Thumb-1 was designed explicitly as compiler target. You can tell this because it has all the orthogonality of a sphere.
-
-
if you need more registers, that’s what cache is for, right?
-
Yes but it's not "more registers" it's "wider SIMD registers" (bs like AVX512).
-
gotcha, same thing in my book I’m glad AVX-512 looks as ridiculous to not-me
-
AVX-512 combines absolutely critical improvements to the the x86 SIMD ISA with wider registers that very few people want.
-
oh right it’s a catch-all term for a bunch of different extensions, right?
-
what’s the useful stuff
-
Real Permutes, conversion between floating point and unsigned, orthogonality of integer arithmetic (i.e. most ops available for every type, signed and unsigned), assorted nice-to-have features for writing math libraries.
-
Conflict detect, predication, extract/deposit enable vectorizing some stuff that was previously unvectorizable.
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.