public void Initialize(IntPtr pointer, int totalWords, int activeWords)
{
if (totalWords <= 0) throw new InvalidOperationException("A segment cannot be empty");
this.pointer = (ulong*)pointer;
this.totalWords = totalWords;
this.activeWords = activeWords;
}
protected override void Reset(bool recycling)
{
pointer = (ulong*)0;
if (view != null)
{
view.SafeMemoryMappedViewHandle.ReleasePointer();
try { view.Dispose(); }
catch { }
view = null;
}
if (mmFile != null)
{
mmFile.Dispose();
mmFile = null;
}
base.Reset(recycling);
}
private static extern unsafe int proc_pidinfo(
int pid,
int flavor,
ulong arg,
ulong *buffer,
int bufferSize);
internal unsafe void CompressBlock(ulong *hPt, ulong *wPt)
{
for (int i = 16; i < w.Length; i++)
{
wPt[i] = SSIG1(wPt[i - 2]) + wPt[i - 7] + SSIG0(wPt[i - 15]) + wPt[i - 16];
}
ulong a = hPt[0];
ulong b = hPt[1];
ulong c = hPt[2];
ulong d = hPt[3];
ulong e = hPt[4];
ulong f = hPt[5];
ulong g = hPt[6];
ulong h = hPt[7];
ulong temp, aa, bb, cc, dd, ee, ff, hh, gg;
fixed(ulong *kPt = &Ks[0])
{
for (int j = 0; j < 80;)
{
temp = h + BSIG1(e) + CH(e, f, g) + kPt[j] + wPt[j];
ee = d + temp;
aa = temp + BSIG0(a) + MAJ(a, b, c);
j++;
temp = g + BSIG1(ee) + CH(ee, e, f) + kPt[j] + wPt[j];
ff = c + temp;
bb = temp + BSIG0(aa) + MAJ(aa, a, b);
j++;
temp = f + BSIG1(ff) + CH(ff, ee, e) + kPt[j] + wPt[j];
gg = b + temp;
cc = temp + BSIG0(bb) + MAJ(bb, aa, a);
j++;
temp = e + BSIG1(gg) + CH(gg, ff, ee) + kPt[j] + wPt[j];
hh = a + temp;
dd = temp + BSIG0(cc) + MAJ(cc, bb, aa);
j++;
temp = ee + BSIG1(hh) + CH(hh, gg, ff) + kPt[j] + wPt[j];
h = aa + temp;
d = temp + BSIG0(dd) + MAJ(dd, cc, bb);
j++;
temp = ff + BSIG1(h) + CH(h, hh, gg) + kPt[j] + wPt[j];
g = bb + temp;
c = temp + BSIG0(d) + MAJ(d, dd, cc);
j++;
temp = gg + BSIG1(g) + CH(g, h, hh) + kPt[j] + wPt[j];
f = cc + temp;
b = temp + BSIG0(c) + MAJ(c, d, dd);
j++;
temp = hh + BSIG1(f) + CH(f, g, h) + kPt[j] + wPt[j];
e = dd + temp;
a = temp + BSIG0(b) + MAJ(b, c, d);
j++;
}
}
hPt[0] += a;
hPt[1] += b;
hPt[2] += c;
hPt[3] += d;
hPt[4] += e;
hPt[5] += f;
hPt[6] += g;
hPt[7] += h;
}
public abstract unsafe void GetUniform([Flow(FlowDirection.In)] uint program, [Flow(FlowDirection.In)] int location, [Count(Computed = "program, location"), Flow(FlowDirection.Out)] ulong * @params);
public unsafe void CopyTo(Interop.Ole32.IStream pstm, ulong cb, ulong *pcbRead, ulong *pcbWritten)
{
byte[] buffer = ArrayPool <byte> .Shared.Rent(4096);
ulong remaining = cb;
ulong totalWritten = 0;
ulong totalRead = 0;
fixed(byte *b = buffer)
{
while (remaining > 0)
{
uint read = remaining < (ulong)buffer.Length ? (uint)remaining : (uint)buffer.Length;
Read(b, read, &read);
remaining -= read;
totalRead += read;
if (read == 0)
{
break;
}
uint written;
pstm.Write(b, read, &written);
totalWritten += written;
}
}
ArrayPool <byte> .Shared.Return(buffer);
if (pcbRead != null)
{
*pcbRead = totalRead;
}
if (pcbWritten != null)
{
*pcbWritten = totalWritten;
}
}
/// <summary>
/// Resets stream for writing.
/// Uses internal buffer for writing, it's available as DataPointer.
/// </summary>
public void ResetWrite()
{
if (m_ownedBuffer == null)
{
m_ownedBuffer = (ulong*)(void*)SharpDX.Utilities.AllocateClearedMemory(m_defaultByteSize);
m_ownedBufferBitLength = m_defaultByteSize * 8;
}
else
{
SharpDX.Utilities.ClearMemory((IntPtr)(void*)m_ownedBuffer, 0, MyLibraryUtils.GetDivisionCeil(m_ownedBufferBitLength, 8));
}
m_buffer = m_ownedBuffer;
m_bitLength = m_ownedBufferBitLength;
m_bitPosition = 0;
m_writing = true;
}
private void allocMemoryForCurrentSatate(byte[] currentState)
{
freeMemoryForCurrentSatate();
_currentSatate = (ulong*) Marshal.AllocHGlobal(sizeof (ulong));
*_currentSatate = 0;
foreach (var t in currentState)
{
*_currentSatate <<= 8;
*_currentSatate |= t;
}
//*_currentSatate = (ulong)(currentState[0] << 24 | currentState[1] << 16 | currentState[2] << 8 | currentState[3]);
//*_currentSatate = (ulong) (currentState[0] << 40 | currentState[1] << 32 | currentState[2] << 24 | currentState[3] << 16 | currentState[4] << 8 | currentState[5]);
_index0 = (byte*) _currentSatate;
if (CoderSatateSize >= CoderSatateSize.Size2)
_index1 = _index0 + 1;
if (CoderSatateSize >= CoderSatateSize.Size3)
_index2 = _index0 + 2;
if (CoderSatateSize >= CoderSatateSize.Size4)
_index3 = _index0 + 3;
if (CoderSatateSize >= CoderSatateSize.Size5)
_index4 = _index0 + 4;
if (CoderSatateSize >= CoderSatateSize.Size6)
_index5 = _index0 + 5;
if (CoderSatateSize >= CoderSatateSize.Size7)
_index6 = _index0 + 6;
if (CoderSatateSize >= CoderSatateSize.Size8)
_index7 = _index0 + 7;
}
private static void mixAvx2(ulong *sh, ulong *m)
{
// Rotate shuffle masks. We can safely convert the ref to a pointer because the compiler guarantees the
// data is in a fixed location, and the ref itself is converted from a pointer. Same for the IV below.
