aaru-dps/Aaru.Checksums
1
0
Fork 0
mirror of https://github.com/aaru-dps/Aaru.Checksums.git synced 2025-12-16 19:24:29 +00:00
Code Issues Packages Projects Releases Wiki Activity

General cleanup and refactor.

Browse Source
This commit is contained in:
Natalia Portillo
2022-03-07 07:36:42 +00:00
parent db8c5c6a53
commit 19c7f3d4ff
22 changed files with 1076 additions and 1544 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
CRC32/arm_simd.cs
View File

@@ -45,18 +45,18 @@
// Copyright 2017 The Chromium Authors. All rights reserved.
// ****************************************************************************/
namespace Aaru.Checksums.CRC32;
using System;
using System.Runtime.Intrinsics.Arm;
namespace Aaru.Checksums.CRC32;
internal static class ArmSimd
static class ArmSimd
{
internal static uint Step64(byte[] buf, long len, uint crc)
{
uint c = crc;
int bufPos = 0;
var bufPos = 0;
while(len >= 64)
{
@@ -87,9 +87,7 @@ internal static class ArmSimd
}
while(len-- > 0)
{
c = Crc32.ComputeCrc32(c, buf[bufPos++]);
}
return c;
}
@@ -98,7 +96,7 @@ internal static class ArmSimd
{
uint c = crc;
int bufPos = 0;
var bufPos = 0;
while(len >= 32)
{
@@ -129,9 +127,7 @@ internal static class ArmSimd
}
while(len-- > 0)
{
c = Crc32.ComputeCrc32(c, buf[bufPos++]);
}
return c;
}

