Merge pull request #151 from SiaFoundation/nate/vzeroupper

Increase sector root performance
SiaFoundation · Apr 4, 2024 · f424825 · f424825
2 parents 1ae0547 + 69d6da0
commit f424825
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 0 deletions.
diff --git a/internal/blake2b/blake2b_amd64.s b/internal/blake2b/blake2b_amd64.s
@@ -1643,4 +1643,6 @@ TEXT ·hashBlocksAVX2(SB), NOSPLIT, $320-24
 	VMOVDQU     Y1, 32(CX)
 	VMOVDQU     Y2, 64(CX)
 	VMOVDQU     Y3, 96(CX)
+	// Clear the upper YMM registers to avoid performance penalties
+	VZEROUPPER
 	RET
diff --git a/internal/blake2b/gen.go b/internal/blake2b/gen.go
@@ -230,6 +230,8 @@ func genHashBlocksAVX2() {
 		}
 	}
 
+	Comment("Clear the upper YMM registers to avoid performance penalties")
+	VZEROUPPER()
 	RET()
 }