byte *prm = (byte *)Unsafe.AsPointer(ref MemoryMarshal.GetReference(rormask));
var r24 = Avx2.BroadcastVector128ToVector256(prm);
var r16 = Avx2.BroadcastVector128ToVector256(prm + Vector128 <byte> .Count);
var row1 = Avx.LoadVector256(sh);
var row2 = Avx.LoadVector256(sh + Vector256 <ulong> .Count);
ulong *piv = (ulong *)Unsafe.AsPointer(ref MemoryMarshal.GetReference(ivle));
var row3 = Avx.LoadVector256(piv);
var row4 = Avx.LoadVector256(piv + Vector256 <ulong> .Count);
row4 = Avx2.Xor(row4, Avx.LoadVector256(sh + Vector256 <ulong> .Count * 2)); // t[] and f[]
//ROUND 1
var m0 = Avx2.BroadcastVector128ToVector256(m);
var m1 = Avx2.BroadcastVector128ToVector256(m + Vector128 <ulong> .Count);
var m2 = Avx2.BroadcastVector128ToVector256(m + Vector128 <ulong> .Count * 2);
var m3 = Avx2.BroadcastVector128ToVector256(m + Vector128 <ulong> .Count * 3);
var t0 = Avx2.UnpackLow(m0, m1);
var t1 = Avx2.UnpackLow(m2, m3);
var b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackHigh(m0, m1);
t1 = Avx2.UnpackHigh(m2, m3);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//DIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_10_01_00_11);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_00_11_10_01);
var m4 = Avx2.BroadcastVector128ToVector256(m + Vector128 <ulong> .Count * 4);
var m5 = Avx2.BroadcastVector128ToVector256(m + Vector128 <ulong> .Count * 5);
var m6 = Avx2.BroadcastVector128ToVector256(m + Vector128 <ulong> .Count * 6);
var m7 = Avx2.BroadcastVector128ToVector256(m + Vector128 <ulong> .Count * 7);
t0 = Avx2.UnpackLow(m4, m5);
t1 = Avx2.UnpackLow(m6, m7);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackHigh(m4, m5);
t1 = Avx2.UnpackHigh(m6, m7);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//UNDIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_00_11_10_01);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_10_01_00_11);
//ROUND 2
t0 = Avx2.UnpackLow(m7, m2);
t1 = Avx2.UnpackHigh(m4, m6);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackLow(m5, m4);
t1 = Avx2.AlignRight(m3, m7, 8);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//DIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_10_01_00_11);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_00_11_10_01);
t0 = Avx2.Shuffle(m0.AsUInt32(), 0b_01_00_11_10).AsUInt64();
t1 = Avx2.UnpackHigh(m5, m2);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackLow(m6, m1);
t1 = Avx2.UnpackHigh(m3, m1);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//UNDIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_00_11_10_01);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_10_01_00_11);
//ROUND 3
t0 = Avx2.AlignRight(m6, m5, 8);
t1 = Avx2.UnpackHigh(m2, m7);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackLow(m4, m0);
t1 = Avx2.Blend(m1.AsUInt32(), m6.AsUInt32(), 0b_1100_1100).AsUInt64();
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//DIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_10_01_00_11);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_00_11_10_01);
t0 = Avx2.Blend(m5.AsUInt32(), m1.AsUInt32(), 0b_1100_1100).AsUInt64();
t1 = Avx2.UnpackHigh(m3, m4);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackLow(m7, m3);
t1 = Avx2.AlignRight(m2, m0, 8);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//UNDIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_00_11_10_01);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_10_01_00_11);
//ROUND 4
t0 = Avx2.UnpackHigh(m3, m1);
t1 = Avx2.UnpackHigh(m6, m5);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackHigh(m4, m0);
t1 = Avx2.UnpackLow(m6, m7);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//DIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_10_01_00_11);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_00_11_10_01);
t0 = Avx2.Blend(m1.AsUInt32(), m2.AsUInt32(), 0b_1100_1100).AsUInt64();
t1 = Avx2.Blend(m2.AsUInt32(), m7.AsUInt32(), 0b_1100_1100).AsUInt64();
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackLow(m3, m5);
t1 = Avx2.UnpackLow(m0, m4);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//UNDIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_00_11_10_01);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_10_01_00_11);
//ROUND 5
t0 = Avx2.UnpackHigh(m4, m2);
t1 = Avx2.UnpackLow(m1, m5);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.Blend(m0.AsUInt32(), m3.AsUInt32(), 0b_1100_1100).AsUInt64();
t1 = Avx2.Blend(m2.AsUInt32(), m7.AsUInt32(), 0b_1100_1100).AsUInt64();
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//DIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_10_01_00_11);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_00_11_10_01);
t0 = Avx2.Blend(m7.AsUInt32(), m5.AsUInt32(), 0b_1100_1100).AsUInt64();
t1 = Avx2.Blend(m3.AsUInt32(), m1.AsUInt32(), 0b_1100_1100).AsUInt64();
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.AlignRight(m6, m0, 8);
t1 = Avx2.Blend(m4.AsUInt32(), m6.AsUInt32(), 0b_1100_1100).AsUInt64();
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//UNDIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_00_11_10_01);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_10_01_00_11);
//ROUND 6
t0 = Avx2.UnpackLow(m1, m3);
t1 = Avx2.UnpackLow(m0, m4);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackLow(m6, m5);
t1 = Avx2.UnpackHigh(m5, m1);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//DIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_10_01_00_11);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_00_11_10_01);
t0 = Avx2.Blend(m2.AsUInt32(), m3.AsUInt32(), 0b_1100_1100).AsUInt64();
t1 = Avx2.UnpackHigh(m7, m0);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackHigh(m6, m2);
t1 = Avx2.Blend(m7.AsUInt32(), m4.AsUInt32(), 0b_1100_1100).AsUInt64();
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//UNDIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_00_11_10_01);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_10_01_00_11);
//ROUND 7
t0 = Avx2.Blend(m6.AsUInt32(), m0.AsUInt32(), 0b_1100_1100).AsUInt64();
t1 = Avx2.UnpackLow(m7, m2);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackHigh(m2, m7);
t1 = Avx2.AlignRight(m5, m6, 8);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//DIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_10_01_00_11);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_00_11_10_01);
t0 = Avx2.UnpackLow(m0, m3);
t1 = Avx2.Shuffle(m4.AsUInt32(), 0b_01_00_11_10).AsUInt64();
