←back to thread

Ask HN: Why hasn’t AMD made a viable CUDA alternative?

183 points spacebanana7 | 5 comments | 01 Apr 25 14:37 UTC | HN request time: 0.21s | source

I appreciate developing ROCm into something competitive with CUDA would require a lot of work, both internally within AMD and with external contributions to the relevant open source libraries.

However the amount of resources at stake is incredible. The delta between NVIDIA's value and AMD's is bigger than the annual GDP of Spain. Even if they needed to hire a few thousand engineers at a few million in comp each, it'd still be a good investment.

1. gmm1990 ◴[01 Apr 25 14:56 UTC] No.43547545[source]▶
The idea that CUDA is the main reason behind Nvidia dominance seems strange to me. If most of the money is coming from Facebook and Microsoft they have their own teams writing code at a lower level than CUDA anyway. Even deepseek was writing stuff lower than that.
replies(2): >>43549250 #>>43565743 #
2. brcmthrowaway ◴[01 Apr 25 17:15 UTC] No.43549250[source]▶
What is lower than CUDA? Compute shaders?
replies(1): >>43549537 #
3. a5ehren ◴[01 Apr 25 17:39 UTC] No.43549537[source]▶
PTX assembly. Deepseek used some of it to do a little bit of work that CUDA didn't have APIs for.
replies(1): >>43553196 #
4. brcmthrowaway ◴[02 Apr 25 02:32 UTC] No.43553196{3}[source]▶
Sadly platform specific
5. ActorNightly ◴[03 Apr 25 06:51 UTC] No.43565743[source]▶
Its not efficient to write low level code. Its like trying to build a web application with C.

CUDA is basically the equivalent of "hey you wanna do matrix multiply fast? just use this and we handle the rest". And its standardized, and when Nvidia comes out with new cards, all the old shit still works.