Rendered at 13:35:18 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
pcmoritz 19 hours ago [-]
For people interested in this: I've been working on an open-source libtpu implementation for https://tenstorrent.com/ accelerators as a side project, it can already run most of torchtpu, see https://github.com/pcmoritz/libtt (based on the tenstorrent software stack). It shows all the things involved in lowering torch to tile based accelerators.
Thanks for sharing, this is super cool! Right now my approach to correctness it to just run as many tests as possible and fix them (torchtpu greatly expands the amount of available tests), but formal verification is also super cool. Having a look at your book https://lean.brettkoonce.com/blueprint.pdf
Thanks for the encouragement, if you have the hardware and end up trying it out I'd love to get your feedback (feel free to drop an issue in the repo) :)
yhjc2692 18 hours ago [-]
I would love to do that, but unfortunately don’t have one on hand (yet?). In fact, I’m curious - did you buy a TT device to work on this, or are you using a cloud provider like Koyeb?
pcmoritz 18 hours ago [-]
I first built my own to try it out (the exact hardware is described in https://relativity.cc/hardware/ under "one Blackhole card") and when I liked the hardware I got a quietbox 2. It is very high quality but a little expensive (but inexpensive compared to other LLM hardware and it has excellent interconnects which is very rare in the consumer/dev space).
dalemhurley 17 hours ago [-]
Google and Apple are going to move consumers and ordinary business users to edge inference. This is more disruptive to the upcoming IPOs than most investors understand.
lofties 13 hours ago [-]
Google, Apple, and don't forget our friends DeepSeek!
anuptalwalkar 14 hours ago [-]
I would love to see them move off into edge computing. There should be some autonomy and siloed experience in day to day usage.
tjwebbnorfolk 15 hours ago [-]
I would love to be able to buy a TPU for inference
They are just wildly outdated for most tasks. Even projects like frigate suggest buying different hardware. Coral was released in 2019 and hasn't been updated once. With a jetson nano or even a raspi ai hat you get way more tops
emmelaich 12 hours ago [-]
You can buy a Google Coral TPU, does that count?
Dr4kn 7 hours ago [-]
With them being released in 2019 with zero hardware revisions they are just outdated for almost every workload
npalli 21 hours ago [-]
Interesting, Amazon doing the same for AWS Trainium.
Pytorch conference is happening today for explaining the coincidental timing
sroussey 19 hours ago [-]
That explains the 5x speed increases huggingface mentioned for Mac using metal.
mathisfun123 18 hours ago [-]
Link?
fancyfredbot 19 hours ago [-]
Cool! XLA is pretty good + potentially a better compiler infrastructure than inductor which will lead to improved performance on all hardware.
Also I guess this is the final nail in the coffin of Tensorflow.
tingkuanpei 37 minutes ago [-]
Very interesting work. Google should have done this a long time ago. PyTorch's ecosystem is just so rich.
ismailmaj 18 hours ago [-]
Amazing news, I remember the days when you had to fight XLA to make PyTorch run on TPUs, it seems we finally got first-party support!
Since HW specifics live mostly in the kernel layer (Pallas or CUDA), I imagine the Jax vs PyTorch question is more of a style preference now?
nico 18 hours ago [-]
Pardon my ignorance, but who else besides Google has TPUs? Are they commercially available somewhere? What’s their main application right now outside of Google?
cavisne 18 hours ago [-]
You can rent them on cloud + they sell directly to Anthropic. Its pretty well known that Anthropic is a heavy user of TPU's (along with Trainium & GPU's)
atty 10 hours ago [-]
Google is selling TPUs to enterprise clients for on prem deployments. We’ve evaluated them but decided to stick with Nvidia, at least for one more generation. We’re a relatively large user of GCP.
Google is selling TPU shovels via Cloud services and also physically to select partners. So this makes sense.
There will be an abundance of shovels. Nebius already crashed by almost half from its hype peak.
airza 20 hours ago [-]
it doesn't seem like there's an abundance of shovels coming anytime soon
tjwebbnorfolk 15 hours ago [-]
Everyone says there's a massive compute shortage. But where is it showing up? For whom is there a shortage?
