3 ms·

>~75–80% of S&P 500 gains Dot com bust was 50% of S&P / 80% on Nasdaq... we survived it.. For Nvidia: all it will take is for one smart grad student to find a

>~75–80% of S&P 500 gains Dot com bust was 50% of S&P / 80% on Nasdaq... we survived it.. For Nvidia: all it will take is for one smart grad student to find a better training algorithm to destroy 75% of their value. Why I think a better algorithm is out there: Total installed GPU Tflops = 4 billion. Tflops of human brain = 1 million

Yoric 1y ago
> For Nvidia: all it will take is for one smart grad student to find a better training algorithm to destroy 75% of their value. I don't think so. We've already seen several generations of better training algorithms, but they all rely on CUDA, hence on NVidia.
- jhallenworld 1y ago
  For inference they already face competition: that deal OpenAI announced with AMD .. at a certain point it will be worth broadening away from CUDA.
  - Yoric 1y ago
    Possibly, yes, but so far, it's barely noticeable, hardly a threat.
  - p4ul 1y ago
    Yep, and don't forget about Mojo, which has the potential to make such a transition less painful.
cantor_S_drug 1y ago
But the flip side is excess capacity will be used for other fields which are not getting a chance.
- Yoric 1y ago
  On the other hand, GPUs die down fairly quickly, so this excess capacity had better prove useful rather quickly.
adidoit 1y ago
Is that true? I was under the impression that one reason NVIDIA is so entrenched is that their GPUs are highly flexible. Versus the ASIC offerings of other specialized players are more focused on transformers remaining the dominant architecture.
- jhallenworld 1y ago
  Maybe the better algorithm will be able to run on GPUs.
baq 1y ago
the 1 million is debatable, but you can't argue with the power consumption of 20W. that said, if all the current infrastructure could run brain-equivalent models at 20W each, I'd wager we would have much more demand than currently.
- Windchaser 1y ago
  There's speed, efficiency, and determinism. Pick 2
quantumspandex 1y ago
Computation on human brain is on a totally different substrate than silicon. It's in memory and highly error prone. It's questionable a mere algorithm would get us there without a fundamental change in computer architecture. (in terms of Intelligence / W)

Yoric 1y ago

> For Nvidia: all it will take is for one smart grad student to find a better training algorithm to destroy 75% of their value. I don't think so. We've already seen several generations of better training algorithms, but they all rely on CUDA, hence on NVidia.

jhallenworld 1y ago
For inference they already face competition: that deal OpenAI announced with AMD .. at a certain point it will be worth broadening away from CUDA.
- Yoric 1y ago
  Possibly, yes, but so far, it's barely noticeable, hardly a threat.
- p4ul 1y ago
  Yep, and don't forget about Mojo, which has the potential to make such a transition less painful.

cantor_S_drug 1y ago

But the flip side is excess capacity will be used for other fields which are not getting a chance.

Yoric 1y ago
On the other hand, GPUs die down fairly quickly, so this excess capacity had better prove useful rather quickly.

adidoit 1y ago

Is that true? I was under the impression that one reason NVIDIA is so entrenched is that their GPUs are highly flexible. Versus the ASIC offerings of other specialized players are more focused on transformers remaining the dominant architecture.

jhallenworld 1y ago
Maybe the better algorithm will be able to run on GPUs.

baq 1y ago

the 1 million is debatable, but you can't argue with the power consumption of 20W. that said, if all the current infrastructure could run brain-equivalent models at 20W each, I'd wager we would have much more demand than currently.

Windchaser 1y ago
There's speed, efficiency, and determinism. Pick 2

quantumspandex 1y ago

Computation on human brain is on a totally different substrate than silicon. It's in memory and highly error prone. It's questionable a mere algorithm would get us there without a fundamental change in computer architecture. (in terms of Intelligence / W)