Skip to content

Commit 3215e5e

Browse files
lemirelemire
committed
Minor optimizations
1 parent a352a15 commit 3215e5e

File tree

1 file changed

+24
-21
lines changed

1 file changed

+24
-21
lines changed

‎src/UTF8.cs

Lines changed: 24 additions & 21 deletions
Original file line numberDiff line numberDiff line change
@@ -68,6 +68,7 @@ private static (int utfAdjust, int scalarAdjust) GetFinalScalarUtfAdjustments(by
6868
// We scan the input from buf to len, possibly going back howFarBack bytes, to find the end of
6969
// a valid UTF-8 sequence. We return buf + len if the buffer is valid, otherwise we return the
7070
// pointer to the first invalid byte.
71+
[MethodImpl(MethodImplOptions.AggressiveInlining)]
7172
private unsafe static byte* SimpleRewindAndValidateWithErrors(int howFarBack, byte* buf, int len)
7273
{
7374
int extraLen = 0;
@@ -90,7 +91,6 @@ private static (int utfAdjust, int scalarAdjust) GetFinalScalarUtfAdjustments(by
9091
{
9192
return buf - howFarBack;
9293
}
93-
9494
int pos = 0;
9595
int nextPos;
9696
uint codePoint = 0;
@@ -598,7 +598,7 @@ private unsafe static (int utfadjust, int scalaradjust) calculateErrorPathadjust
598598
}
599599
else
600600
{
601-
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(processedLength - 3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
601+
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
602602
}
603603
if (invalidBytePointer < pInputBuffer + processedLength)
604604
{
@@ -624,16 +624,17 @@ private unsafe static (int utfadjust, int scalaradjust) calculateErrorPathadjust
624624

625625

626626
// We may still have an error.
627-
if (processedLength < inputLength || !Sse42.TestZ(prevIncomplete, prevIncomplete))
627+
bool hasIncompete = !Sse42.TestZ(prevIncomplete, prevIncomplete);
628+
if (processedLength < inputLength || hasIncompete)
628629
{
629630
byte* invalidBytePointer;
630-
if (processedLength == 0)
631+
if (processedLength == 0 || !hasIncompete)
631632
{
632633
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(0, pInputBuffer + processedLength, inputLength - processedLength);
633634
}
634635
else
635636
{
636-
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(processedLength - 3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
637+
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
637638

638639
}
639640
if (invalidBytePointer != pInputBuffer + inputLength)
@@ -813,7 +814,7 @@ private unsafe static (int utfadjust, int scalaradjust) calculateErrorPathadjust
813814
if (!Avx2.TestZ(prevIncomplete, prevIncomplete))
814815
{
815816
int off = processedLength >= 3 ? processedLength - 3 : processedLength;
816-
byte* invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(16 - 3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
817+
byte* invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(32 - 3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
817818
// So the code is correct up to invalidBytePointer
818819
if (invalidBytePointer < pInputBuffer + processedLength)
819820
{
@@ -877,7 +878,7 @@ private unsafe static (int utfadjust, int scalaradjust) calculateErrorPathadjust
877878
}
878879
else
879880
{
880-
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(processedLength - 3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
881+
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
881882
}
882883
if (invalidBytePointer < pInputBuffer + processedLength)
883884
{
@@ -899,17 +900,17 @@ private unsafe static (int utfadjust, int scalaradjust) calculateErrorPathadjust
899900
}
900901
}
901902
// We may still have an error.
902-
if (processedLength < inputLength || !Avx2.TestZ(prevIncomplete, prevIncomplete))
903+
bool hasIncompete = !Avx2.TestZ(prevIncomplete, prevIncomplete);
904+
if (processedLength < inputLength || hasIncompete)
903905
{
904906
byte* invalidBytePointer;
905-
if (processedLength == 0)
907+
if (processedLength == 0 || !hasIncompete)
906908
{
907909
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(0, pInputBuffer + processedLength, inputLength - processedLength);
908910
}
909911
else
910912
{
911-
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(processedLength - 3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
912-
913+
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
913914
}
914915
if (invalidBytePointer != pInputBuffer + inputLength)
915916
{
@@ -1215,7 +1216,7 @@ private unsafe static (int utfadjust, int scalaradjust) calculateErrorPathadjust
12151216
}
12161217
else
12171218
{
1218-
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(processedLength - 3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
1219+
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
12191220
}
12201221
if (invalidBytePointer < pInputBuffer + processedLength)
12211222
{
@@ -1237,16 +1238,17 @@ private unsafe static (int utfadjust, int scalaradjust) calculateErrorPathadjust
12371238
}
12381239
}
12391240
// We may still have an error.
1240-
if (processedLength < inputLength || Avx512BW.CompareGreaterThan(prevIncomplete, Vector512<byte>.Zero).ExtractMostSignificantBits() != 0)
1241+
bool hasIncompete = Avx512BW.CompareGreaterThan(prevIncomplete, Vector512<byte>.Zero).ExtractMostSignificantBits() != 0;
1242+
if (processedLength < inputLength || hasIncompete)
12411243
{
12421244
byte* invalidBytePointer;
1243-
if (processedLength == 0)
1245+
if (processedLength == 0 || !hasIncompete)
12441246
{
12451247
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(0, pInputBuffer + processedLength, inputLength - processedLength);
12461248
}
12471249
else
12481250
{
1249-
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(processedLength - 3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
1251+
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
12501252

