[PATCH] D52347: [X86][BtVer2] Fix latency and resource cycles of AVX 256-bit zero-idioms.

[Andrea Di Biagio via Phabricator via llvm-commits]

andreadb added inline comments.


================
Comment at: test/CodeGen/X86/avx-schedule.ll:5508
+; ZNVER1-NEXT:    retq # sched: [1:0.50]
+  call void asm sideeffect "vxorps %ymm0, %ymm0, %ymm0\0Avxorpd %ymm1, %ymm1, %ymm1\0Avandnps %ymm2, %ymm2, %ymm2\0Avandnpd %ymm3, %ymm3, %ymm3", ""()
+  ret void
----------------
lebedev.ri wrote:
> There is no way to achieve this without inline asm?
> When committing, please commit the baseline first.
Not that I know of. In this case, I am pretty sure that there is no other way.

I will add the test in a separate commit.

Thanks for the review!


https://reviews.llvm.org/D52347