4 ms·

"I estimate having software on par with NVDA would raise their market cap by 100B. Then you estimate what the chance it that @tinygrad can close that gap,

"I estimate having software on par with NVDA would raise their market cap by 100B. Then you estimate what the chance it that @__tinygrad__ can close that gap, say it's 0.1%, probably a very low estimate when you see what we have done so far, but still... That's worth 100M. And they won't even send us 2 ~100k boxes. In what world does that make sense, except in a world where decisions are made based on pride instead of ROI. Culture issue." https://x.com/__tinygrad__/status/1879620242315317304 https://x.com/__tinygrad__/status/1879620242315317304

AshamedCaptain 2y ago
I would really like to see a concrete, legit way to materialize a "100M raise in market cap" into actual ROI ...
- rikafurude21 2y ago
  When the market cap rises, price of shares goes up? Do you know what a market cap is?
  - carlmr 2y ago
    Yes, but the company doesn't get more money from that. The only, way to get money out of it is by selling shares at the new price. However it would also raise future revenue, which should be what's reflected by the market. So it would still be something that's good for the company, but not nearly 100B good.
    - rikafurude21 2y ago
      You dont think AMD being competitive with Nvidia (3,37 trillion USD MC) would be "nearly 100B good"? Believe it or not the only reason thats not the case is good bug-free software. Thats what tinygrad is doing
      - carlmr 2y ago
        Again, market cap is not what the company has. It's what the market believes the company should be worth accounting for future earnings.
latchkey 2y ago
This is his opinion, nothing more, nothing less. He currently has a partially implemented piece of software that hasn't seen a release since November and isn't performant at all. Take the free offer, prove everyone wrong and then start to tell us how great you are. https://x.com/HotAisle/status/1880507210217750550 https://x.com/HotAisle/status/1880507210217750550
- FeepingCreature 2y ago
  To be fair, having seen his software evolve, and having seen ROCm evolve, I'm more optimistic for his software in a year than yours. He picked his problem better. The whole reason that tinygrad is, well, tiny, is that it limits the amount of overhead to onboard people and perform maintenance and rewrites. My strong impression is that the ROCm codebase is simply much too large for AMD's dev resources. You're trying to race NVidia on their turf with less resources. It's brave, but foolish. I can see how Tinygrad could succeed. The story makes sense. AMD's doesn't, neither logically nor empirically. NVidia would have to seriously fumble.
  - llm_trw 2y ago
    >NVidia would have to seriously fumble. Worked for AMD in the CPU market. That said I'm deeply worried about anyone whose based their company on amd gpus. The only reason why they do well in hpc is because there's an army of dreadfully underpaid and over performing grand students to pick up the slack from AMD. Trying to do that in a corporate environment is company suicide.
    - modeless 2y ago
      TSMC is more responsible for AMD's success vs. Intel than AMD is. Unfortunately for AMD, Nvidia uses TSMC too.
      - jorvi 2y ago
        3D-Cache blows Intel out of the water and has absolutely nothing to do with TSMC. Same goes for the clever chiplet design.
catgary 2y ago
AMD already has major ongoing projects with OpenXLA/IREE. Lots of established engineers/researchers, and it’s in collaboration with Google/AWS. Hotz is delusional if he thinks that he can do better by ripping off Karpathy’s toy autograd implementation.
- cyberax 2y ago
  > AMD already has major ongoing projects with OpenXLA/IREE. And how's that been going? The AMD stock price compared to NVidia seems to speak volumes about the efficacy of these projects. IREE has been around for 5 years, without producing anything overtly practical. They seem to be focused more on academic jobs and citations. It's also focused on the general case of a compiler for "all" AI-type tasks, supporting everything from WASM to CUDA. OpenXLA seems to be a bit more practical, but I spent the last 2 hours trying to make it work on my AMD card (Radeon Pro W7900) and failing. I personally don't like Tinygrad's approach of doing their own thing rather than integrating into PyTorch/JAX/..., but it at least is _practical_ with a reasonable end-goal. Is it going to be successful? Who knows. But it's more practical than anything AMD has done within the recent 5 years.
  - catgary 2y ago
    I am an ML scientist, my company and several others are using IREE to deploy our models to edge devices. It is the most promising technology in this area. Those academic publications are a sign that the people involved actually know what they’re doing, and are making sure their work holds up to scrutiny.
- juliangoldsmith 2y ago
  How does Tinygrad fall short? Performance is fine [0]. It's much smaller than Pytorch and all, but that's kind of in the name. I've been hearing about MLIR and OpenXLA for years through Tensorflow, but I've never seen an actual application using them. What out there makes use of them? I'd originally hoped it'd allow Tensorflow to support alternate backends, but that doesn't seem to be the case. 0: https://cprimozic.net/notes/posts/machine-learning-benchmarks-on-the-7900-xtx/ https://cprimozic.net/notes/posts/machine-learning-benchmark...

