summary refs log tree commit diff stats
path: root/results/scraper/fex/3796
diff options
context:
space:
mode:
authorChristian Krinitsin <mail@krinitsin.com>2025-07-17 09:10:43 +0200
committerChristian Krinitsin <mail@krinitsin.com>2025-07-17 09:10:43 +0200
commitf2ec263023649e596c5076df32c2d328bc9393d2 (patch)
tree5dd86caab46e552bd2e62bf9c4fb1a7504a44db4 /results/scraper/fex/3796
parent63d2e9d409831aa8582787234cae4741847504b7 (diff)
downloadqemu-analysis-main.tar.gz
qemu-analysis-main.zip
add downloaded fex bug-reports HEAD main
Diffstat (limited to 'results/scraper/fex/3796')
-rw-r--r--results/scraper/fex/379611
1 files changed, 11 insertions, 0 deletions
diff --git a/results/scraper/fex/3796 b/results/scraper/fex/3796
new file mode 100644
index 000000000..906a26818
--- /dev/null
+++ b/results/scraper/fex/3796
@@ -0,0 +1,11 @@
+AVX128: Blends with immediate selector completely unoptimized
+This includes the whole family of blends:

+16-bit: VPBLENDW

+32-bit: VPBLENDD, VBLENDPS

+64-bit: VBLENDPD

+

+64-bit blend is definitely the easiest where we should be able to move or zip depending.

+

+32-bit and 16-bit get a bit more spicy.

+

+These are all currently implemented with just a loop of inserts in to a zero register, so they are quite low performance atm.
\ No newline at end of file