[PATCH] D37329: [X86][AVX512] Add simple tests for all AVX512 shuffle instructions.

[Zvi Rackover via Phabricator via llvm-commits]

zvi added inline comments.


================
Comment at: test/CodeGen/X86/avx512-shuffles/duplicate-low.ll:875
+}
+define <4 x float> @test_masked_4xfloat_dup_low_mem_mask1(<4 x float>* %vp, <4 x float> %vec2) {
+; CHECK-LABEL: test_masked_4xfloat_dup_low_mem_mask1:
----------------
This test doesn't really check any AVX512 support, because it will be folded early in the target-indpenendent dagcombine.


================
Comment at: test/CodeGen/X86/avx512-shuffles/duplicate-low.ll:1002
+}
+define <8 x float> @test_masked_8xfloat_dup_low_mask1(<8 x float> %vec, <8 x float> %vec2) {
+; CHECK-LABEL: test_masked_8xfloat_dup_low_mask1:
----------------
Same here


================
Comment at: test/CodeGen/X86/avx512-shuffles/in_lane_permute.ll:1311
+  %shuf = shufflevector <2 x double> %vec, <2 x double> undef, <2 x i32> <i32 1, i32 0>
+  %res = select <2 x i1> <i1 0, i1 0>, <2 x double> %shuf, <2 x double> %vec2
+  ret <2 x double> %res
----------------
The all-zero/all-one mask tests don't contribute much for AVX512 testing


https://reviews.llvm.org/D37329