Optimizing for neoverse

Ubuntu · Ubuntu · commit 34171110bfda · 2024-06-23T23:46:50.000Z
diff --git a/src/UTF8.cs b/src/UTF8.cs
@@ -1362,8 +1362,9 @@ private unsafe static (int utfadjust, int scalaradjust) calculateErrorPathadjust
                     {
 
                         Vector128<byte> currentBlock = AdvSimd.LoadVector128(pInputBuffer + processedLength);
-
-                        if (AdvSimd.Arm64.MaxAcross(currentBlock).ToScalar() <= 127)
+                        if (AdvSimd.Arm64.MaxAcross(Vector128.AsUInt32(AdvSimd.And(currentBlock, v80))).ToScalar() == 0)
+                        // We could it with (AdvSimd.Arm64.MaxAcross(currentBlock).ToScalar() <= 127) but it is slower on some
+                        // hardware.
                         {
                             // We have an ASCII block, no need to process it, but
                             // we need to check if the previous block was incomplete.

Original file line number	Diff line number	Diff line change
`@@ -1362,8 +1362,9 @@ private unsafe static (int utfadjust, int scalaradjust) calculateErrorPathadjust`
`1362`	`1362`	`{`
`1363`	`1363`
`1364`	`1364`	`Vector128<byte> currentBlock = AdvSimd.LoadVector128(pInputBuffer + processedLength);`
`1365`		`-`
`1366`		`- if (AdvSimd.Arm64.MaxAcross(currentBlock).ToScalar() <= 127)`
	`1365`	`+ if (AdvSimd.Arm64.MaxAcross(Vector128.AsUInt32(AdvSimd.And(currentBlock, v80))).ToScalar() == 0)`
	`1366`	`+ // We could it with (AdvSimd.Arm64.MaxAcross(currentBlock).ToScalar() <= 127) but it is slower on some`
	`1367`	`+ // hardware.`
`1367`	`1368`	`{`
`1368`	`1369`	`// We have an ASCII block, no need to process it, but`
`1369`	`1370`	`// we need to check if the previous block was incomplete.`