The reality is, once a baseline of perf is achieved, everything above it is bonus. The application runs. It doesn't crash. Market that shit. And here we are, playing a half finished product and throwing 1500 bucks of GPU at it to get the frames we want GPUs just travel along with the demand in gaming, simple as. There is no Huang's Law. There is Huang's 80+% market share and the resulting ability to dictate nonsense that feels like a law - while stocks last.

I did not say that GPU demand has stagnated, I said that current generation CPUs bottleneck the 4090 in all but a handful of very intensive scenarios involving super high resolutions and complex workloads. It's a different claim, one that you'll pretty easy to measure when we're talking high refresh scenarios.

 

[R0H1T]

We'd be arguing semantics, I brought this as a somewhat reasonable/realistic benchmark, but the truth is that it will change wildly with the SIMD optimizations available to each processor and is extremely workload-dependant.

I find it reasonable to believe GPU growth has outpaced CPU growth, but idk, YMMV.

It's actually great that you showed this, if we have (near)perfect scaling on 96c or 128c chips then top end EPYC should score 200k-300k on R23 multi. Or you could pick up some random benchmark which shows even higher growth like with say AI acceleration?

And that's the point of contention ~ GPU's IMO have not outpaced CPU growth in the last decade or so

And limiting the chips to just MSDT also wouldn't make sense because they're artificially limited, in terms of core count, by Intel/AMD due to the market they serve.

The scaling part is important because the vast majority of tasks on a GPU scales nicely for larger chips.

 

[Dr. Dro]

It's actually great that you showed this, if we have (near)perfect scaling on 96c or 128c chips then top end EPYC should score 200k-300k on R23 multi. Or you could pick up some random benchmark which shows even higher growth like with say AI acceleration?

And that's the point of contention ~ GPU's IMO have not outpaced CPU growth in the last decade or so

And limiting the chips to just MSDT also wouldn't make sense because they're artificially limited, in terms of core count, by Intel/AMD due to the market they serve.

The scaling part is important because the vast majority of tasks on a GPU scales nicely for larger chips.

Agreed. The market situation itself is also quite different now. Ryzen is a good pick for the vast majority of people today, and both companies offer healthy products on all segments right now. AMD is even ahead at the server market, which is quite the achievement.