Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Nvidia Ampere GA102 GPU Architecture [pdf]

nvidia.com

Post reply on HN
Print view View original

11–20 of 51 posts

Re: Nvidia Ampere GA102 GPU Architecture [pdf]

#11

Post by zamadatix »

dragontamer wrote: post #5

I posted this a day or two ago: The A100 whitepaper "spoiled" a lot of these factoids already. ( https://www.nvidia.com/content/dam/en-zz/Solutions/Data-Cent ...) The new bit seems to be the doubling of FP32 "CUDA cores" (I really hate that word: when Intel or AMD double their CPU pipelines it doesn't mean that they're selling more cores, it means their cores got wider... anyway). A100 didn't have this feature (I ass…

GPU accelerated direct storage access was previously part of Telsa/DGX-2 as a feature named "GPUDirect storage" https://developer.nvidia.com/blog/gpudirect-storage/.

It's a feature the new consoles are doing so it'll be widely supported.

Re: Nvidia Ampere GA102 GPU Architecture [pdf]

#12

Post by Guthur »

modeless wrote: post #6
amelius wrote: post #2

These GPU architectures are too complicated for my taste. It reminds me of the days of segmented memory, and makes me feel like having to jump through hoops all the time. I'm curious if people have tried to invent something more elegant, and where these approaches have failed. Also, the fact that NVidia has probably patented this particular architecture makes it less interesting for me to really dig into.

It's a good question actually. Intel tried to make a GPU called Larrabee that was mostly a bunch of small x86 cores with giant vector units. Turns out that it couldn't compete in rendering performance on existing games (in 2010) without the fixed function units that GPUs have, so they canceled it as a GPU. It did result in the AVX-512 instruction set though. I think the idea still has promise but there's a chicken an…

Modern x86 processors are far from simple though and so it's arguable that it's not significantly less complicated.

Though to be fair I'm not sure it's really all that complex relatively to modern high end processors. Most of gpu is just the same unit repeated.

For mind boggling complexity in my mind is the manufacturing process undertaken by the likes of TSMC.

Re: Nvidia Ampere GA102 GPU Architecture [pdf]

#13

Post by jagger27 »

dragontamer wrote: post #5

I posted this a day or two ago: The A100 whitepaper "spoiled" a lot of these factoids already. ( https://www.nvidia.com/content/dam/en-zz/Solutions/Data-Cent ...) The new bit seems to be the doubling of FP32 "CUDA cores" (I really hate that word: when Intel or AMD double their CPU pipelines it doesn't mean that they're selling more cores, it means their cores got wider... anyway). A100 didn't have this feature (I ass…

> The FP32 doubling, is one of the most important bits here. But fortunately for programmers, this doesn't really change how you do your code.

Early benchmarks are showing games under-performing quite a bit in the worst cases. The crux of the issue is that it's not /exactly/ a no-compromise doubling of FP32. Each data path per SM can either do 2xFP32 or 1xINT32/1xFP32 per clock cycle. So if your game or application has any significant INT32 operations scheduled, all of a sudden you're back to the number of FP32 cores you had last generation, though you get the benefit of parallel INT32 execution.

It's a pretty cool architecture overall though.

Re: Nvidia Ampere GA102 GPU Architecture [pdf]

#14

Post by CyberDildonics »

01100011 wrote: post #3
amelius wrote: post #2

These GPU architectures are too complicated for my taste. It reminds me of the days of segmented memory, and makes me feel like having to jump through hoops all the time. I'm curious if people have tried to invent something more elegant, and where these approaches have failed. Also, the fact that NVidia has probably patented this particular architecture makes it less interesting for me to really dig into.

When you run up against the limits of Moore's law and the end of Dennard Scaling you have to get dirty and do what it takes to get more performance. Personally, I think CPU architecture became too complicated for my taste after the 68k. So what?

You're posting this from a 68k?

Re: Nvidia Ampere GA102 GPU Architecture [pdf]

#15

Post by CyberDildonics »

modeless wrote: post #6
amelius wrote: post #2

These GPU architectures are too complicated for my taste. It reminds me of the days of segmented memory, and makes me feel like having to jump through hoops all the time. I'm curious if people have tried to invent something more elegant, and where these approaches have failed. Also, the fact that NVidia has probably patented this particular architecture makes it less interesting for me to really dig into.

It's a good question actually. Intel tried to make a GPU called Larrabee that was mostly a bunch of small x86 cores with giant vector units. Turns out that it couldn't compete in rendering performance on existing games (in 2010) without the fixed function units that GPUs have, so they canceled it as a GPU. It did result in the AVX-512 instruction set though. I think the idea still has promise but there's a chicken an…

They did actually make it and it was not supposed to be a GPU..

