For those in the industry:
When a new generation like this is released, will a typical AI company replace the current GPUs? Is there a chance to acquire the older versions for private use or is it too early for that?
Nvidia CEO Introduces Nvidia Ampere Architecture, Nvidia A100 GPU
21–30 of 347 posts
Re: Nvidia CEO Introduces Nvidia Ampere Architecture, Nvidia A100 GPU
#22Probably even more closed than ever. They tend to become more and more restrictive with every new hardware generation. I wonder where their promised open source announcement they preannounced before.
I suspect some of it is driven by trying to keep the gaming/ai and desktop/server markets from overlapping. Market segmentation. If it were more open, that would be harder.
Re: Nvidia CEO Introduces Nvidia Ampere Architecture, Nvidia A100 GPU
#23Probably even more closed than ever. They tend to become more and more restrictive with every new hardware generation. I wonder where their promised open source announcement they preannounced before.
Yeah. That's been my general problem with adopting NVidia for anything. They make good hardware, but there's a lot of lock-in, and not a lot of transparency. That introduces business risk.
I'm not in a position where I need GPGPU, but if there wasn't that risk, and generally there were mature, open standards, I'd definitely use it. The major breakpoint would be when libraries like Numpy do it natively, and better yet, when Python can fork out list comprehensions to a GPU. I think at that point, the flood gates will open up, and NVidia's marketshare will explode from specialized applications to everywhere.
Intel stumbled into it by accident, but got it right with x86. Define an open(ish) standard, and produce superior chips to that standard. Without AMD, Cyrix, Via, and the other knock-offs, there would be no Intel at this point.
Intel keeps getting it right with numerical libraries. They're open. They work well. They work on AMD. But because Intel is building them, Intel has that slight bit of advantage. If Intel's open libraries are even 5% better on Intel, that's a huge market edge.
Re: Nvidia CEO Introduces Nvidia Ampere Architecture, Nvidia A100 GPU
#24If the demonstrated speed ups translate to real world performance, then I’m truly blown away. Looks like Nvidia will be holding onto the AI crown a while longer.
The only thing I wonder is how difficult is it to take advantage of some of the new arch features, such as TF32 format or sparsity tensor ops.
Re: Nvidia CEO Introduces Nvidia Ampere Architecture, Nvidia A100 GPU
#25Probably even more closed than ever. They tend to become more and more restrictive with every new hardware generation. I wonder where their promised open source announcement they preannounced before.
The large majority of researchers and business getting into NVidia products doesn't seem to find it that relevant, rather what tools, GPU programming languages and hardware they are able to put their hands on.
Re: Nvidia CEO Introduces Nvidia Ampere Architecture, Nvidia A100 GPU
#26Stealing Epic Games/Unreal Engine's thunder.
If Nvidia was a human, they'd be the type to propose at someone else's wedding.
Re: Nvidia CEO Introduces Nvidia Ampere Architecture, Nvidia A100 GPU
#27If the demonstrated speed ups translate to real world performance, then I’m truly blown away. Looks like Nvidia will be holding onto the AI crown a while longer. The only thing I wonder is how difficult is it to take advantage of some of the new arch features, such as TF32 format or sparsity tensor ops.
Have you looked at Apex.Amp? TF32 sounds to be along similar lines, and the PyTorch usage is a breeze.
Re: Nvidia CEO Introduces Nvidia Ampere Architecture, Nvidia A100 GPU
#28Probably even more closed than ever. They tend to become more and more restrictive with every new hardware generation. I wonder where their promised open source announcement they preannounced before.
Yeah. That's been my general problem with adopting NVidia for anything. They make good hardware, but there's a lot of lock-in, and not a lot of transparency. That introduces business risk. I'm not in a position where I need GPGPU, but if there wasn't that risk, and generally there were mature, open standards, I'd definitely use it. The major breakpoint would be when libraries like Numpy do it natively, and better yet…
Given that I hardly saw any clone vendors other than AMD, I really doubt that they had any influence on Intel's market share.
What worked out was IBM not being able to prevent PC clones, but given the wide adoption of laptops, tablets and phones that hardly matters nowadays.
Re: Nvidia CEO Introduces Nvidia Ampere Architecture, Nvidia A100 GPU
#297 times V100 performance for BERT. That is insane!
Mind you, this is an optimised version of BERT, from what I could see in their blog post.
Re: Nvidia CEO Introduces Nvidia Ampere Architecture, Nvidia A100 GPU
#30Earlier quoted context omitted.
I'm curious: why exactly do you need double precision digits? Not dismissing, just wondering what kind of application needs it.
Physics simulations. There's a rule of thumb that to get an n-bit accurate result after a long chain of calculations, intermediate results should be stored with 2n bits. Often using the full dynamic range of a float is necessary because the magnitude of different physical phenomena varies so wildly. I guess people do store intermediate results in floats in order to take advantage of GPU acceleration. However, once yo…
I've always argued that if you are getting close to having to worry about underflow, overflow etc. then you have an ill-conditioned problem and just increasing the size of your intermediate results won't help you a huge amount because you need more precision from your inputs. There are very few fields where you need more than the 7 decimal digits afforded by floats. Maybe the only exceptions are in astrophysics.