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
Add your own comment

172 Comments on NVIDIA's Next-Generation "Ampere" GPUs Could Have 18 TeraFLOPs of Compute Performance

#76
cucker tarlson
ratirt
AMD is aware and I can assure you they have a plan for all of this.
:rolleyes:
to refresh rdna1 in 2020 ?
Posted on Reply
#77
ratirt
cucker tarlson
:rolleyes:
to refresh rdna1 in 2020 ?
Could be. Just like NV wants to refresh Turing with Ampere :)

kings
But AMD cannot wait, they are the ones who have to gain market share, increase profits, try to change the mindshare, etc ... It is AMD that is playing catch up, so if they have big guns, there is no interest in waiting.

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.
BTW is NV that is supposed to wait or AMD? What "wait" here means? If I were AMD I would wait for NV's release. To see where I stand with their new GPU and the odds are it will be very good. Wait and make adjustments to whatever they've got and if it turns out AMD can't compete, adjust the graphics tiers and lower price by making cuts to at least stay competitive at the price/performance ratio. Needless to say, most people think that AMD is the black sheep in the GPU market so AMD needs to bite on the bullet and bite pretty damn good.
Posted on Reply
#78
cucker tarlson
ratirt
Could be. Just like NV wants to refresh Turing with Ampere :)
source ?
I don't know how your getting all this information cause so far amd mentioned rdna2 is rdna1 with rt and vrs but you claims it'srevolutionary for some reason,while nvidia said nothing about ampere and yet you know it's disappointing.
Posted on Reply
#79
ratirt
cucker tarlson
source ?
I don't know how your getting all this information cause so far amd mentioned rdna2 is rdna1 with rt and vrs but you claims it'srevolutionary for some reason,while nvidia said nothing about ampere and yet you know it's disappointing.
Source?
Since we are at the source. Any source what Ampere actually is?
BTW it is still sarcasm.
Posted on Reply
#80
cucker tarlson
ratirt
If I were AMD I would wait for NV's release. To see where I stand with their new GPU and the odds are it will be very good.
they had nothiong for tu104 and up,so yeah,keep waiting to see if ampere is slower than turing maybe.

ratirt
Source?
where did you get that ampere is turing refresh ?

at this point you're pretty much trolling.
Posted on Reply
#81
ratirt
cucker tarlson
they had nothiong for tu104 and up,so yeah,keep waiting to see if ampere is slower than turing maybe.
Who said it is going to be slower? I said it will be very good card. AMD is waiting to see what its up against. That is how I take it and that is why Navi RDNA has never been released.
Posted on Reply
#82
cucker tarlson
ratirt
that is why Navi RDNA has never been released.
what ?

dude,stop trolling every gpu thread with false information.
Posted on Reply
#83
efikkan
kings
We will see, the last time AMD said it was going to compete seriously in the high end, we had Vega, 15 months after Pascal, barely matching the GTX 1080.

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.
Yeah, this seems to be a recurring thing with AMD lately. The new one is disappointing; then rumors of a bigger version immediately surfaces. This has been the case with Polaris, Vega and Navi.

ratirt
Well, NV is releasing its new GPU first. At this point anything is possible.
Well, nothing seems to indicate Nvidia's next gen is imminent. Consumer products are probably shipping in the second half of the year, but that's still a guess.

ratirt
Even though It has been confirmed the new RDNA2 is coming out this year, we all know this may change. I think AMD is waiting for the new NV GPU (that's a maybe). If they find out the new NV is not that improved over the Turing, maybe they will do something else? It is hard to predict what each company will actually do.
They certainly can postpone or cancel products, or overclock them like they recently did. But the technical details of Navi 2x were already set in stone when it was taped out, probably early last year. People generally overestimate these companies' ability to adapt on the fly. They can certainly adjust clock speeds (within a range), disable/enable some features and change pricing on a moment's notice, but pretty much everything else is decided 2-3 years ahead. Even changes in production of different wafers usually takes >4 months until it reaches the market.

ratirt
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.
Of course, this should be the case for everyone. :)

At this point, no one (even at Nvidia) knows exactly how these new cards will perform.
Posted on Reply
#84
Juankato1987
We are always worried about NV pricing... lol... hopelly a RTX 3050 might be the game changer RTX at 200$ or less????
and 1660's performance at 150$ may be????
who knows.....
Posted on Reply
#85
ratirt
Juankato1987
We are always worried about NV pricing... lol... hopelly a RTX 3050 might be the game changer RTX at 200$ or less????
and 1660's performance at 150$ may be????
who knows.....
do you really believe that ? I really hope so it will happen.
efikkan
They certainly can postpone or cancel products, or overclock them like they recently did. But the technical details of Navi 2x were already set in stone when it was taped out, probably early last year. People generally overestimate these companies' ability to adapt on the fly. They can certainly adjust clock speeds (within a range), disable/enable some features and change pricing on a moment's notice, but pretty much everything else is decided 2-3 years ahead. Even changes in production of different wafers usually takes >4 months until it reaches the market.
Read it as it goes with the quotes.

