[PATCH] D29874: [X86] Generate VZEROUPPER for Skylake-avx512

[Craig Topper via Phabricator via llvm-commits]

craig.topper added inline comments.


================
Comment at: lib/Target/X86/X86.td:241
                        "Use software floating point features.">;
-// On at least some AMD processors, there is no performance hazard to writing
-// only the lower parts of a YMM register without clearing the upper part.
-def FeatureFastPartialYMMWrite
-    : SubtargetFeature<"fast-partial-ymm-write", "HasFastPartialYMMWrite",
-                       "true", "Partial writes to YMM registers are fast">;
+// On Some X86 processors, there is no performance hazard to writing only the
+// lower parts of a YMM or ZMM register without clearing the upper part.
----------------
Minor nit. The word "Some" shouldn't be capitalized.


================
Comment at: lib/Target/X86/X86VZeroUpper.cpp:222
+    // explicitly specified.
+    if (IsCall && !callHasRegMask(MI))
       continue;
----------------
Why did this change to not check specific register clobbers?


================
Comment at: lib/Target/X86/X86VZeroUpper.cpp:225
 
     // The VZEROUPPER instruction resets the upper 128 bits of all AVX
     // registers. In addition, the processor changes back to Clean state, after
----------------
This comments is stale with ZMM. It's not upper 128 bits.


https://reviews.llvm.org/D29874