4 ms·

Nvidia has been powering the AI(ML) progress for the last 10 years and it's because they have the best hardware/software combo for processing vectors. There's

by ru552 3y ago

Nvidia has been powering the AI(ML) progress for the last 10 years and it's because they have the best hardware/software combo for processing vectors. There's nothing lucky about being continuously dominant in a space (GPUs) for 15+ years.

A4ET8a8uTh0 3y ago
Hard disagree. Like with everything, there are multiple factors at play, but at certain size, decisions almost don't matter, because there is no one to upset the apple cart. As a counter, look at MS, they have been around for more than 10 years and there is plenty of luck, hubris, good decisions, bad decisions that did not matter, because they effectively held a dominant position. Being dominant equals to luck.
- Agingcoder 3y ago
  Well in that particular case they became dominant in the gpgpu space because they had a vastly superior offering ( cuda ), which they then capitalized on, while their competitors produced inferior products. I read my first gpgpu papers around 2005 I think , but the one company that ran with it was Nvidia. I don’t see this as luck. Once you are dominant though, you can stay that way for a long time.
  - A4ET8a8uTh0 3y ago
    We disagree, but that is ok. I am admittedly a little biased against nvidia. FWIW, your arguments are good. I think here it is a matter of perspective on what constitutes luck.
    - Agingcoder 3y ago
      https://en.wikipedia.org/wiki/BrookGPU?wprov=sfti1# https://en.wikipedia.org/wiki/BrookGPU?wprov=sfti1# If you’re interested in the early gpgpu work, which interestingly enough worked both on Ati( now amd ) and nvidia hardware. I agree to disagree :-)
Agingcoder 3y ago
This is true, but they’ve been lucky because a gigantic new market appeared which is very friendly to their hardware ( and they make sure it is like you said ). That market could also never have happened at all.
- KaoruAoiShiho 3y ago
  That makes no sense, there's no way AI "could never have happened".
  - Agingcoder 3y ago
    Well it could have happened but have required a non gpu-friendly architecture, it could have happened in 80 years from now and nvidia be long gone, etc. So from their perspective, it wouldn’t have happened at all. But I understand my comment was ambiguous.
    - KaoruAoiShiho 3y ago
      No that doesn't make sense because NVIDIA physically built this field with their own hands. So there's no way it could have "not happened" or "not been GPU friendly" lol. Here is Jensen talking about building the industry 9 months before the founding of OpenAI: https://www.youtube.com/watch?v=_iBLoNG0qHk https://www.youtube.com/watch?v=_iBLoNG0qHk
      - Agingcoder 3y ago
        I disagree, nvidia didn’t build ai. Gpgpu dates back to the early 2000s, and it happens that deep learning as a computational problem is very well suited to GPUs.People figured this out around 2012-2013, and just like with cuda a few years before, nvidia figured out there was a huge market for them, because the underlying compute problem was well suited to their hardware. Ai could have used different models which are not computationally friendly to gpus ( or may not even require a lot of computational power ) and nvidia wouldn’t have been so successful at ai, much like intel’s cpus are not well suited to current deep learning models. I remember discussing with friends in the hpc community what was going to happen to nvidia - we were wondering how they could grow their compute business. And bam, massive compute needs for ai, problem solved.

by ru552 3y ago

A4ET8a8uTh0 3y ago

Hard disagree. Like with everything, there are multiple factors at play, but at certain size, decisions almost don't matter, because there is no one to upset the apple cart. As a counter, look at MS, they have been around for more than 10 years and there is plenty of luck, hubris, good decisions, bad decisions that did not matter, because they effectively held a dominant position. Being dominant equals to luck.

Agingcoder 3y ago
Well in that particular case they became dominant in the gpgpu space because they had a vastly superior offering ( cuda ), which they then capitalized on, while their competitors produced inferior products. I read my first gpgpu papers around 2005 I think , but the one company that ran with it was Nvidia. I don’t see this as luck. Once you are dominant though, you can stay that way for a long time.
- A4ET8a8uTh0 3y ago
  We disagree, but that is ok. I am admittedly a little biased against nvidia. FWIW, your arguments are good. I think here it is a matter of perspective on what constitutes luck.
  - Agingcoder 3y ago
    https://en.wikipedia.org/wiki/BrookGPU?wprov=sfti1# https://en.wikipedia.org/wiki/BrookGPU?wprov=sfti1# If you’re interested in the early gpgpu work, which interestingly enough worked both on Ati( now amd ) and nvidia hardware. I agree to disagree :-)

Agingcoder 3y ago

This is true, but they’ve been lucky because a gigantic new market appeared which is very friendly to their hardware ( and they make sure it is like you said ). That market could also never have happened at all.

KaoruAoiShiho 3y ago
That makes no sense, there's no way AI "could never have happened".
- Agingcoder 3y ago
  Well it could have happened but have required a non gpu-friendly architecture, it could have happened in 80 years from now and nvidia be long gone, etc. So from their perspective, it wouldn’t have happened at all. But I understand my comment was ambiguous.
  - KaoruAoiShiho 3y ago
    No that doesn't make sense because NVIDIA physically built this field with their own hands. So there's no way it could have "not happened" or "not been GPU friendly" lol. Here is Jensen talking about building the industry 9 months before the founding of OpenAI: https://www.youtube.com/watch?v=_iBLoNG0qHk https://www.youtube.com/watch?v=_iBLoNG0qHk
    - Agingcoder 3y ago
      I disagree, nvidia didn’t build ai. Gpgpu dates back to the early 2000s, and it happens that deep learning as a computational problem is very well suited to GPUs.People figured this out around 2012-2013, and just like with cuda a few years before, nvidia figured out there was a huge market for them, because the underlying compute problem was well suited to their hardware. Ai could have used different models which are not computationally friendly to gpus ( or may not even require a lot of computational power ) and nvidia wouldn’t have been so successful at ai, much like intel’s cpus are not well suited to current deep learning models. I remember discussing with friends in the hpc community what was going to happen to nvidia - we were wondering how they could grow their compute business. And bam, massive compute needs for ai, problem solved.