With all do respect. I don't underestimate anything. The specs for RDNA2 are not there. Just as NV's aren't. I'm trying to be objective but it does not bode well with some members here.

Question. What are those specs? NAVI 2x? what is that mean? BTW I can say same thing about NV. Just as our friend Tarlson said Ampere it is something totally different from Turing in terms of arch (not RDNA2 becuse that is just a refresh) Well, The AMPERE, I see RT cores, Cuda, Tensore. Is this something to justify new? It looks same to me as Turing in every way. I'm not here to mark what you say but please use arguments. Comparison of the two at this point is crazy. Companies can do whatever they want. It is a market and they will do anything to sell stuff.

Agreed. Anything can happen.
Posted on Reply
#86
kapone32
All of those suppositions that AMD CANNOT catch Nvidia remind me of all the flak on the internet before Ryzen launched. I read a lot of comments like "Ryzen will never catch Intel". Today I don't think anyone would state Intel's CPUs are summarily better than AMD's offerings (Unless you are a fanboy) . AMD has always been a company with great ideas but bad leadership (Just read the history of AMD (ARM & GLO FO).

Today's AMD GPU division is not the company that refreshed Tahiti for 5 or 6 generations. The evidence of this was Polaris that was just as fast as Tahiti with 1/2 the power draw (That was the biggest caveat to AMD cards). Vega was good but released during the mining craze that made it's price go through the roof this made it less of a proposition vs Nvidia.

The 5700XT is currently AMD's fastest card. Not that we can but if we look at Ryzen it started at 8 cores and 16 cores max with TR4 in 2017. It is 2020 and the core count has increased to 16 and 64 cores max for HEDT. That is in 3 years, much more than Intel has done in 10. I have confidence in Lisa Su and have not heard her say anything that has not come to pass. I am not saying it is definite but I do believe the propensity is high, besides we need Nvidia to bring prices back to where they were before the mining craze.
Posted on Reply
#87
cucker tarlson
kapone32
All of those suppositions that AMD CANNOT catch Nvidia remind me of all the flak on the internet before Ryzen launched. I read a lot of comments like "Ryzen will never catch Intel". Today I don't think anyone would state Intel's CPUs are summarily better than AMD's offerings (Unless you are a fanboy) . AMD has always been a company with great ideas but bad leadership (Just read the history of AMD (ARM & GLO FO).

Today's AMD GPU division is not the company that refreshed Tahiti for 5 or 6 generations. The evidence of this was Polaris that was just as fast as Tahiti with 1/2 the power draw (That was the biggest caveat to AMD cards). Vega was good but released during the mining craze that made it's price go through the roof this made it less of a proposition vs Nvidia.

The 5700XT is currently AMD's fastest card. Not that we can but if we look at Ryzen it started at 8 cores and 16 cores max with TR4 in 2017. It is 2020 and the core count has increased to 16 and 64 cores max for HEDT. That is in 3 years, much more than Intel has done in 10. I have confidence in Lisa Su and have not heard her say anything that has not come to pass. I am not saying it is definite but I do believe the propensity is high, besides we need Nvidia to bring prices back to where they were before the mining craze.
except nvidia is not intel
Posted on Reply
#88
kapone32
cucker tarlson
except nvidia is not intel
I am well aware of that but is Nvidia looking over their shoulder? WIth the release of the 2060KO it would seem so.
Posted on Reply
#89
cucker tarlson
kapone32
I am well aware of that but is Nvidia looking over their shoulder? WIth the release of the 2060KO it would seem so.
well they didn't really do anything for 2080ti since launch,and it's more ridicoulously priced than any gpu I ever remembered
Posted on Reply
#90
kapone32
cucker tarlson
well they didn't really do anything for 2080ti since launch,and it's more ridicoulously priced than any gpu I ever remembered
[/QUOTE