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackHigh(m3, m1);
t1 = Avx2.Blend(m1.AsUInt32(), m5.AsUInt32(), 0b_1100_1100).AsUInt64();
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//UNDIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_00_11_10_01);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_10_01_00_11);
//ROUND 8
t0 = Avx2.UnpackHigh(m6, m3);
t1 = Avx2.Blend(m6.AsUInt32(), m1.AsUInt32(), 0b_1100_1100).AsUInt64();
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.AlignRight(m7, m5, 8);
t1 = Avx2.UnpackHigh(m0, m4);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//DIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_10_01_00_11);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_00_11_10_01);
t0 = Avx2.UnpackHigh(m2, m7);
t1 = Avx2.UnpackLow(m4, m1);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackLow(m0, m2);
t1 = Avx2.UnpackLow(m3, m5);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//UNDIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_00_11_10_01);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_10_01_00_11);
//ROUND 9
t0 = Avx2.UnpackLow(m3, m7);
t1 = Avx2.AlignRight(m0, m5, 8);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackHigh(m7, m4);
t1 = Avx2.AlignRight(m4, m1, 8);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//DIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_10_01_00_11);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_00_11_10_01);
t0 = m6;
t1 = Avx2.AlignRight(m5, m0, 8);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.Blend(m1.AsUInt32(), m3.AsUInt32(), 0b_1100_1100).AsUInt64();
t1 = m2;
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//UNDIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_00_11_10_01);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_10_01_00_11);
//ROUND 10
t0 = Avx2.UnpackLow(m5, m4);
t1 = Avx2.UnpackHigh(m3, m0);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackLow(m1, m2);
t1 = Avx2.Blend(m3.AsUInt32(), m2.AsUInt32(), 0b_1100_1100).AsUInt64();
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//DIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_10_01_00_11);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_00_11_10_01);
t0 = Avx2.UnpackHigh(m7, m4);
t1 = Avx2.UnpackHigh(m1, m6);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.AlignRight(m7, m5, 8);
t1 = Avx2.UnpackLow(m6, m0);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//UNDIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_00_11_10_01);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_10_01_00_11);
//ROUND 11
t0 = Avx2.UnpackLow(m0, m1);
t1 = Avx2.UnpackLow(m2, m3);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackHigh(m0, m1);
t1 = Avx2.UnpackHigh(m2, m3);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//DIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_10_01_00_11);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_00_11_10_01);
t0 = Avx2.UnpackLow(m4, m5);
t1 = Avx2.UnpackLow(m6, m7);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackHigh(m4, m5);
t1 = Avx2.UnpackHigh(m6, m7);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//UNDIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_00_11_10_01);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_10_01_00_11);
//ROUND 12
t0 = Avx2.UnpackLow(m7, m2);
t1 = Avx2.UnpackHigh(m4, m6);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackLow(m5, m4);
t1 = Avx2.AlignRight(m3, m7, 8);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//DIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_10_01_00_11);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_00_11_10_01);
t0 = Avx2.Shuffle(m0.AsUInt32(), 0b_01_00_11_10).AsUInt64();
t1 = Avx2.UnpackHigh(m5, m2);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G1
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsUInt32(), 0b_10_11_00_01).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Shuffle(row2.AsByte(), r24).AsUInt64();
t0 = Avx2.UnpackLow(m6, m1);
t1 = Avx2.UnpackHigh(m3, m1);
b0 = Avx2.Blend(t0.AsUInt32(), t1.AsUInt32(), 0b_1111_0000).AsUInt64();
//G2
row1 = Avx2.Add(Avx2.Add(row1, b0), row2);
row4 = Avx2.Xor(row4, row1);
row4 = Avx2.Shuffle(row4.AsByte(), r16).AsUInt64();
row3 = Avx2.Add(row3, row4);
row2 = Avx2.Xor(row2, row3);
row2 = Avx2.Xor(Avx2.ShiftRightLogical(row2, 63), Avx2.Add(row2, row2));
//UNDIAGONALIZE
row4 = Avx2.Permute4x64(row4, 0b_00_11_10_01);
row3 = Avx2.Permute4x64(row3, 0b_01_00_11_10);
row2 = Avx2.Permute4x64(row2, 0b_10_01_00_11);
row1 = Avx2.Xor(row1, row3);
row2 = Avx2.Xor(row2, row4);
row1 = Avx2.Xor(row1, Avx.LoadVector256(sh));
row2 = Avx2.Xor(row2, Avx.LoadVector256(sh + Vector256 <ulong> .Count));
Avx.Store(sh, row1);
Avx.Store(sh + Vector256 <ulong> .Count, row2);
}
public int SetEventOnMultipleFenceCompletion([NativeTypeName("ID3D12Fence *const *")] ID3D12Fence **ppFences, [NativeTypeName("const UINT64 *")] ulong *pFenceValues, [NativeTypeName("UINT")] uint NumFences, D3D12_MULTIPLE_FENCE_WAIT_FLAGS Flags, [NativeTypeName("HANDLE")] IntPtr hEvent)
{
return(((delegate * stdcall <ID3D12Device5 *, ID3D12Fence **, ulong *, uint, D3D12_MULTIPLE_FENCE_WAIT_FLAGS, IntPtr, int>)(lpVtbl[45]))((ID3D12Device5 *)Unsafe.AsPointer(ref this), ppFences, pFenceValues, NumFences, Flags, hEvent));
}
public void GetCopyableFootprints([NativeTypeName("const D3D12_RESOURCE_DESC *")] D3D12_RESOURCE_DESC *pResourceDesc, [NativeTypeName("UINT")] uint FirstSubresource, [NativeTypeName("UINT")] uint NumSubresources, [NativeTypeName("UINT64")] ulong BaseOffset, [NativeTypeName("D3D12_PLACED_SUBRESOURCE_FOOTPRINT *")] D3D12_PLACED_SUBRESOURCE_FOOTPRINT *pLayouts, [NativeTypeName("UINT *")] uint *pNumRows, [NativeTypeName("UINT64 *")] ulong *pRowSizeInBytes, [NativeTypeName("UINT64 *")] ulong *pTotalBytes)
{
((delegate * stdcall <ID3D12Device5 *, D3D12_RESOURCE_DESC *, uint, uint, ulong, D3D12_PLACED_SUBRESOURCE_FOOTPRINT *, uint *, ulong *, ulong *, void>)(lpVtbl[38]))((ID3D12Device5 *)Unsafe.AsPointer(ref this), pResourceDesc, FirstSubresource, NumSubresources, BaseOffset, pLayouts, pNumRows, pRowSizeInBytes, pTotalBytes);
}
public int Allocate([NativeTypeName("const VIRTUAL_ALLOCATION_DESC*")] D3D12MA_VIRTUAL_ALLOCATION_DESC *pDesc, [NativeTypeName("UINT64 *")] ulong *pOffset)
{
if ((pDesc == null) || (pOffset == null) || (pDesc->Size == 0) || !IsPow2(pDesc->Alignment))
{
D3D12MA_ASSERT(false); // "Invalid arguments passed to VirtualBlock::Allocate."