12511253
}
12521254
if (invalidBytePointer != pInputBuffer + inputLength)
@@ -1431,7 +1433,7 @@ private unsafe static (int utfadjust, int scalaradjust) calculateErrorPathadjust
14311433
}
14321434
else
14331435
{
1434-
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(processedLength - 3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
1436+
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
14351437
}
14361438
if (invalidBytePointer < pInputBuffer + processedLength)
14371439
{
@@ -1457,18 +1459,17 @@ private unsafe static (int utfadjust, int scalaradjust) calculateErrorPathadjust
14571459
n4 += negn4add;
14581460
}
14591461
}
1460-
1461-
// We may still have an error.
1462-
if (processedLength < inputLength || AdvSimd.Arm64.MaxAcross(prevIncomplete).ToScalar() != 0)
1462+
bool hasIncompete = AdvSimd.Arm64.MaxAcross(Vector128.AsUInt32(prevIncomplete)).ToScalar() != 0;
1463+
if (processedLength < inputLength || hasIncompete)
14631464
{
14641465
byte* invalidBytePointer;
1465-
if (processedLength == 0)
1466+
if (processedLength == 0 || !hasIncompete)
14661467
{
14671468
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(0, pInputBuffer + processedLength, inputLength - processedLength);
14681469
}
14691470
else
14701471
{
1471-
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(processedLength - 3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
1472+
invalidBytePointer = SimdUnicode.UTF8.SimpleRewindAndValidateWithErrors(3, pInputBuffer + processedLength - 3, inputLength - processedLength + 3);
14721473
}
14731474
if (invalidBytePointer != pInputBuffer + inputLength)
14741475
{
@@ -1497,6 +1498,7 @@ private unsafe static (int utfadjust, int scalaradjust) calculateErrorPathadjust
14971498
return GetPointerToFirstInvalidByteScalar(pInputBuffer + processedLength, inputLength - processedLength, out utf16CodeUnitCountAdjustment, out scalarCountAdjustment);
14981499
}
14991500

1501+
[MethodImpl(MethodImplOptions.AggressiveInlining)]
15001502
private static unsafe void removeCounters(byte* start, byte* end, ref int n4, ref int contbytes)
15011503
{
15021504
for (byte* p = start; p < end; p++)
@@ -1512,6 +1514,7 @@ private static unsafe void removeCounters(byte* start, byte* end, ref int n4, re
15121514
}
15131515
}
15141516

1517+
[MethodImpl(MethodImplOptions.AggressiveInlining)]
15151518
private static unsafe void addCounters(byte* start, byte* end, ref int n4, ref int contbytes)
15161519
{
15171520
for (byte* p = start; p < end; p++)

0 commit comments

Comments
 (0)