I believe AMD is doing the right thing by trying to get the midrange as those GPUs sell way more than high end. As far as i am concerned the 2080Ti, 2080S Super and retail 1080TIs are way too expensive. This is why I hope that RDNA2 or Big Navi do indeed compete at lower prices. I want to replace my Vega 64(s) but not at the cost of an entire decent build.
Posted on Reply
#91
gamefoo21
renz496
OpenCL exist for a decade now. Back in 2009 they said give it two years. Two years and no one ever remember CUDA ever exist. Part of the issue is CUDA is fully controlled by nvidia. That in one aspect will speed up CUDA development and optimization on nvidia hardware. With OpenCL everyone needs to agree what spec to be accepted into the API and this where the major problem are. Because some feature might work nicely on one hardware but not others. And they will not want this to happen because that will only give certain IHV the comepetitive edge while leave out the others. So while they are disputing how the feature should be added to the spec closed API like CUDA already moving forward to implement next feature and improvement. For professional client they will favor the one that really works for them even if they really like open source. Because in pro world time is money. And second if only certain vendor going to be good with that open source API what the point? Because to get maximize percormance they will still going to be locked on that certian hardware when the point of using open source solution is not to locked themselves on one vendor only.
Nvidia definitely doesn't help the situation. They purposefully cripple OpenCL on their GPUs.

They refuse to support newer versions of it. Radeons have supported OpenCL 2.0 since the 7790.

It's like nV making an incompatible Variable Refresh tech, to the open standard in DisplayPort. Intel adopted the open spec, the HDMI group came on board, and adopted it. Suddenly nV made gSync work with Adaptive Vsync no NV port corruption necessary.

NV will block and slow down OpenCL as much as they can. I suspect you'll see their tune change if the Intel and AMD really start challenging NV.

I still hate that NV does everything they can to make PhysX all but impossible if your main GPU isn't NV.

cucker tarlson
well they didn't really do anything for 2080ti since launch,and it's more ridicoulously priced than any gpu I ever remembered
Titan V?
Posted on Reply
#93
efikkan
ratirt
With all do respect. I don't underestimate anything. The specs for RDNA2 are not there. Just as NV's aren't. I'm trying to be objective but it does not bode well with some members here.
I don't pretend to know the specs of either if that's what you're implying.

I was just trying to point out that AMD's ability to adapt their lineup to what Nvidia launches is very limited, and vice versa of course. Beyond price and clock speeds, there is little they can adjust over night. It's not like they can whip up a new chip design in a couple of months.

ratirt
Question. What are those specs? NAVI 2x? what is that mean?
We know of three Navi 2x chips so far; Navi 21, Navi 22 and Navi 23 from Linux driver patches. If there are more than these three chips coming, then will not be launching anytime soon.

ratirt
BTW I can say same thing about NV. Just as our friend Tarlson said Ampere it is something totally different from Turing in terms of arch (not RDNA2 becuse that is just a refresh) Well, The AMPERE, I see RT cores, Cuda, Tensore. Is this something to justify new? It looks same to me as Turing in every way.
I don't know exactly where you see these details, to my knowledge Nvidia haven't released anything public about "Ampere" so far.

But regardless, they've been using "Cuda cores" for many generations (since Tesla?), but their performance characteristics have changed radically over time. Just from Pascal to Turing there were major efficiency gains.

I don't know if Nvidia's next gen will be a small tweak of Turing or if it's another major architecture, but deriving this from vague rumors is utterly pointless. I've seen nothing substantive about Nvidia's next gen so far, and a pro-tip; whenever you see rumors pointing in every direction, it's usually an indication that "all of them" are just BS, it's a typical symptom of the information vacuum we're in right now. This is why I have recommended to be cautious about most "Ampere" or Navi 2x rumors so far. When we are getting closer to a real launch, we will start to see believable rumors pointing in the same direction.

