[PATCH] [X86] Also create v2f32 FMIN/FMAX nodes from fcmp/select.

Quentin Colombet qcolombet at apple.com
Thu Jan 8 11:49:35 PST 2015


Hi Ahmed,

LGTM with maybe more verbose tests.

Also, shouldn’t we fix some vector cost model since you mentioned the vectorizer shouldn’t have kicked in some benchmarks?
File free to file a PR for that (or to fix it in a following commit :)).

Thanks,
-Quentin


================
Comment at: test/CodeGen/X86/sse-minmax.ll:942
@@ +941,3 @@
+; UNSAFE-LABEL: test_maxps_illegal_v2f32:
+; UNSAFE: maxps
+define <2 x float> @test_maxps_illegal_v2f32(<2 x float> %x, <2 x float> %y) {
----------------
I know the surrounding tests are already written like this, but I would like something more verbose that check we just generate this one instruction (+ret).

http://reviews.llvm.org/D6557

EMAIL PREFERENCES
  http://reviews.llvm.org/settings/panel/emailpreferences/






More information about the llvm-commits mailing list