diff options
author | gdkchan <gab.dark.100@gmail.com> | 2020-07-12 00:07:01 -0300 |
---|---|---|
committer | GitHub <noreply@github.com> | 2020-07-12 05:07:01 +0200 |
commit | 4d02a2d2c0451b4de1f6de3bbce54c457cacebe2 (patch) | |
tree | 120fe4fb8cfa1ac1c6ef4c97d92be47e955e8c0f /Ryujinx.Graphics.Nvdec.Vp9/Dsp/InvTxfm.cs | |
parent | 38b26cf4242999fa7d8c550993ac0940cd03d55f (diff) |
New NVDEC and VIC implementation (#1384)
* Initial NVDEC and VIC implementation
* Update FFmpeg.AutoGen to 4.3.0
* Add nvdec dependencies for Windows
* Unify some VP9 structures
* Rename VP9 structure fields
* Improvements to Video API
* XML docs for Common.Memory
* Remove now unused or redundant overloads from MemoryAccessor
* NVDEC UV surface read/write scalar paths
* Add FIXME comments about hacky things/stuff that will need to be fixed in the future
* Cleaned up VP9 memory allocation
* Remove some debug logs
* Rename some VP9 structs
* Remove unused struct
* No need to compile Ryujinx.Graphics.Host1x with unsafe anymore
* Name AsyncWorkQueue threads to make debugging easier
* Make Vp9PictureInfo a ref struct
* LayoutConverter no longer needs the depth argument (broken by rebase)
* Pooling of VP9 buffers, plus fix a memory leak on VP9
* Really wish VS could rename projects properly...
* Address feedback
* Remove using
* Catch OperationCanceledException
* Add licensing informations
* Add THIRDPARTY.md to release too
Co-authored-by: Thog <me@thog.eu>
Diffstat (limited to 'Ryujinx.Graphics.Nvdec.Vp9/Dsp/InvTxfm.cs')
-rw-r--r-- | Ryujinx.Graphics.Nvdec.Vp9/Dsp/InvTxfm.cs | 2868 |
1 files changed, 2868 insertions, 0 deletions
diff --git a/Ryujinx.Graphics.Nvdec.Vp9/Dsp/InvTxfm.cs b/Ryujinx.Graphics.Nvdec.Vp9/Dsp/InvTxfm.cs new file mode 100644 index 00000000..b4ad4344 --- /dev/null +++ b/Ryujinx.Graphics.Nvdec.Vp9/Dsp/InvTxfm.cs @@ -0,0 +1,2868 @@ +using System; +using System.Diagnostics; +using System.Runtime.CompilerServices; +using Ryujinx.Graphics.Nvdec.Vp9.Common; +using static Ryujinx.Graphics.Nvdec.Vp9.Dsp.TxfmCommon; + +namespace Ryujinx.Graphics.Nvdec.Vp9.Dsp +{ + internal static class InvTxfm + { + // 12 signal input bits + 7 2D forward transform amplify bits + 5 1D inverse + // transform amplify bits + 1 bit for contingency in rounding and quantizing + private const int HighbdValidTxfmMagnitudeRange = (1 << 25); + + [MethodImpl(MethodImplOptions.AggressiveInlining)] + private static int DetectInvalidHighbdInput(ReadOnlySpan<int> input, int size) + { + int i; + for (i = 0; i < size; ++i) + { + if (Math.Abs(input[i]) >= HighbdValidTxfmMagnitudeRange) + { + return 1; + } + } + + return 0; + } + + [MethodImpl(MethodImplOptions.AggressiveInlining)] + private static long CheckRange(long input) + { + // For valid VP9 input streams, intermediate stage coefficients should always + // stay within the range of a signed 16 bit integer. Coefficients can go out + // of this range for invalid/corrupt VP9 streams. + Debug.Assert(short.MinValue <= input); + Debug.Assert(input <= short.MaxValue); + return input; + } + + [MethodImpl(MethodImplOptions.AggressiveInlining)] + public static long HighbdCheckRange(long input, int bd) + { + // For valid highbitdepth VP9 streams, intermediate stage coefficients will + // stay within the ranges: + // - 8 bit: signed 16 bit integer + // - 10 bit: signed 18 bit integer + // - 12 bit: signed 20 bit integer + int intMax = (1 << (7 + bd)) - 1; + int intMin = -intMax - 1; + Debug.Assert(intMin <= input); + Debug.Assert(input <= intMax); + + return input; + } + + [MethodImpl(MethodImplOptions.AggressiveInlining)] + private static int WrapLow(long x) + { + return (short)CheckRange(x); + } + + [MethodImpl(MethodImplOptions.AggressiveInlining)] + private static int HighbdWrapLow(long x, int bd) + { + return ((int)HighbdCheckRange(x, bd) << (24 - bd)) >> (24 - bd); + } + + [MethodImpl(MethodImplOptions.AggressiveInlining)] + public static byte ClipPixelAdd(byte dest, long trans) + { + trans = WrapLow(trans); + return BitUtils.ClipPixel(dest + (int)trans); + } + + [MethodImpl(MethodImplOptions.AggressiveInlining)] + public static ushort HighbdClipPixelAdd(ushort dest, long trans, int bd) + { + trans = HighbdWrapLow(trans, bd); + return BitUtils.ClipPixelHighbd(dest + (int)trans, bd); + } + + [MethodImpl(MethodImplOptions.AggressiveInlining)] + private static long DctConstRoundShift(long input) + { + long rv = BitUtils.RoundPowerOfTwo(input, DctConstBits); + return rv; + } + + public static void Iwht4x416Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + /* 4-point reversible, orthonormal inverse Walsh-Hadamard in 3.5 adds, + 0.5 shifts per pixel. */ + int i; + Span<int> output = stackalloc int[16]; + long a1, b1, c1, d1, e1; + ReadOnlySpan<int> ip = input; + Span<int> op = output; + + for (i = 0; i < 4; i++) + { + a1 = ip[0] >> UnitQuantShift; + c1 = ip[1] >> UnitQuantShift; + d1 = ip[2] >> UnitQuantShift; + b1 = ip[3] >> UnitQuantShift; + a1 += c1; + d1 -= b1; + e1 = (a1 - d1) >> 1; + b1 = e1 - b1; + c1 = e1 - c1; + a1 -= b1; + d1 += c1; + op[0] = WrapLow(a1); + op[1] = WrapLow(b1); + op[2] = WrapLow(c1); + op[3] = WrapLow(d1); + ip = ip.Slice(4); + op = op.Slice(4); + } + + Span<int> ip2 = output; + for (i = 0; i < 4; i++) + { + a1 = ip2[4 * 0]; + c1 = ip2[4 * 1]; + d1 = ip2[4 * 2]; + b1 = ip2[4 * 3]; + a1 += c1; + d1 -= b1; + e1 = (a1 - d1) >> 1; + b1 = e1 - b1; + c1 = e1 - c1; + a1 -= b1; + d1 += c1; + dest[stride * 0] = ClipPixelAdd(dest[stride * 0], WrapLow(a1)); + dest[stride * 1] = ClipPixelAdd(dest[stride * 1], WrapLow(b1)); + dest[stride * 2] = ClipPixelAdd(dest[stride * 2], WrapLow(c1)); + dest[stride * 3] = ClipPixelAdd(dest[stride * 3], WrapLow(d1)); + + ip2 = ip2.Slice(1); + dest = dest.Slice(1); + } + } + + public static void Iwht4x41Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i; + long a1, e1; + Span<int> tmp = stackalloc int[4]; + ReadOnlySpan<int> ip = input; + Span<int> op = tmp; + + a1 = ip[0] >> UnitQuantShift; + e1 = a1 >> 1; + a1 -= e1; + op[0] = WrapLow(a1); + op[1] = op[2] = op[3] = WrapLow(e1); + + Span<int> ip2 = tmp; + for (i = 0; i < 4; i++) + { + e1 = ip2[0] >> 1; + a1 = ip2[0] - e1; + dest[stride * 0] = ClipPixelAdd(dest[stride * 0], a1); + dest[stride * 1] = ClipPixelAdd(dest[stride * 1], e1); + dest[stride * 2] = ClipPixelAdd(dest[stride * 2], e1); + dest[stride * 3] = ClipPixelAdd(dest[stride * 3], e1); + ip2 = ip2.Slice(1); + dest = dest.Slice(1); + } + } + + public static void Iadst4(ReadOnlySpan<int> input, Span<int> output) + { + long s0, s1, s2, s3, s4, s5, s6, s7; + int x0 = input[0]; + int x1 = input[1]; + int x2 = input[2]; + int x3 = input[3]; + + if ((x0 | x1 | x2 | x3) == 0) + { + output.Slice(0, 4).Fill(0); + return; + } + + // 32-bit result is enough for the following multiplications. + s0 = SinPi1_9 * x0; + s1 = SinPi2_9 * x0; + s2 = SinPi3_9 * x1; + s3 = SinPi4_9 * x2; + s4 = SinPi1_9 * x2; + s5 = SinPi2_9 * x3; + s6 = SinPi4_9 * x3; + s7 = WrapLow(x0 - x2 + x3); + + s0 = s0 + s3 + s5; + s1 = s1 - s4 - s6; + s3 = s2; + s2 = SinPi3_9 * s7; + + // 1-D transform scaling factor is sqrt(2). + // The overall dynamic range is 14b (input) + 14b (multiplication scaling) + // + 1b (addition) = 29b. + // Hence the output bit depth is 15b. + output[0] = WrapLow(DctConstRoundShift(s0 + s3)); + output[1] = WrapLow(DctConstRoundShift(s1 + s3)); + output[2] = WrapLow(DctConstRoundShift(s2)); + output[3] = WrapLow(DctConstRoundShift(s0 + s1 - s3)); + } + + public static void Idct4(ReadOnlySpan<int> input, Span<int> output) + { + Span<short> step = stackalloc short[4]; + long temp1, temp2; + + // stage 1 + temp1 = ((short)input[0] + (short)input[2]) * CosPi16_64; + temp2 = ((short)input[0] - (short)input[2]) * CosPi16_64; + step[0] = (short)WrapLow(DctConstRoundShift(temp1)); + step[1] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = (short)input[1] * CosPi24_64 - (short)input[3] * CosPi8_64; + temp2 = (short)input[1] * CosPi8_64 + (short)input[3] * CosPi24_64; + step[2] = (short)WrapLow(DctConstRoundShift(temp1)); + step[3] = (short)WrapLow(DctConstRoundShift(temp2)); + + // stage 2 + output[0] = WrapLow(step[0] + step[3]); + output[1] = WrapLow(step[1] + step[2]); + output[2] = WrapLow(step[1] - step[2]); + output[3] = WrapLow(step[0] - step[3]); + } + + public static void Idct4x416Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i, j; + Span<int> output = stackalloc int[4 * 4]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[4]; + Span<int> tempOut = stackalloc int[4]; + + // Rows + for (i = 0; i < 4; ++i) + { + Idct4(input, outptr); + input = input.Slice(4); + outptr = outptr.Slice(4); + } + + // Columns + for (i = 0; i < 4; ++i) + { + for (j = 0; j < 4; ++j) + { + tempIn[j] = output[j * 4 + i]; + } + + Idct4(tempIn, tempOut); + for (j = 0; j < 4; ++j) + { + dest[j * stride + i] = ClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 4)); + } + } + } + + public static void Idct4x41Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i; + long a1; + int output = WrapLow(DctConstRoundShift((short)input[0] * CosPi16_64)); + + output = WrapLow(DctConstRoundShift(output * CosPi16_64)); + a1 = BitUtils.RoundPowerOfTwo(output, 4); + + for (i = 0; i < 4; i++) + { + dest[0] = ClipPixelAdd(dest[0], a1); + dest[1] = ClipPixelAdd(dest[1], a1); + dest[2] = ClipPixelAdd(dest[2], a1); + dest[3] = ClipPixelAdd(dest[3], a1); + dest = dest.Slice(stride); + } + } + + public static void Iadst8(ReadOnlySpan<int> input, Span<int> output) + { + int s0, s1, s2, s3, s4, s5, s6, s7; + long x0 = input[7]; + long x1 = input[0]; + long x2 = input[5]; + long x3 = input[2]; + long x4 = input[3]; + long x5 = input[4]; + long x6 = input[1]; + long x7 = input[6]; + + if ((x0 | x1 | x2 | x3 | x4 | x5 | x6 | x7) == 0) + { + output.Slice(0, 8).Fill(0); + return; + } + + // stage 1 + s0 = (int)(CosPi2_64 * x0 + CosPi30_64 * x1); + s1 = (int)(CosPi30_64 * x0 - CosPi2_64 * x1); + s2 = (int)(CosPi10_64 * x2 + CosPi22_64 * x3); + s3 = (int)(CosPi22_64 * x2 - CosPi10_64 * x3); + s4 = (int)(CosPi18_64 * x4 + CosPi14_64 * x5); + s5 = (int)(CosPi14_64 * x4 - CosPi18_64 * x5); + s6 = (int)(CosPi26_64 * x6 + CosPi6_64 * x7); + s7 = (int)(CosPi6_64 * x6 - CosPi26_64 * x7); + + x0 = WrapLow(DctConstRoundShift(s0 + s4)); + x1 = WrapLow(DctConstRoundShift(s1 + s5)); + x2 = WrapLow(DctConstRoundShift(s2 + s6)); + x3 = WrapLow(DctConstRoundShift(s3 + s7)); + x4 = WrapLow(DctConstRoundShift(s0 - s4)); + x5 = WrapLow(DctConstRoundShift(s1 - s5)); + x6 = WrapLow(DctConstRoundShift(s2 - s6)); + x7 = WrapLow(DctConstRoundShift(s3 - s7)); + + // stage 2 + s0 = (int)x0; + s1 = (int)x1; + s2 = (int)x2; + s3 = (int)x3; + s4 = (int)(CosPi8_64 * x4 + CosPi24_64 * x5); + s5 = (int)(CosPi24_64 * x4 - CosPi8_64 * x5); + s6 = (int)(-CosPi24_64 * x6 + CosPi8_64 * x7); + s7 = (int)(CosPi8_64 * x6 + CosPi24_64 * x7); + + x0 = WrapLow(s0 + s2); + x1 = WrapLow(s1 + s3); + x2 = WrapLow(s0 - s2); + x3 = WrapLow(s1 - s3); + x4 = WrapLow(DctConstRoundShift(s4 + s6)); + x5 = WrapLow(DctConstRoundShift(s5 + s7)); + x6 = WrapLow(DctConstRoundShift(s4 - s6)); + x7 = WrapLow(DctConstRoundShift(s5 - s7)); + + // stage 3 + s2 = (int)(CosPi16_64 * (x2 + x3)); + s3 = (int)(CosPi16_64 * (x2 - x3)); + s6 = (int)(CosPi16_64 * (x6 + x7)); + s7 = (int)(CosPi16_64 * (x6 - x7)); + + x2 = WrapLow(DctConstRoundShift(s2)); + x3 = WrapLow(DctConstRoundShift(s3)); + x6 = WrapLow(DctConstRoundShift(s6)); + x7 = WrapLow(DctConstRoundShift(s7)); + + output[0] = WrapLow(x0); + output[1] = WrapLow(-x4); + output[2] = WrapLow(x6); + output[3] = WrapLow(-x2); + output[4] = WrapLow(x3); + output[5] = WrapLow(-x7); + output[6] = WrapLow(x5); + output[7] = WrapLow(-x1); + } + + public static void Idct8(ReadOnlySpan<int> input, Span<int> output) + { + Span<short> step1 = stackalloc short[8]; + Span<short> step2 = stackalloc short[8]; + long temp1, temp2; + + // stage 1 + step1[0] = (short)input[0]; + step1[2] = (short)input[4]; + step1[1] = (short)input[2]; + step1[3] = (short)input[6]; + temp1 = (short)input[1] * CosPi28_64 - (short)input[7] * CosPi4_64; + temp2 = (short)input[1] * CosPi4_64 + (short)input[7] * CosPi28_64; + step1[4] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[7] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = (short)input[5] * CosPi12_64 - (short)input[3] * CosPi20_64; + temp2 = (short)input[5] * CosPi20_64 + (short)input[3] * CosPi12_64; + step1[5] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[6] = (short)WrapLow(DctConstRoundShift(temp2)); + + // stage 2 + temp1 = (step1[0] + step1[2]) * CosPi16_64; + temp2 = (step1[0] - step1[2]) * CosPi16_64; + step2[0] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[1] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = step1[1] * CosPi24_64 - step1[3] * CosPi8_64; + temp2 = step1[1] * CosPi8_64 + step1[3] * CosPi24_64; + step2[2] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[3] = (short)WrapLow(DctConstRoundShift(temp2)); + step2[4] = (short)WrapLow(step1[4] + step1[5]); + step2[5] = (short)WrapLow(step1[4] - step1[5]); + step2[6] = (short)WrapLow(-step1[6] + step1[7]); + step2[7] = (short)WrapLow(step1[6] + step1[7]); + + // stage 3 + step1[0] = (short)WrapLow(step2[0] + step2[3]); + step1[1] = (short)WrapLow(step2[1] + step2[2]); + step1[2] = (short)WrapLow(step2[1] - step2[2]); + step1[3] = (short)WrapLow(step2[0] - step2[3]); + step1[4] = step2[4]; + temp1 = (step2[6] - step2[5]) * CosPi16_64; + temp2 = (step2[5] + step2[6]) * CosPi16_64; + step1[5] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[6] = (short)WrapLow(DctConstRoundShift(temp2)); + step1[7] = step2[7]; + + // stage 4 + output[0] = WrapLow(step1[0] + step1[7]); + output[1] = WrapLow(step1[1] + step1[6]); + output[2] = WrapLow(step1[2] + step1[5]); + output[3] = WrapLow(step1[3] + step1[4]); + output[4] = WrapLow(step1[3] - step1[4]); + output[5] = WrapLow(step1[2] - step1[5]); + output[6] = WrapLow(step1[1] - step1[6]); + output[7] = WrapLow(step1[0] - step1[7]); + } + + public static void Idct8x864Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i, j; + Span<int> output = stackalloc int[8 * 8]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[8]; + Span<int> tempOut = stackalloc int[8]; + + // First transform rows + for (i = 0; i < 8; ++i) + { + Idct8(input, outptr); + input = input.Slice(8); + outptr = outptr.Slice(8); + } + + // Then transform columns + for (i = 0; i < 8; ++i) + { + for (j = 0; j < 8; ++j) + { + tempIn[j] = output[j * 8 + i]; + } + + Idct8(tempIn, tempOut); + for (j = 0; j < 8; ++j) + { + dest[j * stride + i] = ClipPixelAdd(dest[j * stride + i], + BitUtils.RoundPowerOfTwo(tempOut[j], 5)); + } + } + } + + public static void Idct8x812Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i, j; + Span<int> output = stackalloc int[8 * 8]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[8]; + Span<int> tempOut = stackalloc int[8]; + + // First transform rows + // Only first 4 row has non-zero coefs + for (i = 0; i < 4; ++i) + { + Idct8(input, outptr); + input = input.Slice(8); + outptr = outptr.Slice(8); + } + + // Then transform columns + for (i = 0; i < 8; ++i) + { + for (j = 0; j < 8; ++j) + { + tempIn[j] = output[j * 8 + i]; + } + + Idct8(tempIn, tempOut); + for (j = 0; j < 8; ++j) + { + dest[j * stride + i] = ClipPixelAdd(dest[j * stride + i], + BitUtils.RoundPowerOfTwo(tempOut[j], 5)); + } + } + } + + public static void Idct8x81Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i, j; + long a1; + int output = WrapLow(DctConstRoundShift((short)input[0] * CosPi16_64)); + + output = WrapLow(DctConstRoundShift(output * CosPi16_64)); + a1 = BitUtils.RoundPowerOfTwo(output, 5); + for (j = 0; j < 8; ++j) + { + for (i = 0; i < 8; ++i) + { + dest[i] = ClipPixelAdd(dest[i], a1); + } + + dest = dest.Slice(stride); + } + } + + public static void Iadst16(ReadOnlySpan<int> input, Span<int> output) + { + long s0, s1, s2, s3, s4, s5, s6, s7, s8; + long s9, s10, s11, s12, s13, s14, s15; + long x0 = input[15]; + long x1 = input[0]; + long x2 = input[13]; + long x3 = input[2]; + long x4 = input[11]; + long x5 = input[4]; + long x6 = input[9]; + long x7 = input[6]; + long x8 = input[7]; + long x9 = input[8]; + long x10 = input[5]; + long x11 = input[10]; + long x12 = input[3]; + long x13 = input[12]; + long x14 = input[1]; + long x15 = input[14]; + + if ((x0 | x1 | x2 | x3 | x4 | x5 | x6 | x7 | x8 | x9 | x10 | x11 | x12 | x13 | x14 | x15) == 0) + { + output.Slice(0, 16).Fill(0); + return; + } + + // stage 1 + s0 = x0 * CosPi1_64 + x1 * CosPi31_64; + s1 = x0 * CosPi31_64 - x1 * CosPi1_64; + s2 = x2 * CosPi5_64 + x3 * CosPi27_64; + s3 = x2 * CosPi27_64 - x3 * CosPi5_64; + s4 = x4 * CosPi9_64 + x5 * CosPi23_64; + s5 = x4 * CosPi23_64 - x5 * CosPi9_64; + s6 = x6 * CosPi13_64 + x7 * CosPi19_64; + s7 = x6 * CosPi19_64 - x7 * CosPi13_64; + s8 = x8 * CosPi17_64 + x9 * CosPi15_64; + s9 = x8 * CosPi15_64 - x9 * CosPi17_64; + s10 = x10 * CosPi21_64 + x11 * CosPi11_64; + s11 = x10 * CosPi11_64 - x11 * CosPi21_64; + s12 = x12 * CosPi25_64 + x13 * CosPi7_64; + s13 = x12 * CosPi7_64 - x13 * CosPi25_64; + s14 = x14 * CosPi29_64 + x15 * CosPi3_64; + s15 = x14 * CosPi3_64 - x15 * CosPi29_64; + + x0 = WrapLow(DctConstRoundShift(s0 + s8)); + x1 = WrapLow(DctConstRoundShift(s1 + s9)); + x2 = WrapLow(DctConstRoundShift(s2 + s10)); + x3 = WrapLow(DctConstRoundShift(s3 + s11)); + x4 = WrapLow(DctConstRoundShift(s4 + s12)); + x5 = WrapLow(DctConstRoundShift(s5 + s13)); + x6 = WrapLow(DctConstRoundShift(s6 + s14)); + x7 = WrapLow(DctConstRoundShift(s7 + s15)); + x8 = WrapLow(DctConstRoundShift(s0 - s8)); + x9 = WrapLow(DctConstRoundShift(s1 - s9)); + x10 = WrapLow(DctConstRoundShift(s2 - s10)); + x11 = WrapLow(DctConstRoundShift(s3 - s11)); + x12 = WrapLow(DctConstRoundShift(s4 - s12)); + x13 = WrapLow(DctConstRoundShift(s5 - s13)); + x14 = WrapLow(DctConstRoundShift(s6 - s14)); + x15 = WrapLow(DctConstRoundShift(s7 - s15)); + + // stage 2 + s0 = x0; + s1 = x1; + s2 = x2; + s3 = x3; + s4 = x4; + s5 = x5; + s6 = x6; + s7 = x7; + s8 = x8 * CosPi4_64 + x9 * CosPi28_64; + s9 = x8 * CosPi28_64 - x9 * CosPi4_64; + s10 = x10 * CosPi20_64 + x11 * CosPi12_64; + s11 = x10 * CosPi12_64 - x11 * CosPi20_64; + s12 = -x12 * CosPi28_64 + x13 * CosPi4_64; + s13 = x12 * CosPi4_64 + x13 * CosPi28_64; + s14 = -x14 * CosPi12_64 + x15 * CosPi20_64; + s15 = x14 * CosPi20_64 + x15 * CosPi12_64; + + x0 = WrapLow(s0 + s4); + x1 = WrapLow(s1 + s5); + x2 = WrapLow(s2 + s6); + x3 = WrapLow(s3 + s7); + x4 = WrapLow(s0 - s4); + x5 = WrapLow(s1 - s5); + x6 = WrapLow(s2 - s6); + x7 = WrapLow(s3 - s7); + x8 = WrapLow(DctConstRoundShift(s8 + s12)); + x9 = WrapLow(DctConstRoundShift(s9 + s13)); + x10 = WrapLow(DctConstRoundShift(s10 + s14)); + x11 = WrapLow(DctConstRoundShift(s11 + s15)); + x12 = WrapLow(DctConstRoundShift(s8 - s12)); + x13 = WrapLow(DctConstRoundShift(s9 - s13)); + x14 = WrapLow(DctConstRoundShift(s10 - s14)); + x15 = WrapLow(DctConstRoundShift(s11 - s15)); + + // stage 3 + s0 = x0; + s1 = x1; + s2 = x2; + s3 = x3; + s4 = x4 * CosPi8_64 + x5 * CosPi24_64; + s5 = x4 * CosPi24_64 - x5 * CosPi8_64; + s6 = -x6 * CosPi24_64 + x7 * CosPi8_64; + s7 = x6 * CosPi8_64 + x7 * CosPi24_64; + s8 = x8; + s9 = x9; + s10 = x10; + s11 = x11; + s12 = x12 * CosPi8_64 + x13 * CosPi24_64; + s13 = x12 * CosPi24_64 - x13 * CosPi8_64; + s14 = -x14 * CosPi24_64 + x15 * CosPi8_64; + s15 = x14 * CosPi8_64 + x15 * CosPi24_64; + + x0 = WrapLow(s0 + s2); + x1 = WrapLow(s1 + s3); + x2 = WrapLow(s0 - s2); + x3 = WrapLow(s1 - s3); + x4 = WrapLow(DctConstRoundShift(s4 + s6)); + x5 = WrapLow(DctConstRoundShift(s5 + s7)); + x6 = WrapLow(DctConstRoundShift(s4 - s6)); + x7 = WrapLow(DctConstRoundShift(s5 - s7)); + x8 = WrapLow(s8 + s10); + x9 = WrapLow(s9 + s11); + x10 = WrapLow(s8 - s10); + x11 = WrapLow(s9 - s11); + x12 = WrapLow(DctConstRoundShift(s12 + s14)); + x13 = WrapLow(DctConstRoundShift(s13 + s15)); + x14 = WrapLow(DctConstRoundShift(s12 - s14)); + x15 = WrapLow(DctConstRoundShift(s13 - s15)); + + // stage 4 + s2 = (-CosPi16_64) * (x2 + x3); + s3 = CosPi16_64 * (x2 - x3); + s6 = CosPi16_64 * (x6 + x7); + s7 = CosPi16_64 * (-x6 + x7); + s10 = CosPi16_64 * (x10 + x11); + s11 = CosPi16_64 * (-x10 + x11); + s14 = (-CosPi16_64) * (x14 + x15); + s15 = CosPi16_64 * (x14 - x15); + + x2 = WrapLow(DctConstRoundShift(s2)); + x3 = WrapLow(DctConstRoundShift(s3)); + x6 = WrapLow(DctConstRoundShift(s6)); + x7 = WrapLow(DctConstRoundShift(s7)); + x10 = WrapLow(DctConstRoundShift(s10)); + x11 = WrapLow(DctConstRoundShift(s11)); + x14 = WrapLow(DctConstRoundShift(s14)); + x15 = WrapLow(DctConstRoundShift(s15)); + + output[0] = WrapLow(x0); + output[1] = WrapLow(-x8); + output[2] = WrapLow(x12); + output[3] = WrapLow(-x4); + output[4] = WrapLow(x6); + output[5] = WrapLow(x14); + output[6] = WrapLow(x10); + output[7] = WrapLow(x2); + output[8] = WrapLow(x3); + output[9] = WrapLow(x11); + output[10] = WrapLow(x15); + output[11] = WrapLow(x7); + output[12] = WrapLow(x5); + output[13] = WrapLow(-x13); + output[14] = WrapLow(x9); + output[15] = WrapLow(-x1); + } + + public static void Idct16(ReadOnlySpan<int> input, Span<int> output) + { + Span<short> step1 = stackalloc short[16]; + Span<short> step2 = stackalloc short[16]; + long temp1, temp2; + + // stage 1 + step1[0] = (short)input[0 / 2]; + step1[1] = (short)input[16 / 2]; + step1[2] = (short)input[8 / 2]; + step1[3] = (short)input[24 / 2]; + step1[4] = (short)input[4 / 2]; + step1[5] = (short)input[20 / 2]; + step1[6] = (short)input[12 / 2]; + step1[7] = (short)input[28 / 2]; + step1[8] = (short)input[2 / 2]; + step1[9] = (short)input[18 / 2]; + step1[10] = (short)input[10 / 2]; + step1[11] = (short)input[26 / 2]; + step1[12] = (short)input[6 / 2]; + step1[13] = (short)input[22 / 2]; + step1[14] = (short)input[14 / 2]; + step1[15] = (short)input[30 / 2]; + + // stage 2 + step2[0] = step1[0]; + step2[1] = step1[1]; + step2[2] = step1[2]; + step2[3] = step1[3]; + step2[4] = step1[4]; + step2[5] = step1[5]; + step2[6] = step1[6]; + step2[7] = step1[7]; + + temp1 = step1[8] * CosPi30_64 - step1[15] * CosPi2_64; + temp2 = step1[8] * CosPi2_64 + step1[15] * CosPi30_64; + step2[8] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[15] = (short)WrapLow(DctConstRoundShift(temp2)); + + temp1 = step1[9] * CosPi14_64 - step1[14] * CosPi18_64; + temp2 = step1[9] * CosPi18_64 + step1[14] * CosPi14_64; + step2[9] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[14] = (short)WrapLow(DctConstRoundShift(temp2)); + + temp1 = step1[10] * CosPi22_64 - step1[13] * CosPi10_64; + temp2 = step1[10] * CosPi10_64 + step1[13] * CosPi22_64; + step2[10] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[13] = (short)WrapLow(DctConstRoundShift(temp2)); + + temp1 = step1[11] * CosPi6_64 - step1[12] * CosPi26_64; + temp2 = step1[11] * CosPi26_64 + step1[12] * CosPi6_64; + step2[11] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[12] = (short)WrapLow(DctConstRoundShift(temp2)); + + // stage 3 + step1[0] = step2[0]; + step1[1] = step2[1]; + step1[2] = step2[2]; + step1[3] = step2[3]; + + temp1 = step2[4] * CosPi28_64 - step2[7] * CosPi4_64; + temp2 = step2[4] * CosPi4_64 + step2[7] * CosPi28_64; + step1[4] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[7] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = step2[5] * CosPi12_64 - step2[6] * CosPi20_64; + temp2 = step2[5] * CosPi20_64 + step2[6] * CosPi12_64; + step1[5] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[6] = (short)WrapLow(DctConstRoundShift(temp2)); + + step1[8] = (short)WrapLow(step2[8] + step2[9]); + step1[9] = (short)WrapLow(step2[8] - step2[9]); + step1[10] = (short)WrapLow(-step2[10] + step2[11]); + step1[11] = (short)WrapLow(step2[10] + step2[11]); + step1[12] = (short)WrapLow(step2[12] + step2[13]); + step1[13] = (short)WrapLow(step2[12] - step2[13]); + step1[14] = (short)WrapLow(-step2[14] + step2[15]); + step1[15] = (short)WrapLow(step2[14] + step2[15]); + + // stage 4 + temp1 = (step1[0] + step1[1]) * CosPi16_64; + temp2 = (step1[0] - step1[1]) * CosPi16_64; + step2[0] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[1] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = step1[2] * CosPi24_64 - step1[3] * CosPi8_64; + temp2 = step1[2] * CosPi8_64 + step1[3] * CosPi24_64; + step2[2] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[3] = (short)WrapLow(DctConstRoundShift(temp2)); + step2[4] = (short)WrapLow(step1[4] + step1[5]); + step2[5] = (short)WrapLow(step1[4] - step1[5]); + step2[6] = (short)WrapLow(-step1[6] + step1[7]); + step2[7] = (short)WrapLow(step1[6] + step1[7]); + + step2[8] = step1[8]; + step2[15] = step1[15]; + temp1 = -step1[9] * CosPi8_64 + step1[14] * CosPi24_64; + temp2 = step1[9] * CosPi24_64 + step1[14] * CosPi8_64; + step2[9] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[14] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = -step1[10] * CosPi24_64 - step1[13] * CosPi8_64; + temp2 = -step1[10] * CosPi8_64 + step1[13] * CosPi24_64; + step2[10] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[13] = (short)WrapLow(DctConstRoundShift(temp2)); + step2[11] = step1[11]; + step2[12] = step1[12]; + + // stage 5 + step1[0] = (short)WrapLow(step2[0] + step2[3]); + step1[1] = (short)WrapLow(step2[1] + step2[2]); + step1[2] = (short)WrapLow(step2[1] - step2[2]); + step1[3] = (short)WrapLow(step2[0] - step2[3]); + step1[4] = step2[4]; + temp1 = (step2[6] - step2[5]) * CosPi16_64; + temp2 = (step2[5] + step2[6]) * CosPi16_64; + step1[5] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[6] = (short)WrapLow(DctConstRoundShift(temp2)); + step1[7] = step2[7]; + + step1[8] = (short)WrapLow(step2[8] + step2[11]); + step1[9] = (short)WrapLow(step2[9] + step2[10]); + step1[10] = (short)WrapLow(step2[9] - step2[10]); + step1[11] = (short)WrapLow(step2[8] - step2[11]); + step1[12] = (short)WrapLow(-step2[12] + step2[15]); + step1[13] = (short)WrapLow(-step2[13] + step2[14]); + step1[14] = (short)WrapLow(step2[13] + step2[14]); + step1[15] = (short)WrapLow(step2[12] + step2[15]); + + // stage 6 + step2[0] = (short)WrapLow(step1[0] + step1[7]); + step2[1] = (short)WrapLow(step1[1] + step1[6]); + step2[2] = (short)WrapLow(step1[2] + step1[5]); + step2[3] = (short)WrapLow(step1[3] + step1[4]); + step2[4] = (short)WrapLow(step1[3] - step1[4]); + step2[5] = (short)WrapLow(step1[2] - step1[5]); + step2[6] = (short)WrapLow(step1[1] - step1[6]); + step2[7] = (short)WrapLow(step1[0] - step1[7]); + step2[8] = step1[8]; + step2[9] = step1[9]; + temp1 = (-step1[10] + step1[13]) * CosPi16_64; + temp2 = (step1[10] + step1[13]) * CosPi16_64; + step2[10] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[13] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = (-step1[11] + step1[12]) * CosPi16_64; + temp2 = (step1[11] + step1[12]) * CosPi16_64; + step2[11] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[12] = (short)WrapLow(DctConstRoundShift(temp2)); + step2[14] = step1[14]; + step2[15] = step1[15]; + + // stage 7 + output[0] = WrapLow(step2[0] + step2[15]); + output[1] = WrapLow(step2[1] + step2[14]); + output[2] = WrapLow(step2[2] + step2[13]); + output[3] = WrapLow(step2[3] + step2[12]); + output[4] = WrapLow(step2[4] + step2[11]); + output[5] = WrapLow(step2[5] + step2[10]); + output[6] = WrapLow(step2[6] + step2[9]); + output[7] = WrapLow(step2[7] + step2[8]); + output[8] = WrapLow(step2[7] - step2[8]); + output[9] = WrapLow(step2[6] - step2[9]); + output[10] = WrapLow(step2[5] - step2[10]); + output[11] = WrapLow(step2[4] - step2[11]); + output[12] = WrapLow(step2[3] - step2[12]); + output[13] = WrapLow(step2[2] - step2[13]); + output[14] = WrapLow(step2[1] - step2[14]); + output[15] = WrapLow(step2[0] - step2[15]); + } + + public static void Idct16x16256Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i, j; + Span<int> output = stackalloc int[16 * 16]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[16]; + Span<int> tempOut = stackalloc int[16]; + + // First transform rows + for (i = 0; i < 16; ++i) + { + Idct16(input, outptr); + input = input.Slice(16); + outptr = outptr.Slice(16); + } + + // Then transform columns + for (i = 0; i < 16; ++i) + { + for (j = 0; j < 16; ++j) + { + tempIn[j] = output[j * 16 + i]; + } + + Idct16(tempIn, tempOut); + for (j = 0; j < 16; ++j) + { + dest[j * stride + i] = ClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 6)); + } + } + } + + public static void Idct16x1638Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i, j; + Span<int> output = stackalloc int[16 * 16]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[16]; + Span<int> tempOut = stackalloc int[16]; + + // First transform rows. Since all non-zero dct coefficients are in + // upper-left 8x8 area, we only need to calculate first 8 rows here. + for (i = 0; i < 8; ++i) + { + Idct16(input, outptr); + input = input.Slice(16); + outptr = outptr.Slice(16); + } + + // Then transform columns + for (i = 0; i < 16; ++i) + { + for (j = 0; j < 16; ++j) + { + tempIn[j] = output[j * 16 + i]; + } + + Idct16(tempIn, tempOut); + for (j = 0; j < 16; ++j) + { + dest[j * stride + i] = ClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 6)); + } + } + } + + public static void Idct16x1610Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i, j; + Span<int> output = stackalloc int[16 * 16]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[16]; + Span<int> tempOut = stackalloc int[16]; + + // First transform rows. Since all non-zero dct coefficients