43
CRC32/clmul.cs
View File

@@ -47,14 +47,14 @@
// Copyright (C) 2013 Intel Corporation. All rights reserved.
// ****************************************************************************/
namespace Aaru.Checksums.CRC32;
using System;
using System.Runtime.CompilerServices;
using System.Runtime.Intrinsics;
using System.Runtime.Intrinsics.X86;
namespace Aaru.Checksums.CRC32;
internal static class Clmul
static class Clmul
{
static readonly uint[] _crcK =
{
@@ -85,7 +85,7 @@ internal static class Clmul
static void Fold4(ref Vector128<uint> xmmCRC0, ref Vector128<uint> xmmCRC1, ref Vector128<uint> xmmCRC2,
ref Vector128<uint> xmmCRC3)
{
Vector128<uint> xmmFold4 = Vector128.Create(0xc6e41596, 0x00000001, 0x54442bd4, 0x00000001);
var xmmFold4 = Vector128.Create(0xc6e41596, 0x00000001, 0x54442bd4, 0x00000001);
Vector128<uint> xTmp0 = xmmCRC0;
Vector128<uint> xTmp1 = xmmCRC1;
@@ -130,38 +130,34 @@ internal static class Clmul
Vector128<uint> xmmCRC1 = Vector128<uint>.Zero;
Vector128<uint> xmmCRC2 = Vector128<uint>.Zero;
Vector128<uint> xmmCRC3 = Vector128<uint>.Zero;
int bufPos = 0;
var bufPos = 0;
bool first = true;
var first = true;
/* fold 512 to 32 step variable declarations for ISO-C90 compat. */
Vector128<uint> xmmMask = Vector128.Create(0xFFFFFFFF, 0xFFFFFFFF, 0x00000000, 0x00000000);
Vector128<uint> xmmMask2 = Vector128.Create(0x00000000, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
var xmmMask = Vector128.Create(0xFFFFFFFF, 0xFFFFFFFF, 0x00000000, 0x00000000);
var xmmMask2 = Vector128.Create(0x00000000, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
while((len -= 64) >= 0)
{
xmmT0 = Vector128.Create(BitConverter.ToUInt32(src, bufPos), BitConverter.ToUInt32(src, bufPos + 4),
BitConverter.ToUInt32(src, bufPos + 8),
BitConverter.ToUInt32(src, bufPos + 12));
BitConverter.ToUInt32(src, bufPos + 8), BitConverter.ToUInt32(src, bufPos + 12));
bufPos += 16;
xmmT1 = Vector128.Create(BitConverter.ToUInt32(src, bufPos), BitConverter.ToUInt32(src, bufPos + 4),
BitConverter.ToUInt32(src, bufPos + 8),
BitConverter.ToUInt32(src, bufPos + 12));
BitConverter.ToUInt32(src, bufPos + 8), BitConverter.ToUInt32(src, bufPos + 12));
bufPos += 16;
xmmT2 = Vector128.Create(BitConverter.ToUInt32(src, bufPos), BitConverter.ToUInt32(src, bufPos + 4),
BitConverter.ToUInt32(src, bufPos + 8),
BitConverter.ToUInt32(src, bufPos + 12));
BitConverter.ToUInt32(src, bufPos + 8), BitConverter.ToUInt32(src, bufPos + 12));
bufPos += 16;
Vector128<uint> xmmT3 = Vector128.Create(BitConverter.ToUInt32(src, bufPos),
BitConverter.ToUInt32(src, bufPos + 4),
BitConverter.ToUInt32(src, bufPos + 8),
BitConverter.ToUInt32(src, bufPos + 12));
var xmmT3 = Vector128.Create(BitConverter.ToUInt32(src, bufPos), BitConverter.ToUInt32(src, bufPos + 4),
BitConverter.ToUInt32(src, bufPos + 8),
BitConverter.ToUInt32(src, bufPos + 12));
bufPos += 16;
@@ -184,24 +180,21 @@ internal static class Clmul
/*
* k1
*/
Vector128<uint> crcFold = Vector128.Create(_crcK[0], _crcK[1], _crcK[2], _crcK[3]);
var crcFold = Vector128.Create(_crcK[0], _crcK[1], _crcK[2], _crcK[3]);
Vector128<uint> xTmp0 = Pclmulqdq.CarrylessMultiply(xmmCRC0.AsUInt64(), crcFold.AsUInt64(), 0x10).
AsUInt32();
Vector128<uint> xTmp0 = Pclmulqdq.CarrylessMultiply(xmmCRC0.AsUInt64(), crcFold.AsUInt64(), 0x10).AsUInt32();
xmmCRC0 = Pclmulqdq.CarrylessMultiply(xmmCRC0.AsUInt64(), crcFold.AsUInt64(), 0x01).AsUInt32();
xmmCRC1 = Sse2.Xor(xmmCRC1, xTmp0);
xmmCRC1 = Sse2.Xor(xmmCRC1, xmmCRC0);
Vector128<uint> xTmp1 = Pclmulqdq.CarrylessMultiply(xmmCRC1.AsUInt64(), crcFold.AsUInt64(), 0x10).
AsUInt32();
Vector128<uint> xTmp1 = Pclmulqdq.CarrylessMultiply(xmmCRC1.AsUInt64(), crcFold.AsUInt64(), 0x10).AsUInt32();
xmmCRC1 = Pclmulqdq.CarrylessMultiply(xmmCRC1.AsUInt64(), crcFold.AsUInt64(), 0x01).AsUInt32();
xmmCRC2 = Sse2.Xor(xmmCRC2, xTmp1);
xmmCRC2 = Sse2.Xor(xmmCRC2, xmmCRC1);
Vector128<uint> xTmp2 = Pclmulqdq.CarrylessMultiply(xmmCRC2.AsUInt64(), crcFold.AsUInt64(), 0x10).
AsUInt32();
Vector128<uint> xTmp2 = Pclmulqdq.CarrylessMultiply(xmmCRC2.AsUInt64(), crcFold.AsUInt64(), 0x10).AsUInt32();
xmmCRC2 = Pclmulqdq.CarrylessMultiply(xmmCRC2.AsUInt64(), crcFold.AsUInt64(), 0x01).AsUInt32();
xmmCRC3 = Sse2.Xor(xmmCRC3, xTmp2);