As for Nvidia's next gen, you have to acknowledge that even if it's just a shrunk and tweaked Turing, it's going to cause some real challenge for AMD to compete with. Nvidia currently have a huge lead thanks to their architectural superiority, even with AMD having a node advantage. With the next gen that advantage will be gone for AMD.
Posted on Reply
#94
gamefoo21
cucker tarlson
titan rtx ?
Hmm... Good point...
Posted on Reply
#95
medi01
cucker tarlson
amd proved that even when they have they tools like new,more efficient architecture and a 7nm tsmc node,they still can't deliver in the high end.
Do you actually believe what you type?
Posted on Reply
#96
T4C Fantasy
CPU & GPU DB Maintainer
kapone32
I am well aware of that but is Nvidia looking over their shoulder? WIth the release of the 2060KO it would seem so.
The KO was not a response from Nvidia.. it's a reused chip they do every year with the xx60 series.. the only reason why we know is because the card was disassembled. TU104 2060 is sold by all manufacturers randomly with the same skus.
Posted on Reply
#97
cucker tarlson
medi01
Do you actually believe what you type?
lol,look who's talking :laugh:

do I believe ? yes,strongly.I mean you have to be delusional to think otherwise when amd's last high end competitor was fury x.
with every release their goalposts move,after fury x their high end was vega 64 that went against gtx 1080,now they're gladly selling rtx 2070 (106 die) competitor at +$400 with a limited feature set.
Why cant the red fanbase accept the obvious-amd have introduced the NVIDIA model.they are gladly selling 250mm dies at $400 and making more money than before so they dont event bother with chasing nvidia in the high end when they're milking the mid range
Posted on Reply
#98
ratirt
efikkan
We know of three Navi 2x chips so far; Navi 21, Navi 22 and Navi 23 from Linux driver patches. If there are more than these three chips coming, then will not be launching anytime soon.
Oh. Now I get it. I read it as 2x- two times and that got me confused.
efikkan
I don't know exactly where you see these details, to my knowledge Nvidia haven't released anything public about "Ampere" so far.

But regardless, they've been using "Cuda cores" for many generations (since Tesla?), but their performance characteristics have changed radically over time. Just from Pascal to Turing there were major efficiency gains.

I don't know if Nvidia's next gen will be a small tweak of Turing or if it's another major architecture, but deriving this from vague rumors is utterly pointless. I've seen nothing substantive about Nvidia's next gen so far, and a pro-tip; whenever you see rumors pointing in every direction, it's usually an indication that "all of them" are just BS, it's a typical symptom of the information vacuum we're in right now. This is why I have recommended to be cautious about most "Ampere" or Navi 2x rumors so far. When we are getting closer to a real launch, we will start to see believable rumors pointing in the same direction.

As for Nvidia's next gen, you have to acknowledge that even if it's just a shrunk and tweaked Turing, it's going to cause some real challenge for AMD to compete with. Nvidia currently have a huge lead thanks to their architectural superiority, even with AMD having a node advantage. With the next gen that advantage will be gone for AMD.
You always have to be cautious with the information. My argument was we don't know anything about either of the new cards.
I do acknowledge this and I have mentioned this in the previous posts. I am sure that the new NV release will be fast and great. Not sure about the price though. Actually I'm worried about the price but that's just me.
Posted on Reply
#99
medi01
cucker tarlson
lol,look who's talking :laugh:
Yeah, oh, wait, what? Jesus...

cucker tarlson
do I believe ? yes,strongly.I mean you have to be delusional to think otherwise when amd's last high end competitor was fury x.
So you honestly believe that AMD, which was able to beat almighty Intel even at IPC while flooding market with cores, is not capable of producing 7nm GPU chips that are 500+ mm2 in size?
Like, freaking, for real?



GPU gap was never even remotely as big and bad as the gap we had in CPU market. Only in Raja's time, with modest R&D obviously focusing on Zen, AMD went "missing in action". But it's getting back, with entire set from 5700 to 5500 series looking good. Bigger chips are inevitably coming.
Posted on Reply
#100
cucker tarlson
medi01
Yeah, oh, wait, what? Jesus...


So you honestly believe that AMD, which was able to beat almighty Intel even at IPC while flooding market with cores, is not capable of producing 7nm GPU chips that are 500+ mm2 in size?
Like, freaking, for real?



GPU gap was never even remotely as big and bad as the gap we had in CPU market. Only in Raja's time, with modest R&D obviously focusing on Zen, AMD went "missing in action". But it's getting back, with entire set from 5700 to 5500 series looking good. Bigger chips are inevitably coming.
This is a GPU thread,Intel and epyc have nothing to do with it.
Amd has the design to beat intel in every market,if they can do the same with navi I will be happy to see it rather than hear their marketing talk about disrupting 4k gaming.
They're on 7 nm tsmc and new uarch,why dont they put up? Cause of what I wrote before.They would lose to nvidias +$500 cards cause of lack of features,so they're happy to milk mid range.
Let us see those chips,I am not buying a word their CEO or their fanbase say about high end Radeon.
Posted on Reply
Add your own comment