• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Welcome to Archality - AI Inference by Enthusiasts

Hell yeah baby!

It doesn't get much [H]arder than ASIC development. Conceptually this sounds pretty solid. Nvidia is still largely leaning on its legacy model of shoehorning AI workloads down shader pipelines. I don't even need convincing that a smaller company with a fresh approach could beat their current designs. We've seen it happen before.

What I am worried about is the software side. Everybody and their mother is leaning on CUDA and the frameworks which support it (IE: PyTorch). How do you provide equivalent APIs without stepping on legal landmines and convince people to develop bindings for them? Are there plans to add pull requests to OpenCL or work with the Vulkan Compute people? Will these Archality devices show up as standard PCI devices with their own kernel drivers, run exclusively in user mode via proprietary DMA libraries (bypassing the kernel for performance), or are you shipping an entirely new platform that follows its own paradigm?

It might not be gaming hardware, but it's still hardware, and I'm excited to see developments happening somewhere.
 
Noble Aquarius said:
Please by all means put all your eggs in one basket, A.I is a short cut to learn knowledge, The Stock market will crash in 2029, So Please by all means Feed the illusion.
Click to expand...
Let's pretend you're a true precog and know the future, you're more likely a crack head, but we're still going to pretend, and agree the market will crash in 2029. It's 2026 should not a wise man reap profits while they can?
 
FrgMstr said:
Bingo. Have to think outside of the box. Cerebras and Groq are the only players now and those still require huge silicon and memory footprints.
Click to expand...
take my money :D

the same way crypto used up all the GPUs which quickly became overshadowed/rendered all but useless by dedicated ASICs I could definitely see the same thing happening in AI

let me know when you are hiring and/or accepting investors, I have been working hard behind the dumpster at wendys
 
FLECOM said:
take my money :D

the same way crypto used up all the GPUs which quickly became overshadowed/rendered all but useless by dedicated ASICs I could definitely see the same thing happening in AI

let me know when you are hiring and/or accepting investors, I have been working hard behind the dumpster at wendys
Click to expand...

kac77 said:
Interesting. Will keep an eye on this.
Click to expand...

Got our latest simulations back. WOW.
 
Just read the thread. Not sure what took me so long. But basically you're trying to do what ASIC's did for crypto mining. Kick enough ass at performance so that we can have our generalized PC architecture parts back because everyone else will be moving on to that? Do I have this correct?

I'm curious to know why the other chip companies haven't thought of just doing that.

I think it would be interesting also to see distributed computing carry this load too. Like folding. Where you have enthusiast rigs dedicated to chipping in (sorry for the shit pun) to host a large LLM for everyone to benefit from.
 
FrgMstr said:
Got our latest simulations back. WOW.
Click to expand...
Exciting. I hope this works out for you. Not only would it be a huge win financially, but the rest of us benefit from it too. I don't know man. Be careful. I have a feeling that you're gonna be bumping up against the "you'll own nothing and be happy" crowd.
 
You must log in or register to reply here.
Back
Top