Monday, February 3rd 2020
NVIDIA's Next-Generation "Ampere" GPUs Could Have 18 TeraFLOPs of Compute Performance
NVIDIA will soon launch its next-generation lineup of graphics cards based on a new and improved "Ampere" architecture. With the first Tesla server cards that are a part of the Ampere lineup going inside Indiana University Big Red 200 supercomputer, we now have some potential specifications and information about its compute performance. Thanks to the Twitter user dylan552p(@dylan522p), who did some math about the potential compute performance of the Ampere GPUs based on NextPlatform's report, we discovered that Ampere is potentially going to feature up to 18 TeraFLOPs of FP64 compute performance.
With Big Red 200 supercomputer being based on Cray's Shasta supercomputer building block, it is being deployed in two phases. The first phase is the deployment of 672 dual-socket nodes powered by AMD's EPYC 7742 "Rome" processors. These CPUs provide 3.15 PetaFLOPs of combined FP64 performance. With a total of 8 PetaFLOPs planned to be achieved by the Big Red 200, that leaves just a bit under 5 PetaFLOPs to be had using GPU+CPU enabled system. Considering the configuration of a node that contains one next-generation AMD "Milan" 64 core CPU, and four of NVIDIA's "Ampere" GPUs alongside it. If we take for a fact that Milan boosts FP64 performance by 25% compared to Rome, then the math shows that the 256 GPUs that will be delivered in the second phase of Big Red 200 deployment will feature up to 18 TeraFLOPs of FP64 compute performance. Even if "Milan" doubles the FP64 compute power of "Rome", there will be around 17.6 TeraFLOPs of FP64 performance for the GPU.
Sources:
@dylan522p(Twitter), The Next Platform
With Big Red 200 supercomputer being based on Cray's Shasta supercomputer building block, it is being deployed in two phases. The first phase is the deployment of 672 dual-socket nodes powered by AMD's EPYC 7742 "Rome" processors. These CPUs provide 3.15 PetaFLOPs of combined FP64 performance. With a total of 8 PetaFLOPs planned to be achieved by the Big Red 200, that leaves just a bit under 5 PetaFLOPs to be had using GPU+CPU enabled system. Considering the configuration of a node that contains one next-generation AMD "Milan" 64 core CPU, and four of NVIDIA's "Ampere" GPUs alongside it. If we take for a fact that Milan boosts FP64 performance by 25% compared to Rome, then the math shows that the 256 GPUs that will be delivered in the second phase of Big Red 200 deployment will feature up to 18 TeraFLOPs of FP64 compute performance. Even if "Milan" doubles the FP64 compute power of "Rome", there will be around 17.6 TeraFLOPs of FP64 performance for the GPU.
172 Comments on NVIDIA's Next-Generation "Ampere" GPUs Could Have 18 TeraFLOPs of Compute Performance
please read more carefully
This is not good at all.
I'm hoping for rdna2 rt card in h2 but I wouldn't put it past amd to do some sort of stupid 580->590-like refresh with 5700xt+/5700xtx and move the real rdna 2 cards to 2021.
Not to mention the "overclockers dream" or "poor Volta" thing, etc ... What CEOs say is not reliable and especially at AMD we have had recent cases of reality not matching what is said. Therefore, caution and control over expectations are necessary.
all this amd ceo "we'll be big next year" talk that people keep telling me to believe makes me nauseous.
I'll believe it when I see big rdna2 cards this year,if it's gonna be rdna1 refresh to go against two year old turing cards it'll be a miserable failure yet again.
Even if NV and AMD release their cards I will still wait for benchmarks. I'm not in a hurry and I need to see what each can do to make a purchase decision.
Also, I think there was always this correlation between consoles and graphics. How I always pictured it, the new stuff was always introduced as a GPU for PC and then it would find it's way to consoles. That is reasonable since consoles as dedicated gaming platforms, don't need that much processing power as PCs do and (despite PS4pro), all graphics options are locked.
Why do you feel sorry? Most of the people purchasing it, know what they've purchased and for what reason. AMD was not aiming for sorry though and what I hear, people are happy with the performance for what they had to pay for it. Performance per $ and value is good in the tier where the card is currently. Also mixing up consoles and PC graphics in the way you do is not right.
I can say, I'm sorry for people who purchased Turing in 2020 since new NV GPU is going to be released soon. I don't think that would have been justified since they are happy with what they've got.
Try this for a change.
This Ampere is nothing more than Turing on a 7nm node. Nothing special about it and I wouldn't expect much. Would you agree with my statement? that was a sarcasm but sure, I give you that one :)
The longer they wait, the more time they give Nvidia to respond and even subject themselves to Nvidia already having something better by then. It's a dangerous game for them, I don't think AMD is doing this on purpose. I agree, for me It´s clear by now that 1st gen Navi will not go beyond this, otherwise they had already launched them. And it´s not because they are waiting, it´s because it is probably not feasible to scale up to the level they need. AMD did not mind launching a Radeon VII and discontinuing it 4 months later, for that reason, I don´t believe that AMD is waiting for something new from Nvidia, they simply have nothing to offer.
Maybe with RDNA 2 things will be much different, we all hope so, but that is also unknown. People tend to think that RDNA 2 is going to be a revolution, but there is no evidence of that, only rumors. It may just be RDNA 1 with minor tweaks and RT/VRS, we don´t know. The same applies to Nvidia's rumors, we don't know if it's just going to be a slight tweak on Turing or something quite different.
you commented like a thousand times in this thread no,cause turing on 7nm would be amazing
pascal was maxwell on a shrunk die and it put amd so far behind that they still can't catch up with 1080Ti in 2020 unless you're counting radeon 7 one person thinks it and keeps spamming every thread with it
fixed that for you.
What the OP said about PS5in the thread says Based on RDNA2 which can mean anything. Either way I haven't seen it confirmed. I see. that explains everything.