Re: Nvidia Ampere GA102 GPU Architecture [pdf]

#16

Post by apatheticonion »

01100011 wrote: post #3
amelius wrote: post #2

These GPU architectures are too complicated for my taste. It reminds me of the days of segmented memory, and makes me feel like having to jump through hoops all the time. I'm curious if people have tried to invent something more elegant, and where these approaches have failed. Also, the fact that NVidia has probably patented this particular architecture makes it less interesting for me to really dig into.

When you run up against the limits of Moore's law and the end of Dennard Scaling you have to get dirty and do what it takes to get more performance. Personally, I think CPU architecture became too complicated for my taste after the 68k. So what?

As someone with the most basic understanding of CPUs and assembly, why do you say that?

Re: Nvidia Ampere GA102 GPU Architecture [pdf]

#17

Post by PaulKeeble »

jagger27 wrote: post #13
dragontamer wrote: post #5

I posted this a day or two ago: The A100 whitepaper "spoiled" a lot of these factoids already. ( https://www.nvidia.com/content/dam/en-zz/Solutions/Data-Cent ...) The new bit seems to be the doubling of FP32 "CUDA cores" (I really hate that word: when Intel or AMD double their CPU pipelines it doesn't mean that they're selling more cores, it means their cores got wider... anyway). A100 didn't have this feature (I ass…

> The FP32 doubling, is one of the most important bits here. But fortunately for programmers, this doesn't really change how you do your code. Early benchmarks are showing games under-performing quite a bit in the worst cases. The crux of the issue is that it's not /exactly/ a no-compromise doubling of FP32. Each data path per SM can either do 2xFP32 or 1xINT32/1xFP32 per clock cycle. So if your game or application h…

Its not uncommon for GPU workloads in games to max out about 20% INT32 calculations, but alas its enough to drop the FP32 performance quite a bit. I suspect Nvidia next time will probably separate out the INT32 and 2x FP32 units and gradually move towards going towards a better ratio of hardware that better suits the usual workload split.

Re: Nvidia Ampere GA102 GPU Architecture [pdf]

#18

Post by kevingadd »

dragontamer wrote: post #5

I posted this a day or two ago: The A100 whitepaper "spoiled" a lot of these factoids already. ( https://www.nvidia.com/content/dam/en-zz/Solutions/Data-Cent ...) The new bit seems to be the doubling of FP32 "CUDA cores" (I really hate that word: when Intel or AMD double their CPU pipelines it doesn't mean that they're selling more cores, it means their cores got wider... anyway). A100 didn't have this feature (I ass…

RTX I/O is a parallel to a key feature in both of the new consoles (XBox Series and PS5) so I suspect cross-platform titles will have support for it.

Re: Nvidia Ampere GA102 GPU Architecture [pdf]

#19

Post by modeless »

Guthur wrote: post #12
modeless wrote: post #6

Earlier quoted context omitted.

It's a good question actually. Intel tried to make a GPU called Larrabee that was mostly a bunch of small x86 cores with giant vector units. Turns out that it couldn't compete in rendering performance on existing games (in 2010) without the fixed function units that GPUs have, so they canceled it as a GPU. It did result in the AVX-512 instruction set though. I think the idea still has promise but there's a chicken an…

Modern x86 processors are far from simple though and so it's arguable that it's not significantly less complicated. Though to be fair I'm not sure it's really all that complex relatively to modern high end processors. Most of gpu is just the same unit repeated. For mind boggling complexity in my mind is the manufacturing process undertaken by the likes of TSMC.

The Larrabee cores were intentionally simpler than even most 2010 CPUs.

Yeah, modern semiconductor fabrication is pretty much the pinnacle of human achievement. My favorite video on the subject: https://www.youtube.com/watch?v=NGFhc8R_uO4

Re: Nvidia Ampere GA102 GPU Architecture [pdf]

#20

Post by modeless »

CyberDildonics wrote: post #15
modeless wrote: post #6

Earlier quoted context omitted.

It's a good question actually. Intel tried to make a GPU called Larrabee that was mostly a bunch of small x86 cores with giant vector units. Turns out that it couldn't compete in rendering performance on existing games (in 2010) without the fixed function units that GPUs have, so they canceled it as a GPU. It did result in the AVX-512 instruction set though. I think the idea still has promise but there's a chicken an…

They did actually make it and it was not supposed to be a GPU..

They pivoted to HPC when the GPU thing didn't work out, and it has since been discontinued.
Post reply on HN