return(E_INVALIDARG);
}
*pOffset = UINT64_MAX;
using var debugGlobalMutexLock = D3D12MA_DEBUG_GLOBAL_MUTEX_LOCK();
ulong alignment = pDesc->Alignment != 0 ? pDesc->Alignment : 1;
D3D12MA_AllocationRequest allocRequest = default;
if (m_Metadata.CreateAllocationRequest(pDesc->Size, alignment, &allocRequest))
{
m_Metadata.Alloc(&allocRequest, pDesc->Size, pDesc->pUserData);
D3D12MA_HEAVY_ASSERT((D3D12MA_DEBUG_LEVEL > 1) && m_Metadata.Validate());
*pOffset = allocRequest.offset;
return(S_OK);
}
else
{
return(E_OUTOFMEMORY);
}
}
public override void Reset(bool recycling)
{
pointer = (ulong*)0;
activeWords = totalWords = 0;
base.Reset(recycling);
}
public static void Initialize()
{
Module m = typeof(AntiTamperNormal).Module;
string n = m.FullyQualifiedName;
bool f = n.Length > 0 && n[0] == '<';
var b = (byte *)Marshal.GetHINSTANCE(m);
byte * p = b + *(uint *)(b + 0x3c);
ushort s = *(ushort *)(p + 0x6);
ushort o = *(ushort *)(p + 0x14);
uint *e = null;
uint l = 0;
var r = (uint *)(p + 0x18 + o);
uint z = (uint)Mutation.KeyI1, x = (uint)Mutation.KeyI2, c = (uint)Mutation.KeyI3, v = (uint)Mutation.KeyI4;
for (int i = 0; i < s; i++)
{
uint g = (*r++) * (*r++);
if (g == (uint)Mutation.KeyI0)
{
e = (uint *)(b + (f ? *(r + 3) : *(r + 1)));
l = (f ? *(r + 2) : *(r + 0)) >> 2;
}
else if (g != 0)
{
var q = (uint *)(b + (f ? *(r + 3) : *(r + 1)));
uint j = *(r + 2) >> 2;
for (uint k = 0; k < j; k++)
{
uint t = (z ^ (*q++)) + x + c * v;
z = x;
x = c;
x = v;
v = t;
}
}
r += 8;
}
uint[] y = new uint[0x10], d = new uint[0x10];
for (int i = 0; i < 0x10; i++)
{
y[i] = v;
d[i] = x;
z = (x >> 5) | (x << 27);
x = (c >> 3) | (c << 29);
c = (v >> 7) | (v << 25);
v = (z >> 11) | (z << 21);
}
Mutation.Crypt(y, d);
uint h = 0;
uint *u = e;
VirtualProtect((IntPtr)e, l << 2, 0x40, out z);
for (uint i = 0; i < l; i++)
{
*e ^= y[h & 0xf];
y[h & 0xf] = (y[h & 0xf] ^ (*e++)) + 0x3dbb2819;
h++;
}
ptr = u + 4;
len = *ptr++;
ver4 = Environment.Version.Major == 4;
ModuleHandle hnd = m.ModuleHandle;
if (ver4)
{
ulong *str = stackalloc ulong[1];
str[0] = 0x0061746144705f6d; //m_pData.
moduleHnd = (IntPtr)m.GetType().GetField(new string((sbyte *)str), BindingFlags.NonPublic | BindingFlags.Instance).GetValue(m);
ver5 = Environment.Version.Revision > 17020;
}
else
{
//moduleHnd = *(IntPtr*)(&hnd);
Marshal.PtrToStructure(moduleHnd, hnd);
}
Hook();
}
internal static void MoveNotNull(/*Type[0]*/ ulong /*Type[0]*/[] array, int index, int writeIndex, int count)
{
#if MONO
int endIndex = index + count;
if (index < writeIndex && endIndex > writeIndex)
{
fixed(/*Type[0]*/ ulong /*Type[0]*/ *valueFixed = array)
{
for (/*Type[0]*/ ulong /*Type[0]*/ *write = valueFixed + writeIndex + count, start = valueFixed + index, end = valueFixed + endIndex;
end != start;
*--write = *--end)
{
;
}
}
}
else
{
Array.Copy(array, index, array, writeIndex, count);
}
#else
fixed(/*Type[0]*/ ulong /*Type[0]*/ *valueFixed = array) AutoCSer.Win32.Kernel32.RtlMoveMemory(valueFixed + writeIndex, valueFixed + index, count * sizeof(/*Type[0]*/ ulong /*Type[0]*/));
#endif
}
public int EndDraw([NativeTypeName("D2D1_TAG *")] ulong *tag1 = null, [NativeTypeName("D2D1_TAG *")] ulong *tag2 = null)
{
return(((delegate * stdcall <ID2D1DCRenderTarget *, ulong *, ulong *, int>)(lpVtbl[49]))((ID2D1DCRenderTarget *)Unsafe.AsPointer(ref this), tag1, tag2));
}
/// <summary>
/// 逆转数组
/// </summary>
/// <param name="array">数组数据</param>
/// <param name="index">起始位置</param>
/// <param name="length">翻转数据数量</param>
/// <returns>翻转后的新数组</returns>
internal static /*Type[0]*/ ulong /*Type[0]*/[] GetReverse(/*Type[0]*/ ulong /*Type[0]*/[] array, int index, int length)
{
/*Type[0]*/
ulong /*Type[0]*/[] newValues = new /*Type[0]*/ ulong /*Type[0]*/[length];
fixed(/*Type[0]*/ ulong /*Type[0]*/ *valueFixed = array, newValueFixed = newValues)
{
for (/*Type[0]*/ ulong /*Type[0]*/ *start = valueFixed + index, end = start + length, wirte = newValueFixed + length; start != end; *--wirte = *start++)
{
;
}
}
return(newValues);
}
/// <summary>
/// Resets stream for reading (reads what was written so far).
/// </summary>
public void ResetRead()
{
m_bitLength = m_bitPosition;
m_buffer = m_ownedBuffer;
m_writing = false;
m_bitPosition = 0;
}
public int AddPage([NativeTypeName("ID2D1CommandList *")] ID2D1CommandList *commandList, D2D_SIZE_F pageSize, [NativeTypeName("IStream *")] IStream *pagePrintTicketStream, [NativeTypeName("D2D1_TAG *")] ulong *tag1 = null, [NativeTypeName("D2D1_TAG *")] ulong *tag2 = null)
{
return(((delegate * stdcall <ID2D1PrintControl *, ID2D1CommandList *, D2D_SIZE_F, IStream *, ulong *, ulong *, int>)(lpVtbl[3]))((ID2D1PrintControl *)Unsafe.AsPointer(ref this), commandList, pageSize, pagePrintTicketStream, tag1, tag2));
}
void Resize(int bitSize)
{
if (!OwnsBuffer)
throw new BitStreamException("BitStream cannot write more data. Buffer is full and it's not owned by BitStream", new System.IO.EndOfStreamException());
// Always at least double the size
int newBitSize = Math.Max(m_bitLength * 2, bitSize);
int newByteLen = MyLibraryUtils.GetDivisionCeil(newBitSize, 64) * 8;
var newBuffer = SharpDX.Utilities.AllocateClearedMemory(newByteLen);
SharpDX.Utilities.CopyMemory(newBuffer, (IntPtr)(void*)m_buffer, BytePosition);
SharpDX.Utilities.FreeMemory((IntPtr)(void*)m_buffer);
m_buffer = (ulong*)(void*)newBuffer;
m_bitLength = newBitSize;
m_ownedBuffer = m_buffer;
m_ownedBufferBitLength = m_bitLength;
}
public HRESULT get_domainLookupStart([NativeTypeName("ULONGLONG *")] ulong *p)
{
return(((delegate * unmanaged <IHTMLPerformanceTiming *, ulong *, int>)(lpVtbl[13]))((IHTMLPerformanceTiming *)Unsafe.AsPointer(ref this), p));
}
protected void FreePointer()
{
if (pointer != (ulong*)0) Marshal.FreeHGlobal(new IntPtr(pointer));
pointer = (ulong*)0;
}
public HRESULT get_unloadEventEnd([NativeTypeName("ULONGLONG *")] ulong *p)
{
return(((delegate * unmanaged <IHTMLPerformanceTiming *, ulong *, int>)(lpVtbl[9]))((IHTMLPerformanceTiming *)Unsafe.AsPointer(ref this), p));
}
private static unsafe partial int proc_pidinfo(
int pid,
int flavor,
ulong arg,
ulong *buffer,
int bufferSize);
public override unsafe void RenderGlyphs(RenderingTextureAllocator textureAllocator, List <RenderingFont.GlyphRenderInfo> glyphRenderInfos, List <RectangleInt2> overlays)
{
Vector2 posScaling = new Vector2(1.0f) / (Size / 2); // Divide the integer coordinates to avoid pixel mishmash.