Every time I ask Claude to do something, it does it just fine. I never have to wait in any sort of queue that would be common in other kinds of shortages.
eitally 14 hours ago [-]
It's showing up for enterprises (and ISVs) who need capacity from cloud vendors and can't get it. Sometimes there are some GPU shapes available, but not always what you either need or want to pay for (e.g. a TPU vs a V100 vs a B300), and also not always in spot instances (frequently not).
The hyperscalers are all so oversubscribed that they require commits from enterprises so if you only have a transient need you might be out of luck [from your vendor of choice]. Additionally, not all GPU service providers are equal. Semi Analysis has some great articles explaining why.
octoberfranklin 12 hours ago [-]
But where is it showing up?
Pricing -- exactly where shortages show up in a free market economy.
juleiie 17 hours ago [-]
Shovels and gold aren’t the problem, humans are the problem
either AI achieves hyper-efficiency and triggers systematic economic demand destruction due to unemployment, or it underperforms and sparks a massive tech-sector asset bubble burst
If AI is as good as valuations - it crashes. If AI isn’t as good, it crashes. all eggs in one basket economy.
Every future leads to a massive financial shitstorm
AI is priced in as a technology to end all markets. The last bull market you will ever see.
It’s a miracle markets didn’t notice this “damned if you succeed damned if you don’t” type of inevitable checkmate
I have been building a bunker
bsder 13 hours ago [-]
> It’s a miracle markets didn’t notice this “damned if you succeed damned if you don’t” type of inevitable checkmate
They learned from 2008 that Too Big Too Fail(tm) means the government bails you out.
This was the problem with not OBLITERATING the banks in 2008. Had the US government taken over the banks and wiped the investors, there would be more pushback right now.
omgwtfbyobbq 16 hours ago [-]
It's possible. If goog's AI overview is accurate, it seems somewhat similar to speculative euphoria seen in the past (eg companies involved in electricity prior to the great depression). It's larger (currently 35% for AI versus 25% for electric/related cos), I'm guessing because of more credit. At the same time, the housing component of the market prior to the GFC was about 23%, and it wasn't as economically damaging as the GD, so hopefully the AI bubble unwinding won't be as bad.
esseph 16 hours ago [-]
> so hopefully the AI bubble unwinding won't be as bad
It's currently larger than the 1880s railroad boom.
"Overbuilding and speculative debt ultimately culminated in the Panic of 1893, plunging the nation into a severe economic depression."
omgwtfbyobbq 15 hours ago [-]
Not according to Goog's. The railroad bubble of the 1880s was supposedly 60+% of the market. It may not be accurate though.
"During the 1880s, private railroad enterprises represented over 60% of the total capitalization of the entire U.S. stock market, peaking at 63% in 1881 and 1885."
fasterik 16 hours ago [-]
[dead]
Lucasoato 21 hours ago [-]
Are there TPUs outside Google Cloud?
boulos 20 hours ago [-]
Yes. Google is famously selling TPUs on prem to big customers (e.g., Anthropic) and Crux AI is now a venture to do just that.
sschueller 21 hours ago [-]
Google coral USB and m.2 devices are TPUs.
Archit3ch 19 hours ago [-]
"Have you seen any software run on Coral TPUs?"
"No, but I have seen the house that the project built, and it is terrific." ~ some Google director
Is a pretty successful open source NVR that used corals when it launched (and many people still use them with it)
qlte 7 hours ago [-]
Yeah I have one for Frigate on a Raspberry Pi. Has worked well but definitely showing its age performance wise. I got a cheap mini PC (right before prices like quadrupled) which has about the same performance on an iGPU). I wouldn't be surprised if Frigate was the funnel for like 90% of Coral purchases or more.
boulos 20 hours ago [-]
Only in marketing name. They're not related otherwise.
goda90 20 hours ago [-]
Maybe just marketing, but they call the NPUs in their Pixel phones "TPUs"
fulafel 8 hours ago [-]
TPUs are really matrix processors (they process 2d tensors right?), it's technically fair to call most matrix processors TPUs.
westurner 21 hours ago [-]
TPU / neural compute accelerator / Math coprocessor;
> How does Cerebras WSE-3 with 44GB of 'L2' on-chip SRAM compare to Google's TPUs, Tesla's TPUs, NorthPole, Groq LPU, Tenstorrent's, and AMD's NPU designs?
