NVIDIA Announces Grace CPU For ARM-Based AI/HPC Processor

[qarium]

Unless the transition from x86 to ARM on desktop/workstation is smooth, almost nobody currently depending on x86 apps will be willing to make an ARM CPU their main desktop/workstation CPU.
An example of a smooth transition: i386 to amd64.

right... but ARM can emulate X86_64... you or others maybe depending on X86 but maybe you don't depend on high-performance X86...

if you can manage that your high performance apps are in ARM ISA and your legacy x86 apps don't need performance then this is a win.

[Dawn]

So what you're saying is that if I have workloads that are single threaded because they cannot be designed to be multi-threaded, ARM is still a good while away from being competitive? Bring on the hybrid x86 + ARM systems then :P

ARM ST is actually pretty good these days (even outside of Apple, where it's excellent.) X1 is, for same-clock integer, about on par with Zen 3, though for efficiency reasons nobody really runs it over 3GHz or so yet.

[qarium]

So what you're saying is that if I have workloads that are single threaded because they cannot be designed to be multi-threaded, ARM is still a good while away from being competitive? Bring on the hybrid x86 + ARM systems then :P

right now ARM can not outperform X86 in singlecore benchmarks.

yea such hybride is not a bad idea i already told AMD they should build a chiplet hybride of RDNA and CDNA with a IO Crossbridge for HBM2 ram shared to both the RDNA chip and CDNA chip.

such hybrid designs make more sense than most people can imagine.

but i do not believe there will be a ARM+X86 hybrid... but a IBM openPower + ARM hybrid should be simple

but yes imagine a dual socket system in the first socket your X86 and in the second socket your ARM...

and then the software runs on the cpu who is best for.

[piotrj3]

right now ARM can not outperform X86 in singlecore benchmarks.

yea such hybride is not a bad idea i already told AMD they should build a chiplet hybride of RDNA and CDNA with a IO Crossbridge for HBM2 ram shared to both the RDNA chip and CDNA chip.

such hybrid designs make more sense than most people can imagine.

but i do not believe there will be a ARM+X86 hybrid... but a IBM openPower + ARM hybrid should be simple

but yes imagine a dual socket system in the first socket your X86 and in the second socket your ARM...

and then the software runs on the cpu who is best for.

Our computers are already hybrids. SSDs have own processor to manage data. GPU is own kind of processor with own principles. Samsung I think was supposed to make a prototype RAM with some sort of logic as well implemented in it for simpler computing. Heck even AMD and Intel use own "inner" processor for security workloads, and it is ARM architecture.

Your computers are already hybrids.

[TemplarGR]

Yet I sit here running an M1 based Mac Book Air that is as good as my high end desktop and runs all the software I need to run on a laptop. The future is already here, it just requires people to swallow the right pill and wake up from their coma.

No it is not as good as your high end desktop. Can you play Cyberpunk 2077 on it? Nope. Can you run Linux on it?

"it runs all the software i need to run on a laptop" means nothing. My android smartphone i bought for 115 euros a few months ago does everything i want to run on a smartphone, your point? A desktop needs much more than that. Not only in terms of performance but also compatibility.

[TemplarGR]

ARM is not on Desktop or Workstation YET....

ARM right now is only in mobile market and server market.

but on server market a 80-128core ARM cpu is at 4000€ and has similar performance to a 40core 10nm 8000€ Xeon.

by this alone you can see ARM is the future... a 16core AMD 2950X is at ~700€ yes you need 32cores for ARM to be able to compete but it could only cost you ~400€

the X86_64 Tax is that high.

LOL. What an ignorant statement that only a non-professional would make....

Servers are nothing like the desktop market. Single core performance is paramount here. It is the reason Bulldozer lost to Core despite double the cores (minus fpu which was shared per 2 cores). The vast majority of software (games included) on the desktop are single threaded or dual threaded at best. AT BEST. Even today. Perhaps they can utilize more cores by lowering the load for individual cores but still don't really exploit them in any serious capacity. ARM is dead in the water if it offers significantly less per thread performance, period.

And no, the x86 tax is miniscule. And people talk about x86 emulation on ARM like it won't have a performance penalty....

[TemplarGR]

ARM ST is actually pretty good these days (even outside of Apple, where it's excellent.) X1 is, for same-clock integer, about on par with Zen 3, though for efficiency reasons nobody really runs it over 3GHz or so yet.

Clock for clock performance is not telling the full story. And not running them higher may not be just about efficiency, but also simple stability. you can get great IPC in a processor but be unable to clock it high because of the design.

For a clear example to illustrate, Pentium 4 (netburst) had lower IPC than AMD's Athlon (and Pentium 3) but could be clocked significantly higher because of its longer pipeline. Athlon had a shorter pipeline, so it was clocked lower. It wasn't for just "efficiency" reasons (after all, both Intel and AMD -FX9xxxx series- have proven they don't care about efficiency much), but also for stability reasons. At some point, you just get errors and instability if you clock a design for more than it can go.

ARM is similar in the sense that it is designed for lower clocks and lower die sizes. The designs are not meant to be clock champions, and i am not even sure they could even reach stable clocks that high. And it lacks the SIMDs/FP performance too.

[zxy_thf]

ARM doesn't give anything to you, except avoiding duopoly of AMD-Intel.

Actually this advantage is also not that clear, if we take (potential) vendor locked-in into consideration.
We may switch Xeon with Epyc and enjoy improved performance/dollar, but when we switched one ARM from another one, there is no guarantee that they share the shame extensions and have similar performance behavior.

Considered Apple's success from walled garden, I don't believe ARM's vendors won't want to lock you into their own production line.

[Jumbotron]

Ugh enough with this cringy tag line...

OK...I admit I was wrong. Let me amend.

The Age of ARM HAS BEEN HERE.

For PDAs, then Smart Phones and Tablets.
For IoT, Smart Devices, Watches, Sensors, Nerworking, etc.

The Age of ARM is ARRIVING for EVERY OTHER platform not mentioned above, like Supercomputers, HPC, AI, Edge, VR, AR, Auto both augmented and fully Self Driving. Also in Chromebooks, ALL APPLE products, increasingly in Microsoft products.

Because the x86 Desktop PC is an increasingly marginalized platform in light of ARM based Smart Phones, Tablets and soon every SINGLE Apple Personal Compute product including Desktops, by 2030 more than 50% of all Personal Compute products both desktop and laptops Windows, Apple and Google combined will be ARM based.

[qarium]

Actually this advantage is also not that clear, if we take (potential) vendor locked-in into consideration.
We may switch Xeon with Epyc and enjoy improved performance/dollar, but when we switched one ARM from another one, there is no guarantee that they share the shame extensions and have similar performance behavior.
Considered Apple's success from walled garden, I don't believe ARM's vendors won't want to lock you into their own production line.

thats not possible because vendor lock in only works if you have high marketshare.

ARM any kind of does not have this marketshare right now.