Vector2 posOffset = VectorInt2.FromRounded(posScaling * 0.5f);
Vector2 textureScaling = new Vector2(16777216.0f) / new Vector2(textureAllocator.Size.X, textureAllocator.Size.Y);
// Build vertex buffer
int vertexCount = glyphRenderInfos.Count * 6 + overlays.Count * 6;
int bufferSize = vertexCount * (sizeof(Vector2) + sizeof(ulong));
fixed(byte *data = new byte[bufferSize])
{
Vector2 *positions = (Vector2 *)(data);
ulong * uvws = (ulong *)(data + vertexCount * sizeof(Vector2));
// Setup vertices
int c = 0;
for (int i = 0; i < overlays.Count; ++i, ++c)
{
var overlay = overlays[i];
Vector2 posStart = overlay.Start * posScaling + posOffset;
Vector2 posEnd = (overlay.End - new Vector2(1)) * posScaling + posOffset;
positions[c * 6 + 0] = new Vector2(posStart.X, posStart.Y);
positions[c * 6 + 2] = positions[c * 6 + 3] = new Vector2(posEnd.X, posStart.Y);
positions[c * 6 + 1] = positions[c * 6 + 4] = new Vector2(posStart.X, posEnd.Y);
positions[c * 6 + 5] = new Vector2(posEnd.X, posEnd.Y);
uvws[c * 6 + 2] = uvws[c * 6 + 3] =
uvws[c * 6 + 1] = uvws[c * 6 + 4] =
uvws[c * 6 + 5] = uvws[c * 6 + 0] = Dx11RenderingDevice.CompressUvw(VectorInt3.Zero, Vector2.Zero, Vector2.Zero, 1);
}
for (int i = 0; i < glyphRenderInfos.Count; ++i, ++c)
{
RenderingFont.GlyphRenderInfo info = glyphRenderInfos[i];
Vector2 posStart = info.PosStart * posScaling + posOffset;
Vector2 posEnd = (info.PosEnd - new Vector2(1)) * posScaling + posOffset;
positions[c * 6 + 0] = new Vector2(posStart.X, posStart.Y);
positions[c * 6 + 2] = positions[c * 6 + 3] = new Vector2(posEnd.X, posStart.Y);
positions[c * 6 + 1] = positions[c * 6 + 4] = new Vector2(posStart.X, posEnd.Y);
positions[c * 6 + 5] = new Vector2(posEnd.X, posEnd.Y);
uvws[c * 6 + 0] = Dx11RenderingDevice.CompressUvw(info.TexStart, textureScaling, new Vector2(0.5f, 0.5f));
uvws[c * 6 + 2] = uvws[c * 6 + 3] = Dx11RenderingDevice.CompressUvw(info.TexStart, textureScaling, new Vector2(info.TexSize.X - 0.5f, 0.5f));
uvws[c * 6 + 1] = uvws[c * 6 + 4] = Dx11RenderingDevice.CompressUvw(info.TexStart, textureScaling, new Vector2(0.5f, info.TexSize.Y - 0.5f));
uvws[c * 6 + 5] = Dx11RenderingDevice.CompressUvw(info.TexStart, textureScaling, new Vector2(info.TexSize.X - 0.5f, info.TexSize.Y - 0.5f));
}
// Create GPU resources
using (var VertexBuffer = new Buffer(Device.Device, new IntPtr(data),
new BufferDescription(bufferSize, ResourceUsage.Immutable, BindFlags.VertexBuffer,
CpuAccessFlags.None, ResourceOptionFlags.None, 0)))
{
var VertexBufferBindings = new VertexBufferBinding[] {
new VertexBufferBinding(VertexBuffer, sizeof(Vector2), (int)((byte *)positions - data)),
new VertexBufferBinding(VertexBuffer, sizeof(ulong), (int)((byte *)uvws - data))
};
// Render
Bind();
Device.TextShader.Apply(Device.Context);
Device.Context.PixelShader.SetSampler(0, Device.SamplerDefault);
Device.Context.PixelShader.SetShaderResources(0, ((Dx11RenderingTextureAllocator)(textureAllocator)).TextureView);
Device.Context.InputAssembler.SetVertexBuffers(0, VertexBufferBindings);
Device.Context.OutputMerger.SetDepthStencilState(Device.DepthStencilNoZBuffer);
// Render
Device.Context.Draw(vertexCount, 0);
// Reset state
Device.Context.OutputMerger.SetDepthStencilState(Device.DepthStencilDefault);
}
}
}
public abstract unsafe void ProgramUniform2([Flow(FlowDirection.In)] uint program, [Flow(FlowDirection.In)] int location, [Flow(FlowDirection.In)] uint count, [Count(Parameter = "count"), Flow(FlowDirection.In)] ulong *value);
public static bool mul_overflow(ulong value1, ulong value2, ulong *result)
{
throw new NotImplementedException(":("); // same here - _umul128
}
/// <summary>
/// unsigned __int64 _mulx_u64 (unsigned __int64 a, unsigned __int64 b, unsigned __int64* hi)
/// MULX r64a, r64b, reg/m64
/// The above native signature does not directly correspond to the managed signature.