"I estimate having software on par with NVDA would raise their market cap by 100B. Then you estimate what the chance it that @__tinygrad__ can close that gap,

AshamedCaptain 2y ago

I would really like to see a concrete, legit way to materialize a "100M raise in market cap" into actual ROI ...

rikafurude21 2y ago
When the market cap rises, price of shares goes up? Do you know what a market cap is?
- carlmr 2y ago
  Yes, but the company doesn't get more money from that. The only, way to get money out of it is by selling shares at the new price. However it would also raise future revenue, which should be what's reflected by the market. So it would still be something that's good for the company, but not nearly 100B good.
  - rikafurude21 2y ago
    You dont think AMD being competitive with Nvidia (3,37 trillion USD MC) would be "nearly 100B good"? Believe it or not the only reason thats not the case is good bug-free software. Thats what tinygrad is doing
    - carlmr 2y ago
      Again, market cap is not what the company has. It's what the market believes the company should be worth accounting for future earnings.

latchkey 2y ago

This is his opinion, nothing more, nothing less. He currently has a partially implemented piece of software that hasn't seen a release since November and isn't performant at all. Take the free offer, prove everyone wrong and then start to tell us how great you are. https://x.com/HotAisle/status/1880507210217750550 https://x.com/HotAisle/status/1880507210217750550

FeepingCreature 2y ago
To be fair, having seen his software evolve, and having seen ROCm evolve, I'm more optimistic for his software in a year than yours. He picked his problem better. The whole reason that tinygrad is, well, tiny, is that it limits the amount of overhead to onboard people and perform maintenance and rewrites. My strong impression is that the ROCm codebase is simply much too large for AMD's dev resources. You're trying to race NVidia on their turf with less resources. It's brave, but foolish. I can see how Tinygrad could succeed. The story makes sense. AMD's doesn't, neither logically nor empirically. NVidia would have to seriously fumble.
- llm_trw 2y ago
  >NVidia would have to seriously fumble. Worked for AMD in the CPU market. That said I'm deeply worried about anyone whose based their company on amd gpus. The only reason why they do well in hpc is because there's an army of dreadfully underpaid and over performing grand students to pick up the slack from AMD. Trying to do that in a corporate environment is company suicide.
  - modeless 2y ago
    TSMC is more responsible for AMD's success vs. Intel than AMD is. Unfortunately for AMD, Nvidia uses TSMC too.
    - jorvi 2y ago
      3D-Cache blows Intel out of the water and has absolutely nothing to do with TSMC. Same goes for the clever chiplet design.

catgary 2y ago

AMD already has major ongoing projects with OpenXLA/IREE. Lots of established engineers/researchers, and it’s in collaboration with Google/AWS. Hotz is delusional if he thinks that he can do better by ripping off Karpathy’s toy autograd implementation.

cyberax 2y ago
> AMD already has major ongoing projects with OpenXLA/IREE. And how's that been going? The AMD stock price compared to NVidia seems to speak volumes about the efficacy of these projects. IREE has been around for 5 years, without producing anything overtly practical. They seem to be focused more on academic jobs and citations. It's also focused on the general case of a compiler for "all" AI-type tasks, supporting everything from WASM to CUDA. OpenXLA seems to be a bit more practical, but I spent the last 2 hours trying to make it work on my AMD card (Radeon Pro W7900) and failing. I personally don't like Tinygrad's approach of doing their own thing rather than integrating into PyTorch/JAX/..., but it at least is _practical_ with a reasonable end-goal. Is it going to be successful? Who knows. But it's more practical than anything AMD has done within the recent 5 years.
- catgary 2y ago
  I am an ML scientist, my company and several others are using IREE to deploy our models to edge devices. It is the most promising technology in this area. Those academic publications are a sign that the people involved actually know what they’re doing, and are making sure their work holds up to scrutiny.
juliangoldsmith 2y ago
How does Tinygrad fall short? Performance is fine [0]. It's much smaller than Pytorch and all, but that's kind of in the name. I've been hearing about MLIR and OpenXLA for years through Tensorflow, but I've never seen an actual application using them. What out there makes use of them? I'd originally hoped it'd allow Tensorflow to support alternate backends, but that doesn't seem to be the case. 0: https://cprimozic.net/notes/posts/machine-learning-benchmarks-on-the-7900-xtx/ https://cprimozic.net/notes/posts/machine-learning-benchmark...