Re:Run CUDA, unmodified, on AMD GPUs
by karolist on Monday July 15, 2024 @04:13PM
(#40971091)
Was there a large entity steering x86 spec alone with a huge feature lead against their competition, free to steer the spec in any ways they choose? Also, hardware is not opensource software, you get big players onboard and they will be able to implement the spec they want every gen, software has more moving parts and unaligned parties involved.
[ Parent
| on HN ]
Re:Run CUDA, unmodified, on AMD GPUs
by cherryteastain on Monday July 15, 2024 @04:22PM
(#40971173)
> Was there a large entity steering x86 spec alone with a huge feature lead against their competition, free to steer the spec in any ways they choose?
Ever heard of Intel?
I had't considered that angle. Is your point that Intel was the creator of x86, but software chose to support it, then AMD had nothing else but to play catch up in x86 support to be part of the software target market? If so and factual (I've no idea), fair point, I didn't know.
It was exactly the same instruction set.
C compilers didn't offer an "AMD" CPU target* until AMD came out with the "AMD64" instruction set. Today we call this "x86_64" or "x64".
* Feel free to point out some custom multimedia vector extensions for Athlons or something, but the point remains.
And Intel named its licenced implementation of AMD64 as IA-32e, just to make it clear to everyone that it is based on Intel architecture 32bit version with an extension.
Luckily they dropped that name few years later