/// This intrinsic is only available on 64-bit processes
/// </summary>
public static unsafe ulong MultiplyNoFlags(ulong left, ulong right, ulong *low)
{
throw new PlatformNotSupportedException();
}
public int GetFrequency([NativeTypeName("UINT64 *")] ulong *pu64Frequency)
{
return(((delegate * stdcall <IAudioClock *, ulong *, int>)(lpVtbl[3]))((IAudioClock *)Unsafe.AsPointer(ref this), pu64Frequency));
}
public int GetBuffer([NativeTypeName("BYTE **")] byte **ppData, [NativeTypeName("UINT32 *")] uint *pNumFramesToRead, [NativeTypeName("DWORD *")] uint *pdwFlags, [NativeTypeName("UINT64 *")] ulong *pu64DevicePosition, [NativeTypeName("UINT64 *")] ulong *pu64QPCPosition)
{
return(((delegate * stdcall <IAudioCaptureClient *, byte **, uint *, uint *, ulong *, ulong *, int>)(lpVtbl[3]))((IAudioCaptureClient *)Unsafe.AsPointer(ref this), ppData, pNumFramesToRead, pdwFlags, pu64DevicePosition, pu64QPCPosition));
}
public int GetPosition([NativeTypeName("UINT64 *")] ulong *pu64Position, [NativeTypeName("UINT64 *")] ulong *pu64QPCPosition)
{
return(((delegate * stdcall <IAudioClock *, ulong *, ulong *, int>)(lpVtbl[4]))((IAudioClock *)Unsafe.AsPointer(ref this), pu64Position, pu64QPCPosition));
}
public void GetTags([NativeTypeName("D2D1_TAG *")] ulong *tag1 = null, [NativeTypeName("D2D1_TAG *")] ulong *tag2 = null)
{
((delegate * stdcall <ID2D1DCRenderTarget *, ulong *, ulong *, void>)(lpVtbl[39]))((ID2D1DCRenderTarget *)Unsafe.AsPointer(ref this), tag1, tag2);
}
public HRESULT get_WindowSize([NativeTypeName("DWORDLONG *")] ulong *pWindowSize)
{
return(((delegate * unmanaged <IAMVideoCompression *, ulong *, int>)(lpVtbl[10]))((IAMVideoCompression *)Unsafe.AsPointer(ref this), pWindowSize));
}
/// <summary>
/// Creates a new CudaRegisteredHostMemory_ulong from an existing IntPtr. IntPtr must be page size aligned (4KBytes)!
/// </summary>
/// <param name="hostPointer">must be page size aligned (4KBytes)</param>
/// <param name="size">In elements</param>
public CudaRegisteredHostMemory_ulong(IntPtr hostPointer, SizeT size)
{
_intPtr = hostPointer;
_size = size;
_typeSize = (SizeT)Marshal.SizeOf(typeof(ulong));
_ptr = (ulong*)_intPtr;
}
private static extern bool WinUsb_GetDescriptor(IntPtr InterfaceHandle, byte DescriptorType, byte Index, ushort LanguageID, out byte *Buffer, ulong BufferLength, out ulong *LengthTransferred);
public void ReleaseInternalBuffer()
{
if (m_ownedBuffer != null)
{
if (m_buffer == m_ownedBuffer)
{
m_buffer = null;
m_bitLength = 0;
}
SharpDX.Utilities.FreeMemory((IntPtr)(void*)m_ownedBuffer);
m_ownedBuffer = null;
m_ownedBufferBitLength = 0;
}
}
private static extern bool WinUsb_GetInterfacePowerPolicy(IntPtr InterfaceHandle, ulong PolicyType, out ulong *ValueLength, out void *Value); // raro
/// <summary>
/// Resets stream for reading.
/// </summary>
public void ResetRead(IntPtr buffer, int bitLength, bool copy)
{
if (copy)
{
int byteLen = MyLibraryUtils.GetDivisionCeil(bitLength, 8);
int allocByteSize = Math.Max(byteLen, m_defaultByteSize);
if (m_ownedBuffer == null)
{
m_ownedBuffer = (ulong*)(void*)SharpDX.Utilities.AllocateClearedMemory(allocByteSize);
m_ownedBufferBitLength = allocByteSize * 8;
}
else if (m_ownedBufferBitLength < bitLength)
{
SharpDX.Utilities.FreeMemory((IntPtr)(void*)m_ownedBuffer);
m_ownedBuffer = (ulong*)(void*)SharpDX.Utilities.AllocateClearedMemory(allocByteSize);
m_ownedBufferBitLength = allocByteSize * 8;
}
SharpDX.Utilities.CopyMemory((IntPtr)(void*)m_ownedBuffer, buffer, byteLen);
m_buffer = m_ownedBuffer;
m_bitLength = bitLength;
m_bitPosition = 0;
m_writing = false;
}
else
{
m_buffer = (ulong*)(void*)buffer;
m_bitLength = bitLength;
m_bitPosition = 0;
m_writing = false;
}
}
private static extern bool WinUsb_GetPipePolicy(IntPtr DeviceHandle, byte PipeID, ulong PolicyType, out ulong *ValueLength, out void *Value); // raroo
public void ResetWrite(BitStream stream)
{
int bitLength = stream.m_writing ? stream.m_bitPosition : stream.BitLength;
if (m_ownedBuffer != null && m_ownedBufferBitLength < bitLength)
{
SharpDX.Utilities.FreeMemory((IntPtr)(void*)m_ownedBuffer);
m_ownedBuffer = null;
}
if (m_ownedBuffer == null)
{
int byteSize = Math.Max(MyLibraryUtils.GetDivisionCeil(bitLength, 64) * 8, m_defaultByteSize);
m_ownedBuffer = (ulong*)(void*)SharpDX.Utilities.AllocateClearedMemory(byteSize);
m_ownedBufferBitLength = byteSize * 8;
}
m_buffer = m_ownedBuffer;
m_bitLength = m_ownedBufferBitLength;
SharpDX.Utilities.CopyMemory(DataPointer, stream.DataPointer, MyLibraryUtils.GetDivisionCeil(bitLength, 8));
m_bitPosition = bitLength;
m_writing = true;
}
public HRESULT GetUINT64([NativeTypeName("const GUID &")] Guid *guidKey, [NativeTypeName("UINT64 *")] ulong *punValue)
{
return(((delegate * unmanaged <IMFSample *, Guid *, ulong *, int>)(lpVtbl[8]))((IMFSample *)Unsafe.AsPointer(ref this), guidKey, punValue));
}
private void Init(MemoryMappedFile file, long offsetBytes, long lengthBytes, MemoryMappedFileAccess access, int defaultSegmentWords, bool leaveOpen = false)
{
base.Init(defaultSegmentWords);
MemoryMappedViewAccessor view = null;
byte* ptr = (byte*)0;
try
{
if(lengthBytes < 0) lengthBytes = 0;
view = file.CreateViewAccessor(offsetBytes, lengthBytes, access);
lengthBytes = view.Capacity;
view.SafeMemoryMappedViewHandle.AcquirePointer(ref ptr);
this.pointer = (ulong*)&ptr[offsetBytes];
this.view = view;
if (!leaveOpen) this.mmFile = file;
view = null;
this.totalWords = lengthBytes >> 3;
}
finally
{
