12:59:29scymtymdoug wrote on the sbcl-devel list "I have a series of changes that rename 'sb!vm' to 'sb-vm' etc in the sources.". the first such change has landed
22:01:08stassatsi guess i need to expose avx_supported into lisp, the SIMD-PACK-256 tests fail on the CI
22:01:23pfdietzRecently tweaked the random tester to produce larger numbers sometimes. I need to make further tweaks, for example in MAKE-RANDOM-TYPE-CONTAINING to make sized SIGNED-BYTE and UNSIGNED-BYTE types.
22:02:04pfdietzThis would have caused the random type prop tests on typep to have found this bug.
22:02:55stassats(sb-alien:extern-alien "avx_supported" boolean) already works
22:03:52stassatsgood enough for tests, but maybe something exposed from lisp
22:04:10stassatslike sb-vm::*cpuid-fn1-ecx*, for faster access
22:09:20stassatsand i want to redo the way we specify instructions
22:09:39stassats(which i always do after adding new instructions)
22:09:56stassats(thinking, nah, how many new instructions are there?)
22:10:58stassatsand thinking, i have better things to do
22:11:37stassatsbut if we're exposing the instructions for people to use (sb-simd, etc), better error checking would be nice
22:11:59stassatsbecause i'm not going to manually check all the errors or the arcane intel inconsistencies