They even wrote a whole article about that assumption, and it wasn't very convincing. DAXPY is a blatantly unfair example, feels like they don't actually use SIMD for anything real.https://www.sigarch.org/simd-instructions-considered-harmful/ …
-
-
-
Yes, I read that article and it literally didn't address a single one of the concerns I have about variable-length vector instructions :(
- Show replies
New conversation -
-
-
I think they’re right, though. The reason everything has packed and not variable is that everyone ignored what Cray did in favour of doing the more obvious (and easier) thing.
-
I don't think that's true? I have not seen any discussion of what I think the obvious problems are with arbitrary-length vector instructions, so I'm not convinced that the dismissal is correct...
- Show replies
New conversation -
-
-
What is the name of the book
-
It's from the Volume I: Unprivileged ISA spec here:https://riscv.org/technical/specifications/ …
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.