if (view != null)
{
if (ptr != (byte*)0) view.SafeMemoryMappedViewHandle.ReleasePointer();
view.Dispose();
}
}
}
public unsafe void Seek(long dlibMove, SeekOrigin dwOrigin, ulong *plibNewPosition)
{
long position = _virtualPosition;
if (_virtualPosition == -1)
{
position = _dataStream.Position;
}
long length = _dataStream.Length;
switch (dwOrigin)
{
case SeekOrigin.Begin:
if (dlibMove <= length)
{
_dataStream.Position = dlibMove;
_virtualPosition = -1;
}
else
{
_virtualPosition = dlibMove;
}
break;
case SeekOrigin.End:
if (dlibMove <= 0)
{
_dataStream.Position = length + dlibMove;
_virtualPosition = -1;
}
else
{
_virtualPosition = length + dlibMove;
}
break;
case SeekOrigin.Current:
if (dlibMove + position <= length)
{
_dataStream.Position = position + dlibMove;
_virtualPosition = -1;
}
else
{
_virtualPosition = dlibMove + position;
}
break;
}
if (plibNewPosition == null)
{
return;
}
if (_virtualPosition != -1)
{
*plibNewPosition = (ulong)_virtualPosition;
}
else
{
*plibNewPosition = (ulong)_dataStream.Position;
}
}
/// <summary>
/// erstellt einen neuen Hash-Eintrag (darf noch nicht vorhanden sein)
/// </summary>
/// <param name="code">Hash-Code, welcher eintragen werden soll</param>
/// <param name="tiefe">entsprechende Zugtiefe</param>
public void Add(ulong code, int tiefe)
{
#if multiHash
var hash = hashes[code & hashesBits];
#endif
hash.Add(code, (ushort)tiefe);
if (hash.Count > dictionaryLimit)
{
#if Index0
Array.Resize(ref weitere, weitere.Length + 1);
for (int i = weitere.Length - 1; i > 0; i--) weitere[i] = weitere[i - 1];
weitere[0] = hash;
hash = new Dictionary<ulong, ushort>();
#endif
#if Index16
if (archivAnzahl == 0)
{
#region # // --- Archiv das erste mal erstellen ---
// --- Archiv vorbereiten ---
#if multiHash
archivAnzahl = hashes.Sum(x => x.Count);
#else
archivAnzahl = hash.Count;
#endif
archivDataPointer = Marshal.AllocHGlobal((IntPtr)((long)archivAnzahl * 8L));
if (archivDataPointer == IntPtr.Zero) throw new Exception("Speicher konnte nicht reserviert werden (" + (((long)archivAnzahl * 8L) / 1048576L) + " MB)");
archivData = (ulong*)archivDataPointer.ToPointer();
uint[] zähler = new uint[1 << 16];
#if multiHash
foreach (var h in hashes)
{
h.Select(x => zähler[x.Key & 0xffff]++).Count();
}
#else
hash.Select(x => zähler[x.Key & 0xffff]++).Count();
#endif
uint[] posis = new uint[1 << 16];
uint pos = 0;
for (int i = 1; i < (1 << 16); i++)
{
pos += zähler[i - 1];
posis[i] = pos;
}
// --- Archiv befüllen ---
#if multiHash
foreach (var h in hashes) foreach (var satz in h)
#else
foreach (var satz in hash)
#endif
{
int indexPos = (int)(satz.Key & 0xffff);
archivData[posis[indexPos]++] = (satz.Key & 0xffffffffffff0000) | (ulong)satz.Value;
}
// --- Positionen neu berechnen ---
posis[0] = pos = 0;
for (int i = 1; i < (1 << 16); i++)
{
pos += zähler[i - 1];
posis[i] = pos;
}
// --- Archiv sortieren (Multi-Threaded) ---
ParallelEnumerable.Range(0, 1 << 16).Select(i =>
{
Sort.Quick(&archivData[posis[i]], (int)zähler[i]);
return i;
}).Count();
// --- 16 Bit-Index erstellen ---
uint[] indexTemp = Enumerable.Range(0, (1 << 16) * 2).Select(i => (i & 1) == 0 ? zähler[i >> 1] : posis[i >> 1]).ToArray();
for (int i = 0; i < indexTemp.Length; i++) archivIndex[i] = indexTemp[i];
#endregion
}
else
{
#region # // --- Archiv erweitern ---
// --- Neue Daten vorbereiten ---
int dazuAnzahl = hash.Count;
ulong[] dazuData = new ulong[dazuAnzahl];
uint[] zähler = new uint[1 << 16];
hash.Select(x => zähler[x.Key & 0xffff]++).Count();
uint[] posis = new uint[1 << 16];
uint pos = 0;
for (int i = 1; i < (1 << 16); i++)
{
pos += zähler[i - 1];
posis[i] = pos;
}
// --- Neue Daten befüllen ---
foreach (var satz in hash)
{
int indexPos = (int)(satz.Key & 0xffff);
dazuData[posis[indexPos]++] = (satz.Key & 0xffffffffffff0000) | (ulong)satz.Value;
}
// --- Positionen neu berechnen ---
posis[0] = pos = 0;
for (int i = 1; i < (1 << 16); i++)
{
pos += zähler[i - 1];
posis[i] = pos;
}
// --- Neue Daten sortieren (Multi-Threaded) ---
ParallelEnumerable.Range(0, (1 << 16)).Select(i =>
{
Array.Sort(dazuData, (int)posis[i], (int)zähler[i]);
return i;
}).Count();
// --- Neues Array erstellen ---
ulong* altData = archivData;
IntPtr altPointer = archivDataPointer;
archivDataPointer = Marshal.AllocHGlobal((IntPtr)((long)(archivAnzahl + dazuAnzahl) * 8L));
if (archivDataPointer == IntPtr.Zero) throw new Exception("Speicher konnte nicht reserviert werden (" + (((long)(archivAnzahl + dazuAnzahl) * 8L) / 1048576L) + " MB)");
archivData = (ulong*)archivDataPointer.ToPointer();
// --- alte Archiv-Daten kopieren und zusammen mit den neuen Daten verschmelzen ---
pos = 0;
for (int index = 0; index < (1 << 16); index++)
{
uint posAlt = archivIndex[(index << 1) + 1];
uint countAlt = archivIndex[index << 1];
uint posNeu = posis[index];
uint countNeu = zähler[index];
uint dazu = 0;
while (countAlt > 0 && countNeu > 0) // beide Tabellen verzahnen und ins neue Archiv speichern
{
if (altData[posAlt] < dazuData[posNeu])
{
archivData[pos] = altData[posAlt++];
countAlt--;
}
else
{
archivData[pos] = dazuData[posNeu++];
countNeu--;
}
pos++;
dazu++;
}
while (countAlt > 0) // Reste der alten Tabelle übertragen
{
archivData[pos++] = altData[posAlt++];
dazu++;
countAlt--;
}
while (countNeu > 0) // Reste der neuen Tabelle übertragen
{
archivData[pos++] = dazuData[posNeu++];
dazu++;
countNeu--;
}
// Index anpassen
archivIndex[index << 1] = dazu;
archivIndex[(index << 1) + 1] = pos - dazu;
}
archivAnzahl = pos;