There are TPUs made out of silicon carbide with standard fab.
There is in-flight tensor processing with photons, but that's not quite a TPU.
hackernudes 20 hours ago [-]
I think the commenter was asking if Google sells TPU chip hardware to use outside of Google's own infrastructure and cloud.
g58892881 7 hours ago [-]
great! where can I buy a TPU?
ranger_danger 42 minutes ago [-]
How about open-sourcing the proprietary blackbox Titan security chip, that you pinky-promised to open but never did?
frobisher 7 hours ago [-]
Question: does this make JAX obsolete? I was having fun learning it!
csvance 1 hours ago [-]
It makes PyTorch work nicely on TPU, which is impressive mostly because of how unstructured PyTorch is with its conventions. From a compiler standpoint JAX is much easier to work with since its all pure functions. This means that JAX also has access to fairly cutting edge compiler tech like Enzyme which is quite relevant for scientific machine learning work.
nelaggy 20 hours ago [-]
so if the moat isn't the model and the moat isn't the inference layer, i guess the moat is the hardware now? what does this mean for frontier labs?
badrequest 19 hours ago [-]
hardware has always been the moat in the AI space
AnotherGoodName 18 hours ago [-]
Yeah a lot of people bought into the idea of CUDA being a moat but we’re well past the point where it’s pretty trivial to jump to another equivalent stack on the software side.
jacquesm 18 hours ago [-]
Those are the two main interfaces: the inference hardware and the endpoints. Endpoints are pretty much standardized to the point that you can switch them out in seconds. CUDA still has an edge but every release like this chips away at that. Even so, TPUs are unobtanium so while I appreciate Google's effort here I also recognize that it is not something that will be useful to me.
In that sense AMD is still better positioned to give NVidia a run for their money but given the family connection and AMD's underdog status I am not very hopeful that we'll see meaningful competition in inference hardware any time soon.
NVidia is a much harder nut to crack than OpenAI or Anthropic. They are easily a decade ahead of viable competition. Check out the GB300 if you don't believe that.
kavalg 7 hours ago [-]
While I have the same sentiment, I am also curious why a competitor can't accelerate their developments with AI. In the end, hardware development is to a large extent also software development. Is it the patents that NVIDIA has that are preventing this? Or maybe some trade secrets? AFAIK the NVIDIA hardware is manufactured entirely by TSMC.
nixon_why69 9 hours ago [-]
A decade is a long time and the highest end isn't the only market segment.
High bandwidth flash will be here soon, I bet someone makes an economical competitor thats not quite as fast but much cheaper.
coredog64 18 hours ago [-]
Even if it's allegedly possible, the devil has always been in the details. It always sucks to find out that there's some quirk of your model architecture that breaks an implicit assumption that underpins the export process.
yaoke259 10 hours ago [-]
[dead]
AndrewKemendo 19 hours ago [-]
Will always be true
JacobAsmuth 20 hours ago [-]
Products can also be a moat.
yipinwong 20 hours ago [-]
Dumb question from non-AI engineer.
Is this somehow Google's answer to today's best DeepSeek open-sourcing inferencing engine?
bitpush 20 hours ago [-]
I dont think big companies like Google react to news like that. Very likely this was meant to go out on Tuesday for weeks now. And it just happened to have coincided with another news.
JacobAsmuth 20 hours ago [-]
Yeah they read the news this morning and whipped up a quick Torch->XLA translation layer and coordinated with four different inference engines (who also all wrote their own blog posts) and shoved this out the door quick to stay on top of the news cycle
parthdesai 19 hours ago [-]
Why be rude on purpose? OP clearly prefaced by saying it was a dumb question
caaqil 19 hours ago [-]
[flagged]
luckydata 16 hours ago [-]
[flagged]
mathisfun123 20 hours ago [-]
they're releasing this today because the pytorch conf is this week...
0x00cl 13 hours ago [-]
We might never know, but they probably didn't read the news and started working to get this out, they were probably working on it for a while now.
Personally the question is if they are "saving" other projects or news so they can be released in case a competitor makes a move? Who knows if this was ready days or weeks ago and were waiting on someone to drop some news, so they can shadow those news.
geremiiah 19 hours ago [-]
No, these are two very different objectives.
Tanjreeve 20 hours ago [-]
They do different things and sit in different parts of the stack (training vs inference). TorchTPU seems more like a play to get more people familiar with Googles TPU chips as they want to sell them to people outside of google.
beardedetim 19 hours ago [-]
Yeah I'd assume this is to allow us to train on their TPUs with torch instead of needing to rent a GPU
yipinwong 19 hours ago [-]
ty. that helps a lot and know where to research next.
The Tenstorrent ISA is documented btw https://github.com/tenstorrent/tt-isa-documentation, I wonder if one could do the verification end-to-end with the hardware.
https://github.com/aws-neuron/torch-neuronx
Also I guess this is the final nail in the coffin of Tensorflow.
Since HW specifics live mostly in the kernel layer (Pallas or CUDA), I imagine the Jax vs PyTorch question is more of a style preference now?
There will be an abundance of shovels. Nebius already crashed by almost half from its hype peak.
Every time I ask Claude to do something, it does it just fine. I never have to wait in any sort of queue that would be common in other kinds of shortages.
The hyperscalers are all so oversubscribed that they require commits from enterprises so if you only have a transient need you might be out of luck [from your vendor of choice]. Additionally, not all GPU service providers are equal. Semi Analysis has some great articles explaining why.
Pricing -- exactly where shortages show up in a free market economy.
either AI achieves hyper-efficiency and triggers systematic economic demand destruction due to unemployment, or it underperforms and sparks a massive tech-sector asset bubble burst
If AI is as good as valuations - it crashes. If AI isn’t as good, it crashes. all eggs in one basket economy.
Every future leads to a massive financial shitstorm
AI is priced in as a technology to end all markets. The last bull market you will ever see.
It’s a miracle markets didn’t notice this “damned if you succeed damned if you don’t” type of inevitable checkmate
I have been building a bunker
They learned from 2008 that Too Big Too Fail(tm) means the government bails you out.
This was the problem with not OBLITERATING the banks in 2008. Had the US government taken over the banks and wiped the investors, there would be more pushback right now.
It's currently larger than the 1880s railroad boom.
"Overbuilding and speculative debt ultimately culminated in the Panic of 1893, plunging the nation into a severe economic depression."
"During the 1880s, private railroad enterprises represented over 60% of the total capitalization of the entire U.S. stock market, peaking at 63% in 1881 and 1885."
"No, but I have seen the house that the project built, and it is terrific." ~ some Google director
Is a pretty successful open source NVR that used corals when it launched (and many people still use them with it)
"OpenTPU: Open-Source Reimplementation of Google Tensor Processing Unit (TPU)" https://news.ycombinator.com/item?id=44111452
From https://news.ycombinator.com/item?id=45848676 :
> How does Cerebras WSE-3 with 44GB of 'L2' on-chip SRAM compare to Google's TPUs, Tesla's TPUs, NorthPole, Groq LPU, Tenstorrent's, and AMD's NPU designs?
There are TPUs made out of silicon carbide with standard fab.
There is in-flight tensor processing with photons, but that's not quite a TPU.
In that sense AMD is still better positioned to give NVidia a run for their money but given the family connection and AMD's underdog status I am not very hopeful that we'll see meaningful competition in inference hardware any time soon.
NVidia is a much harder nut to crack than OpenAI or Anthropic. They are easily a decade ahead of viable competition. Check out the GB300 if you don't believe that.
High bandwidth flash will be here soon, I bet someone makes an economical competitor thats not quite as fast but much cheaper.
Is this somehow Google's answer to today's best DeepSeek open-sourcing inferencing engine?
Personally the question is if they are "saving" other projects or news so they can be released in case a competitor makes a move? Who knows if this was ready days or weeks ago and were waiting on someone to drop some news, so they can shadow those news.