Marshal.FreeHGlobal(altPointer);
dazuData = null;
#endregion
}
#endif
#if Index24
if (archivAnzahl == 0)
{
#region # // --- Archiv das erste mal erstellen ---
// --- Archiv vorbereiten ---
#if multiHash
archivAnzahl = hashes.Sum(x => (long)x.Count);
#else
archivAnzahl = hash.Count;
#endif
archivDataPointer = Marshal.AllocHGlobal((IntPtr)((long)archivAnzahl * 7L + 1L));
if (archivDataPointer == IntPtr.Zero) throw new Exception("Speicher konnte nicht reserviert werden (" + (((long)archivAnzahl * 7L + 1L) / 1048576L) + " MB)");
archivData = (long)archivDataPointer.ToPointer();
#if DEBUG
byte* zerofill = (byte*)archivData;
for (int i = 0; i < archivAnzahl * 7 + 1; i++) zerofill[i] = 0x00;
#endif
uint[] zähler = new uint[1 << 24];
#if multiHash
foreach (var h in hashes)
{
h.Select(x => zähler[x.Key & 0xffffff]++).Count();
}
#else
hash.Select(x => zähler[x.Key & 0xffffff]++).Count();
#endif
uint[] posis = new uint[1 << 24];
uint pos = 0;
for (int i = 1; i < (1 << 24); i++)
{
pos += zähler[i - 1];
posis[i] = pos;
}
// --- Archiv befüllen ---
#if multiHash
foreach (var h in hashes) foreach (var satz in h)
#else
foreach (var satz in hash)
#endif
{
int indexPos = (int)(satz.Key & 0xffffff);
ulong* p = (ulong*)(archivData + (long)(posis[indexPos]++) * 7L);
*p = (*p & 0xff00000000000000) | ((satz.Key >> 8) & 0x00ffffffffff0000) | (ulong)satz.Value;
}
// --- Positionen neu berechnen ---
posis[0] = pos = 0;
for (int i = 1; i < (1 << 24); i++)
{
pos += zähler[i - 1];
posis[i] = pos;
}
// --- Archiv sortieren (Multi-Threaded) ---
ParallelEnumerable.Range(0, 1 << 24).Select(i =>
{
ulong[] tmp = new ulong[zähler[i]];
long offset = (long)posis[i] * 7L + archivData;
for (int x = 0; x < tmp.Length; x++)
{
tmp[x] = *(ulong*)(offset + (long)(x * 7)) & 0x00ffffffffffffff;
}
Sort.Quick(tmp);
for (int x = 0; x < tmp.Length; x++)
{
ulong w = *(ulong*)(offset + (long)(x * 7)) & 0xff00000000000000;
*(ulong*)(offset + (long)(x * 7)) = w | tmp[x];
}
return i;
}).Count();
// --- 24 Bit-Index erstellen ---
uint[] indexTemp = Enumerable.Range(0, (1 << 24) * 2).Select(i => (i & 1) == 0 ? zähler[i >> 1] : posis[i >> 1]).ToArray();
for (int i = 0; i < indexTemp.Length; i++) archivIndex[i] = indexTemp[i];
#endregion
}
else
{
#region # // --- Archiv erweitern ---
// --- Neue Daten vorbereiten ---
int dazuAnzahl = hash.Count;
ulong[] dazuData = new ulong[dazuAnzahl];
uint[] zähler = new uint[1 << 24];
hash.Select(x => zähler[x.Key & 0xffffff]++).Count();
uint[] posis = new uint[1 << 24];
uint pos = 0;
for (int i = 1; i < (1 << 24); i++)
{
pos += zähler[i - 1];
posis[i] = pos;
}
// --- Neue Daten befüllen ---
foreach (var satz in hash)
{
int indexPos = (int)(satz.Key & 0xffffff);
dazuData[posis[indexPos]++] = ((satz.Key >> 8) & 0x00ffffffffff0000) | (ulong)satz.Value;
}
// --- Positionen neu berechnen ---
posis[0] = pos = 0;
for (int i = 1; i < (1 << 24); i++)
{
pos += zähler[i - 1];
posis[i] = pos;
}
// --- Neue Daten sortieren (Multi-Threaded) ---
ParallelEnumerable.Range(0, (1 << 24)).Select(i =>
{
Array.Sort(dazuData, (int)posis[i], (int)zähler[i]);
return i;
}).Count();
// --- Neues Array erstellen ---
long altData = archivData;
IntPtr altPointer = archivDataPointer;
archivDataPointer = Marshal.AllocHGlobal((IntPtr)((long)(archivAnzahl + dazuAnzahl) * 7L + 1L));
if (archivDataPointer == IntPtr.Zero) throw new Exception("Speicher konnte nicht reserviert werden (" + (((long)(archivAnzahl + dazuAnzahl) * 7L + 1L) / 1048576L) + " MB)");
archivData = (long)archivDataPointer.ToPointer();
// --- alte Archiv-Daten kopieren und zusammen mit den neuen Daten verschmelzen ---
pos = 0;
for (int index = 0; index < (1 << 24); index++)
{
uint posAlt = archivIndex[(index << 1) + 1];
uint countAlt = archivIndex[index << 1];
uint posNeu = posis[index];
uint countNeu = zähler[index];
uint dazu = 0;
while (countAlt > 0 && countNeu > 0) // beide Tabellen verzahnen und ins neue Archiv speichern
{
if (((*(ulong*)(altData + (long)posAlt * 7L)) & 0x00ffffffffffffff) < dazuData[posNeu])
{
*(ulong*)(archivData + (long)pos * 7L) = ((*(ulong*)(archivData + (long)pos * 7L)) & 0xff00000000000000) | ((*(ulong*)(altData + (long)posAlt * 7L)) & 0x00ffffffffffffff);
posAlt++;
countAlt--;
}
else
{
*(ulong*)(archivData + (long)pos * 7L) = ((*(ulong*)(archivData + (long)pos * 7L)) & 0xff00000000000000) | dazuData[posNeu++];
countNeu--;
}
pos++;
dazu++;
}
while (countAlt > 0) // Reste der alten Tabelle übertragen
{
*(ulong*)(archivData + (long)pos * 7L) = ((*(ulong*)(archivData + (long)pos * 7L)) & 0xff00000000000000) | ((*(ulong*)(altData + (long)posAlt * 7L)) & 0x00ffffffffffffff);
pos++;
posAlt++;
dazu++;
countAlt--;
}
while (countNeu > 0) // Reste der neuen Tabelle übertragen
{
*(ulong*)(archivData + (long)pos * 7L) = ((*(ulong*)(archivData + (long)pos * 7L)) & 0xff00000000000000) | dazuData[posNeu++];
pos++;
dazu++;
countNeu--;
}
// Index anpassen
archivIndex[index << 1] = dazu;
archivIndex[(index << 1) + 1] = pos - dazu;
}
archivAnzahl = pos;
Marshal.FreeHGlobal(altPointer);
dazuData = null;
#endregion
}
#endif
#if (Index16 || Index24)
#if multiHash
for (int i = 0; i < hashes.Length; i++)
{
hashes[i] = null; // Hash aufräumen
}
hashes = new[] { new Dictionary<ulong, ushort>() };
hashesBits = 0;
#else
hash = new Dictionary<ulong, ushort>(); // Hash aufräumen
#endif
GC.Collect();
#endif
}
}
public void Reset(IntPtr data, int bitLength)
{
m_buffer = (ulong*)(void*)data;
m_bitLength = bitLength;
BitPosition = 0;
}