are in + // upper-left 4x4 area, we only need to calculate first 4 rows here. + for (i = 0; i < 4; ++i) + { + Idct16(input, outptr); + input = input.Slice(16); + outptr = outptr.Slice(16); + } + + // Then transform columns + for (i = 0; i < 16; ++i) + { + for (j = 0; j < 16; ++j) + { + tempIn[j] = output[j * 16 + i]; + } + + Idct16(tempIn, tempOut); + for (j = 0; j < 16; ++j) + { + dest[j * stride + i] = ClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 6)); + } + } + } + + public static void Idct16x161Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i, j; + long a1; + int output = WrapLow(DctConstRoundShift((short)input[0] * CosPi16_64)); + + output = WrapLow(DctConstRoundShift(output * CosPi16_64)); + a1 = BitUtils.RoundPowerOfTwo(output, 6); + for (j = 0; j < 16; ++j) + { + for (i = 0; i < 16; ++i) + { + dest[i] = ClipPixelAdd(dest[i], a1); + } + + dest = dest.Slice(stride); + } + } + + public static void Idct32(ReadOnlySpan<int> input, Span<int> output) + { + Span<short> step1 = stackalloc short[32]; + Span<short> step2 = stackalloc short[32]; + long temp1, temp2; + + // stage 1 + step1[0] = (short)input[0]; + step1[1] = (short)input[16]; + step1[2] = (short)input[8]; + step1[3] = (short)input[24]; + step1[4] = (short)input[4]; + step1[5] = (short)input[20]; + step1[6] = (short)input[12]; + step1[7] = (short)input[28]; + step1[8] = (short)input[2]; + step1[9] = (short)input[18]; + step1[10] = (short)input[10]; + step1[11] = (short)input[26]; + step1[12] = (short)input[6]; + step1[13] = (short)input[22]; + step1[14] = (short)input[14]; + step1[15] = (short)input[30]; + + temp1 = (short)input[1] * CosPi31_64 - (short)input[31] * CosPi1_64; + temp2 = (short)input[1] * CosPi1_64 + (short)input[31] * CosPi31_64; + step1[16] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[31] = (short)WrapLow(DctConstRoundShift(temp2)); + + temp1 = (short)input[17] * CosPi15_64 - (short)input[15] * CosPi17_64; + temp2 = (short)input[17] * CosPi17_64 + (short)input[15] * CosPi15_64; + step1[17] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[30] = (short)WrapLow(DctConstRoundShift(temp2)); + + temp1 = (short)input[9] * CosPi23_64 - (short)input[23] * CosPi9_64; + temp2 = (short)input[9] * CosPi9_64 + (short)input[23] * CosPi23_64; + step1[18] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[29] = (short)WrapLow(DctConstRoundShift(temp2)); + + temp1 = (short)input[25] * CosPi7_64 - (short)input[7] * CosPi25_64; + temp2 = (short)input[25] * CosPi25_64 + (short)input[7] * CosPi7_64; + step1[19] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[28] = (short)WrapLow(DctConstRoundShift(temp2)); + + temp1 = (short)input[5] * CosPi27_64 - (short)input[27] * CosPi5_64; + temp2 = (short)input[5] * CosPi5_64 + (short)input[27] * CosPi27_64; + step1[20] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[27] = (short)WrapLow(DctConstRoundShift(temp2)); + + temp1 = (short)input[21] * CosPi11_64 - (short)input[11] * CosPi21_64; + temp2 = (short)input[21] * CosPi21_64 + (short)input[11] * CosPi11_64; + step1[21] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[26] = (short)WrapLow(DctConstRoundShift(temp2)); + + temp1 = (short)input[13] * CosPi19_64 - (short)input[19] * CosPi13_64; + temp2 = (short)input[13] * CosPi13_64 + (short)input[19] * CosPi19_64; + step1[22] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[25] = (short)WrapLow(DctConstRoundShift(temp2)); + + temp1 = (short)input[29] * CosPi3_64 - (short)input[3] * CosPi29_64; + temp2 = (short)input[29] * CosPi29_64 + (short)input[3] * CosPi3_64; + step1[23] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[24] = (short)WrapLow(DctConstRoundShift(temp2)); + + // stage 2 + step2[0] = step1[0]; + step2[1] = step1[1]; + step2[2] = step1[2]; + step2[3] = step1[3]; + step2[4] = step1[4]; + step2[5] = step1[5]; + step2[6] = step1[6]; + step2[7] = step1[7]; + + temp1 = step1[8] * CosPi30_64 - step1[15] * CosPi2_64; + temp2 = step1[8] * CosPi2_64 + step1[15] * CosPi30_64; + step2[8] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[15] = (short)WrapLow(DctConstRoundShift(temp2)); + + temp1 = step1[9] * CosPi14_64 - step1[14] * CosPi18_64; + temp2 = step1[9] * CosPi18_64 + step1[14] * CosPi14_64; + step2[9] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[14] = (short)WrapLow(DctConstRoundShift(temp2)); + + temp1 = step1[10] * CosPi22_64 - step1[13] * CosPi10_64; + temp2 = step1[10] * CosPi10_64 + step1[13] * CosPi22_64; + step2[10] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[13] = (short)WrapLow(DctConstRoundShift(temp2)); + + temp1 = step1[11] * CosPi6_64 - step1[12] * CosPi26_64; + temp2 = step1[11] * CosPi26_64 + step1[12] * CosPi6_64; + step2[11] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[12] = (short)WrapLow(DctConstRoundShift(temp2)); + + step2[16] = (short)WrapLow(step1[16] + step1[17]); + step2[17] = (short)WrapLow(step1[16] - step1[17]); + step2[18] = (short)WrapLow(-step1[18] + step1[19]); + step2[19] = (short)WrapLow(step1[18] + step1[19]); + step2[20] = (short)WrapLow(step1[20] + step1[21]); + step2[21] = (short)WrapLow(step1[20] - step1[21]); + step2[22] = (short)WrapLow(-step1[22] + step1[23]); + step2[23] = (short)WrapLow(step1[22] + step1[23]); + step2[24] = (short)WrapLow(step1[24] + step1[25]); + step2[25] = (short)WrapLow(step1[24] - step1[25]); + step2[26] = (short)WrapLow(-step1[26] + step1[27]); + step2[27] = (short)WrapLow(step1[26] + step1[27]); + step2[28] = (short)WrapLow(step1[28] + step1[29]); + step2[29] = (short)WrapLow(step1[28] - step1[29]); + step2[30] = (short)WrapLow(-step1[30] + step1[31]); + step2[31] = (short)WrapLow(step1[30] + step1[31]); + + // stage 3 + step1[0] = step2[0]; + step1[1] = step2[1]; + step1[2] = step2[2]; + step1[3] = step2[3]; + + temp1 = step2[4] * CosPi28_64 - step2[7] * CosPi4_64; + temp2 = step2[4] * CosPi4_64 + step2[7] * CosPi28_64; + step1[4] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[7] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = step2[5] * CosPi12_64 - step2[6] * CosPi20_64; + temp2 = step2[5] * CosPi20_64 + step2[6] * CosPi12_64; + step1[5] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[6] = (short)WrapLow(DctConstRoundShift(temp2)); + + step1[8] = (short)WrapLow(step2[8] + step2[9]); + step1[9] = (short)WrapLow(step2[8] - step2[9]); + step1[10] = (short)WrapLow(-step2[10] + step2[11]); + step1[11] = (short)WrapLow(step2[10] + step2[11]); + step1[12] = (short)WrapLow(step2[12] + step2[13]); + step1[13] = (short)WrapLow(step2[12] - step2[13]); + step1[14] = (short)WrapLow(-step2[14] + step2[15]); + step1[15] = (short)WrapLow(step2[14] + step2[15]); + + step1[16] = step2[16]; + step1[31] = step2[31]; + temp1 = -step2[17] * CosPi4_64 + step2[30] * CosPi28_64; + temp2 = step2[17] * CosPi28_64 + step2[30] * CosPi4_64; + step1[17] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[30] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = -step2[18] * CosPi28_64 - step2[29] * CosPi4_64; + temp2 = -step2[18] * CosPi4_64 + step2[29] * CosPi28_64; + step1[18] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[29] = (short)WrapLow(DctConstRoundShift(temp2)); + step1[19] = step2[19]; + step1[20] = step2[20]; + temp1 = -step2[21] * CosPi20_64 + step2[26] * CosPi12_64; + temp2 = step2[21] * CosPi12_64 + step2[26] * CosPi20_64; + step1[21] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[26] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = -step2[22] * CosPi12_64 - step2[25] * CosPi20_64; + temp2 = -step2[22] * CosPi20_64 + step2[25] * CosPi12_64; + step1[22] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[25] = (short)WrapLow(DctConstRoundShift(temp2)); + step1[23] = step2[23]; + step1[24] = step2[24]; + step1[27] = step2[27]; + step1[28] = step2[28]; + + // stage 4 + temp1 = (step1[0] + step1[1]) * CosPi16_64; + temp2 = (step1[0] - step1[1]) * CosPi16_64; + step2[0] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[1] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = step1[2] * CosPi24_64 - step1[3] * CosPi8_64; + temp2 = step1[2] * CosPi8_64 + step1[3] * CosPi24_64; + step2[2] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[3] = (short)WrapLow(DctConstRoundShift(temp2)); + step2[4] = (short)WrapLow(step1[4] + step1[5]); + step2[5] = (short)WrapLow(step1[4] - step1[5]); + step2[6] = (short)WrapLow(-step1[6] + step1[7]); + step2[7] = (short)WrapLow(step1[6] + step1[7]); + + step2[8] = step1[8]; + step2[15] = step1[15]; + temp1 = -step1[9] * CosPi8_64 + step1[14] * CosPi24_64; + temp2 = step1[9] * CosPi24_64 + step1[14] * CosPi8_64; + step2[9] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[14] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = -step1[10] * CosPi24_64 - step1[13] * CosPi8_64; + temp2 = -step1[10] * CosPi8_64 + step1[13] * CosPi24_64; + step2[10] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[13] = (short)WrapLow(DctConstRoundShift(temp2)); + step2[11] = step1[11]; + step2[12] = step1[12]; + + step2[16] = (short)WrapLow(step1[16] + step1[19]); + step2[17] = (short)WrapLow(step1[17] + step1[18]); + step2[18] = (short)WrapLow(step1[17] - step1[18]); + step2[19] = (short)WrapLow(step1[16] - step1[19]); + step2[20] = (short)WrapLow(-step1[20] + step1[23]); + step2[21] = (short)WrapLow(-step1[21] + step1[22]); + step2[22] = (short)WrapLow(step1[21] + step1[22]); + step2[23] = (short)WrapLow(step1[20] + step1[23]); + + step2[24] = (short)WrapLow(step1[24] + step1[27]); + step2[25] = (short)WrapLow(step1[25] + step1[26]); + step2[26] = (short)WrapLow(step1[25] - step1[26]); + step2[27] = (short)WrapLow(step1[24] - step1[27]); + step2[28] = (short)WrapLow(-step1[28] + step1[31]); + step2[29] = (short)WrapLow(-step1[29] + step1[30]); + step2[30] = (short)WrapLow(step1[29] + step1[30]); + step2[31] = (short)WrapLow(step1[28] + step1[31]); + + // stage 5 + step1[0] = (short)WrapLow(step2[0] + step2[3]); + step1[1] = (short)WrapLow(step2[1] + step2[2]); + step1[2] = (short)WrapLow(step2[1] - step2[2]); + step1[3] = (short)WrapLow(step2[0] - step2[3]); + step1[4] = step2[4]; + temp1 = (step2[6] - step2[5]) * CosPi16_64; + temp2 = (step2[5] + step2[6]) * CosPi16_64; + step1[5] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[6] = (short)WrapLow(DctConstRoundShift(temp2)); + step1[7] = step2[7]; + + step1[8] = (short)WrapLow(step2[8] + step2[11]); + step1[9] = (short)WrapLow(step2[9] + step2[10]); + step1[10] = (short)WrapLow(step2[9] - step2[10]); + step1[11] = (short)WrapLow(step2[8] - step2[11]); + step1[12] = (short)WrapLow(-step2[12] + step2[15]); + step1[13] = (short)WrapLow(-step2[13] + step2[14]); + step1[14] = (short)WrapLow(step2[13] + step2[14]); + step1[15] = (short)WrapLow(step2[12] + step2[15]); + + step1[16] = step2[16]; + step1[17] = step2[17]; + temp1 = -step2[18] * CosPi8_64 + step2[29] * CosPi24_64; + temp2 = step2[18] * CosPi24_64 + step2[29] * CosPi8_64; + step1[18] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[29] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = -step2[19] * CosPi8_64 + step2[28] * CosPi24_64; + temp2 = step2[19] * CosPi24_64 + step2[28] * CosPi8_64; + step1[19] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[28] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = -step2[20] * CosPi24_64 - step2[27] * CosPi8_64; + temp2 = -step2[20] * CosPi8_64 + step2[27] * CosPi24_64; + step1[20] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[27] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = -step2[21] * CosPi24_64 - step2[26] * CosPi8_64; + temp2 = -step2[21] * CosPi8_64 + step2[26] * CosPi24_64; + step1[21] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[26] = (short)WrapLow(DctConstRoundShift(temp2)); + step1[22] = step2[22]; + step1[23] = step2[23]; + step1[24] = step2[24]; + step1[25] = step2[25]; + step1[30] = step2[30]; + step1[31] = step2[31]; + + // stage 6 + step2[0] = (short)WrapLow(step1[0] + step1[7]); + step2[1] = (short)WrapLow(step1[1] + step1[6]); + step2[2] = (short)WrapLow(step1[2] + step1[5]); + step2[3] = (short)WrapLow(step1[3] + step1[4]); + step2[4] = (short)WrapLow(step1[3] - step1[4]); + step2[5] = (short)WrapLow(step1[2] - step1[5]); + step2[6] = (short)WrapLow(step1[1] - step1[6]); + step2[7] = (short)WrapLow(step1[0] - step1[7]); + step2[8] = step1[8]; + step2[9] = step1[9]; + temp1 = (-step1[10] + step1[13]) * CosPi16_64; + temp2 = (step1[10] + step1[13]) * CosPi16_64; + step2[10] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[13] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = (-step1[11] + step1[12]) * CosPi16_64; + temp2 = (step1[11] + step1[12]) * CosPi16_64; + step2[11] = (short)WrapLow(DctConstRoundShift(temp1)); + step2[12] = (short)WrapLow(DctConstRoundShift(temp2)); + step2[14] = step1[14]; + step2[15] = step1[15]; + + step2[16] = (short)WrapLow(step1[16] + step1[23]); + step2[17] = (short)WrapLow(step1[17] + step1[22]); + step2[18] = (short)WrapLow(step1[18] + step1[21]); + step2[19] = (short)WrapLow(step1[19] + step1[20]); + step2[20] = (short)WrapLow(step1[19] - step1[20]); + step2[21] = (short)WrapLow(step1[18] - step1[21]); + step2[22] = (short)WrapLow(step1[17] - step1[22]); + step2[23] = (short)WrapLow(step1[16] - step1[23]); + + step2[24] = (short)WrapLow(-step1[24] + step1[31]); + step2[25] = (short)WrapLow(-step1[25] + step1[30]); + step2[26] = (short)WrapLow(-step1[26] + step1[29]); + step2[27] = (short)WrapLow(-step1[27] + step1[28]); + step2[28] = (short)WrapLow(step1[27] + step1[28]); + step2[29] = (short)WrapLow(step1[26] + step1[29]); + step2[30] = (short)WrapLow(step1[25] + step1[30]); + step2[31] = (short)WrapLow(step1[24] + step1[31]); + + // stage 7 + step1[0] = (short)WrapLow(step2[0] + step2[15]); + step1[1] = (short)WrapLow(step2[1] + step2[14]); + step1[2] = (short)WrapLow(step2[2] + step2[13]); + step1[3] = (short)WrapLow(step2[3] + step2[12]); + step1[4] = (short)WrapLow(step2[4] + step2[11]); + step1[5] = (short)WrapLow(step2[5] + step2[10]); + step1[6] = (short)WrapLow(step2[6] + step2[9]); + step1[7] = (short)WrapLow(step2[7] + step2[8]); + step1[8] = (short)WrapLow(step2[7] - step2[8]); + step1[9] = (short)WrapLow(step2[6] - step2[9]); + step1[10] = (short)WrapLow(step2[5] - step2[10]); + step1[11] = (short)WrapLow(step2[4] - step2[11]); + step1[12] = (short)WrapLow(step2[3] - step2[12]); + step1[13] = (short)WrapLow(step2[2] - step2[13]); + step1[14] = (short)WrapLow(step2[1] - step2[14]); + step1[15] = (short)WrapLow(step2[0] - step2[15]); + + step1[16] = step2[16]; + step1[17] = step2[17]; + step1[18] = step2[18]; + step1[19] = step2[19]; + temp1 = (-step2[20] + step2[27]) * CosPi16_64; + temp2 = (step2[20] + step2[27]) * CosPi16_64; + step1[20] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[27] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = (-step2[21] + step2[26]) * CosPi16_64; + temp2 = (step2[21] + step2[26]) * CosPi16_64; + step1[21] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[26] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = (-step2[22] + step2[25]) * CosPi16_64; + temp2 = (step2[22] + step2[25]) * CosPi16_64; + step1[22] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[25] = (short)WrapLow(DctConstRoundShift(temp2)); + temp1 = (-step2[23] + step2[24]) * CosPi16_64; + temp2 = (step2[23] + step2[24]) * CosPi16_64; + step1[23] = (short)WrapLow(DctConstRoundShift(temp1)); + step1[24] = (short)WrapLow(DctConstRoundShift(temp2)); + step1[28] = step2[28]; + step1[29] = step2[29]; + step1[30] = step2[30]; + step1[31] = step2[31]; + + // final stage + output[0] = WrapLow(step1[0] + step1[31]); + output[1] = WrapLow(step1[1] + step1[30]); + output[2] = WrapLow(step1[2] + step1[29]); + output[3] = WrapLow(step1[3] + step1[28]); + output[4] = WrapLow(step1[4] + step1[27]); + output[5] = WrapLow(step1[5] + step1[26]); + output[6] = WrapLow(step1[6] + step1[25]); + output[7] = WrapLow(step1[7] + step1[24]); + output[8] = WrapLow(step1[8] + step1[23]); + output[9] = WrapLow(step1[9] + step1[22]); + output[10] = WrapLow(step1[10] + step1[21]); + output[11] = WrapLow(step1[11] + step1[20]); + output[12] = WrapLow(step1[12] + step1[19]); + output[13] = WrapLow(step1[13] + step1[18]); + output[14] = WrapLow(step1[14] + step1[17]); + output[15] = WrapLow(step1[15] + step1[16]); + output[16] = WrapLow(step1[15] - step1[16]); + output[17] = WrapLow(step1[14] - step1[17]); + output[18] = WrapLow(step1[13] - step1[18]); + output[19] = WrapLow(step1[12] - step1[19]); + output[20] = WrapLow(step1[11] - step1[20]); + output[21] = WrapLow(step1[10] - step1[21]); + output[22] = WrapLow(step1[9] - step1[22]); + output[23] = WrapLow(step1[8] - step1[23]); + output[24] = WrapLow(step1[7] - step1[24]); + output[25] = WrapLow(step1[6] - step1[25]); + output[26] = WrapLow(step1[5] - step1[26]); + output[27] = WrapLow(step1[4] - step1[27]); + output[28] = WrapLow(step1[3] - step1[28]); + output[29] = WrapLow(step1[2] - step1[29]); + output[30] = WrapLow(step1[1] - step1[30]); + output[31] = WrapLow(step1[0] - step1[31]); + } + + public static void Idct32x321024Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i, j; + Span<int> output = stackalloc int[32 * 32]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[32]; + Span<int> tempOut = stackalloc int[32]; + + // Rows + for (i = 0; i < 32; ++i) + { + short zeroCoeff = 0; + for (j = 0; j < 32; ++j) + { + zeroCoeff |= (short)input[j]; + } + + if (zeroCoeff != 0) + { + Idct32(input, outptr); + } + else + { + outptr.Slice(0, 32).Fill(0); + } + + input = input.Slice(32); + outptr = outptr.Slice(32); + } + + // Columns + for (i = 0; i < 32; ++i) + { + for (j = 0; j < 32; ++j) + { + tempIn[j] = output[j * 32 + i]; + } + + Idct32(tempIn, tempOut); + for (j = 0; j < 32; ++j) + { + dest[j * stride + i] = ClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 6)); + } + } + } + + public static void Idct32x32135Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i, j; + Span<int> output = stackalloc int[32 * 32]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[32]; + Span<int> tempOut = stackalloc int[32]; + + // Rows + // Only upper-left 16x16 has non-zero coeff + for (i = 0; i < 16; ++i) + { + Idct32(input, outptr); + input = input.Slice(32); + outptr = outptr.Slice(32); + } + + // Columns + for (i = 0; i < 32; ++i) + { + for (j = 0; j < 32; ++j) + { + tempIn[j] = output[j * 32 + i]; + } + + Idct32(tempIn, tempOut); + for (j = 0; j < 32; ++j) + { + dest[j * stride + i] = ClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 6)); + } + } + } + + public static void Idct32x3234Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i, j; + Span<int> output = stackalloc int[32 * 32]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[32]; + Span<int> tempOut = stackalloc int[32]; + + // Rows + // Only upper-left 8x8 has non-zero coeff + for (i = 0; i < 8; ++i) + { + Idct32(input, outptr); + input = input.Slice(32); + outptr = outptr.Slice(32); + } + + // Columns + for (i = 0; i < 32; ++i) + { + for (j = 0; j < 32; ++j) + { + tempIn[j] = output[j * 32 + i]; + } + + Idct32(tempIn, tempOut); + for (j = 0; j < 32; ++j) + { + dest[j * stride + i] = ClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 6)); + } + } + } + + public static void Idct32x321Add(ReadOnlySpan<int> input, Span<byte> dest, int stride) + { + int i, j; + long a1; + int output = WrapLow(DctConstRoundShift((short)input[0] * CosPi16_64)); + + output = WrapLow(DctConstRoundShift(output * CosPi16_64)); + a1 = BitUtils.RoundPowerOfTwo(output, 6); + + for (j = 0; j < 32; ++j) + { + for (i = 0; i < 32; ++i) + { + dest[i] = ClipPixelAdd(dest[i], a1); + } + + dest = dest.Slice(stride); + } + } + + public static void HighbdIwht4x416Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + /* 4-point reversible, orthonormal inverse Walsh-Hadamard in 3.5 adds, + 0.5 shifts per pixel. */ + int i; + Span<int> output = stackalloc int[16]; + long a1, b1, c1, d1, e1; + ReadOnlySpan<int> ip = input; + Span<int> op = output; + + for (i = 0; i < 4; i++) + { + a1 = ip[0] >> UnitQuantShift; + c1 = ip[1] >> UnitQuantShift; + d1 = ip[2] >> UnitQuantShift; + b1 = ip[3] >> UnitQuantShift; + a1 += c1; + d1 -= b1; + e1 = (a1 - d1) >> 1; + b1 = e1 - b1; + c1 = e1 - c1; + a1 -= b1; + d1 += c1; + op[0] = HighbdWrapLow(a1, bd); + op[1] = HighbdWrapLow(b1, bd); + op[2] = HighbdWrapLow(c1, bd); + op[3] = HighbdWrapLow(d1, bd); + ip = ip.Slice(4); + op = op.Slice(4); + } + + ReadOnlySpan<int> ip2 = output; + for (i = 0; i < 4; i++) + { + a1 = ip2[4 * 0]; + c1 = ip2[4 * 1]; + d1 = ip2[4 * 2]; + b1 = ip2[4 * 3]; + a1 += c1; + d1 -= b1; + e1 = (a1 - d1) >> 1; + b1 = e1 - b1; + c1 = e1 - c1; + a1 -= b1; + d1 += c1; + dest[stride * 0] = HighbdClipPixelAdd(dest[stride * 0], HighbdWrapLow(a1, bd), bd); + dest[stride * 1] = HighbdClipPixelAdd(dest[stride * 1], HighbdWrapLow(b1, bd), bd); + dest[stride * 2] = HighbdClipPixelAdd(dest[stride * 2], HighbdWrapLow(c1, bd), bd); + dest[stride * 3] = HighbdClipPixelAdd(dest[stride * 3], HighbdWrapLow(d1, bd), bd); + + ip2 = ip2.Slice(1); + dest = dest.Slice(1); + } + } + + public static void HighbdIwht4x41Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i; + long a1, e1; + Span<int> tmp = stackalloc int[4]; + ReadOnlySpan<int> ip = input; + Span<int> op = tmp; + + a1 = ip[0] >> UnitQuantShift; + e1 = a1 >> 1; + a1 -= e1; + op[0] = HighbdWrapLow(a1, bd); + op[1] = op[2] = op[3] = HighbdWrapLow(e1, bd); + + ReadOnlySpan<int> ip2 = tmp; + for (i = 0; i < 4; i++) + { + e1 = ip2[0] >> 1; + a1 = ip2[0] - e1; + dest[stride * 0] = HighbdClipPixelAdd(dest[stride * 0], a1, bd); + dest[stride * 1] = HighbdClipPixelAdd(dest[stride * 1], e1, bd); + dest[stride * 2] = HighbdClipPixelAdd(dest[stride * 2], e1, bd); + dest[stride * 3] = HighbdClipPixelAdd(dest[stride * 3], e1, bd); + ip2 = ip2.Slice(1); + dest = dest.Slice(1); + } + } + + public static void HighbdIadst4(ReadOnlySpan<int> input, Span<int> output, int bd) + { + long s0, s1, s2, s3, s4, s5, s6, s7; + int x0 = input[0]; + int x1 = input[1]; + int x2 = input[2]; + int x3 = input[3]; + + if (DetectInvalidHighbdInput(input, 4) != 0) + { + Debug.Assert(false, "invalid highbd txfm input"); + output.Slice(0, 4).Fill(0); + return; + } + + if ((x0 | x1 | x2 | x3) == 0) + { + output.Slice(0, 4).Fill(0); + return; + } + + s0 = (long)SinPi1_9 * x0; + s1 = (long)SinPi2_9 * x0; + s2 = (long)SinPi3_9 * x1; + s3 = (long)SinPi4_9 * x2; + s4 = (long)SinPi1_9 * x2; + s5 = (long)SinPi2_9 * x3; + s6 = (long)SinPi4_9 * x3; + s7 = HighbdWrapLow(x0 - x2 + x3, bd); + + s0 = s0 + s3 + s5; + s1 = s1 - s4 - s6; + s3 = s2; + s2 = SinPi3_9 * s7; + + // 1-D transform scaling factor is sqrt(2). + // The overall dynamic range is 14b (input) + 14b (multiplication scaling) + // + 1b (addition) = 29b. + // Hence the output bit depth is 15b. + output[0] = HighbdWrapLow(DctConstRoundShift(s0 + s3), bd); + output[1] = HighbdWrapLow(DctConstRoundShift(s1 + s3), bd); + output[2] = HighbdWrapLow(DctConstRoundShift(s2), bd); + output[3] = HighbdWrapLow(DctConstRoundShift(s0 + s1 - s3), bd); + } + + public static void HighbdIdct4(ReadOnlySpan<int> input, Span<int> output, int bd) + { + Span<int> step = stackalloc int[4]; + long temp1, temp2; + + if (DetectInvalidHighbdInput(input, 4) != 0) + { + Debug.Assert(false, "invalid highbd txfm input"); + output.Slice(0, 4).Fill(0); + return; + } + + // stage 1 + temp1 = (input[0] + input[2]) * (long)CosPi16_64; + temp2 = (input[0] - input[2]) * (long)CosPi16_64; + step[0] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step[1] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = input[1] * (long)CosPi24_64 - input[3] * (long)CosPi8_64; + temp2 = input[1] * (long)CosPi8_64 + input[3] * (long)CosPi24_64; + step[2] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step[3] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + // stage 2 + output[0] = HighbdWrapLow(step[0] + step[3], bd); + output[1] = HighbdWrapLow(step[1] + step[2], bd); + output[2] = HighbdWrapLow(step[1] - step[2], bd); + output[3] = HighbdWrapLow(step[0] - step[3], bd); + } + + public static void HighbdIdct4x416Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i, j; + Span<int> output = stackalloc int[4 * 4]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[4]; + Span<int> tempOut = stackalloc int[4]; + + // Rows + for (i = 0; i < 4; ++i) + { + HighbdIdct4(input, outptr, bd); + input = input.Slice(4); + outptr = outptr.Slice(4); + } + + // Columns + for (i = 0; i < 4; ++i) + { + for (j = 0; j < 4; ++j) + { + tempIn[j] = output[j * 4 + i]; + } + + HighbdIdct4(tempIn, tempOut, bd); + for (j = 0; j < 4; ++j) + { + dest[j * stride + i] = HighbdClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 4), bd); + } + } + } + + public static void HighbdIdct4x41Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i; + long a1; + int output = HighbdWrapLow(DctConstRoundShift(input[0] * (long)CosPi16_64), bd); + + output = HighbdWrapLow(DctConstRoundShift(output * (long)CosPi16_64), bd); + a1 = BitUtils.RoundPowerOfTwo(output, 4); + + for (i = 0; i < 4; i++) + { + dest[0] = HighbdClipPixelAdd(dest[0], a1, bd); + dest[1] = HighbdClipPixelAdd(dest[1], a1, bd); + dest[2] = HighbdClipPixelAdd(dest[2], a1, bd); + dest[3] = HighbdClipPixelAdd(dest[3], a1, bd); + dest = dest.Slice(stride); + } + } + + public static void HighbdIadst8(ReadOnlySpan<int> input, Span<int> output, int bd) + { + long s0, s1, s2, s3, s4, s5, s6, s7; + int x0 = input[7]; + int x1 = input[0]; + int x2 = input[5]; + int x3 = input[2]; + int x4 = input[3]; + int x5 = input[4]; + int x6 = input[1]; + int x7 = input[6]; + + if (DetectInvalidHighbdInput(input, 8) != 0) + { + Debug.Assert(false, "invalid highbd txfm input"); + output.Slice(0, 8).Fill(0); + return; + } + + if ((x0 | x1 | x2 | x3 | x4 | x5 | x6 | x7) == 0) + { + output.Slice(0, 8).Fill(0); + return; + } + + // stage 1 + s0 = (long)CosPi2_64 * x0 + (long)CosPi30_64 * x1; + s1 = (long)CosPi30_64 * x0 - (long)CosPi2_64 * x1; + s2 = (long)CosPi10_64 * x2 + (long)CosPi22_64 * x3; + s3 = (long)CosPi22_64 * x2 - (long)CosPi10_64 * x3; + s4 = (long)CosPi18_64 * x4 + (long)CosPi14_64 * x5; + s5 = (long)CosPi14_64 * x4 - (long)CosPi18_64 * x5; + s6 = (long)CosPi26_64 * x6 + (long)CosPi6_64 * x7; + s7 = (long)CosPi6_64 * x6 - (long)CosPi26_64 * x7; + + x0 = HighbdWrapLow(DctConstRoundShift(s0 + s4), bd); + x1 = HighbdWrapLow(DctConstRoundShift(s1 + s5), bd); + x2 = HighbdWrapLow(DctConstRoundShift(s2 + s6), bd); + x3 = HighbdWrapLow(DctConstRoundShift(s3 + s7), bd); + x4 = HighbdWrapLow(DctConstRoundShift(s0 - s4), bd); + x5 = HighbdWrapLow(DctConstRoundShift(s1 - s5), bd); + x6 = HighbdWrapLow(DctConstRoundShift(s2 - s6), bd); + x7 = HighbdWrapLow(DctConstRoundShift(s3 - s7), bd); + + // stage 2 + s0 = x0; + s1 = x1; + s2 = x2; + s3 = x3; + s4 = (long)CosPi8_64 * x4 + (long)CosPi24_64 * x5; + s5 = (long)CosPi24_64 * x4 - (long)CosPi8_64 * x5; + s6 = (long)(-CosPi24_64) * x6 + (long)CosPi8_64 * x7; + s7 = (long)CosPi8_64 * x6 + (long)CosPi24_64 * x7; + + x0 = HighbdWrapLow(s0 + s2, bd); + x1 = HighbdWrapLow(s1 + s3, bd); + x2 = HighbdWrapLow(s0 - s2, bd); + x3 = HighbdWrapLow(s1 - s3, bd); + x4 = HighbdWrapLow(DctConstRoundShift(s4 + s6), bd); + x5 = HighbdWrapLow(DctConstRoundShift(s5 + s7), bd); + x6 = HighbdWrapLow(DctConstRoundShift(s4 - s6), bd); + x7 = HighbdWrapLow(DctConstRoundShift(s5 - s7), bd); + + // stage 3 + s2 = (long)CosPi16_64 * (x2 + x3); + s3 = (long)CosPi16_64 * (x2 - x3); + s6 = (long)CosPi16_64 * (x6 + x7); + s7 = (long)CosPi16_64 * (x6 - x7); + + x2 = HighbdWrapLow(DctConstRoundShift(s2), bd); + x3 = HighbdWrapLow(DctConstRoundShift(s3), bd); + x6 = HighbdWrapLow(DctConstRoundShift(s6), bd); + x7 = HighbdWrapLow(DctConstRoundShift(s7), bd); + + output[0] = HighbdWrapLow(x0, bd); + output[1] = HighbdWrapLow(-x4, bd); + output[2] = HighbdWrapLow(x6, bd); + output[3] = HighbdWrapLow(-x2, bd); + output[4] = HighbdWrapLow(x3, bd); + output[5] = HighbdWrapLow(-x7, bd); + output[6] = HighbdWrapLow(x5, bd); + output[7] = HighbdWrapLow(-x1, bd); + } + + public static void HighbdIdct8(ReadOnlySpan<int> input, Span<int> output, int bd) + { + Span<int> step1 = stackalloc int[8]; + Span<int> step2 = stackalloc int[8]; + long temp1, temp2; + + if (DetectInvalidHighbdInput(input, 8) != 0) + { + Debug.Assert(false, "invalid highbd txfm input"); + output.Slice(0, 8).Fill(0); + return; + } + + // stage 1 + step1[0] = input[0]; + step1[2] = input[4]; + step1[1] = input[2]; + step1[3] = input[6]; + temp1 = input[1] * (long)CosPi28_64 - input[7] * (long)CosPi4_64; + temp2 = input[1] * (long)CosPi4_64 + input[7] * (long)CosPi28_64; + step1[4] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[7] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = input[5] * (long)CosPi12_64 - input[3] * (long)CosPi20_64; + temp2 = input[5] * (long)CosPi20_64 + input[3] * (long)CosPi12_64; + step1[5] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[6] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + // stage 2 & stage 3 - even half + HighbdIdct4(step1, step1, bd); + + // stage 2 - odd half + step2[4] = HighbdWrapLow(step1[4] + step1[5], bd); + step2[5] = HighbdWrapLow(step1[4] - step1[5], bd); + step2[6] = HighbdWrapLow(-step1[6] + step1[7], bd); + step2[7] = HighbdWrapLow(step1[6] + step1[7], bd); + + // stage 3 - odd half + step1[4] = step2[4]; + temp1 = (step2[6] - step2[5]) * (long)CosPi16_64; + temp2 = (step2[5] + step2[6]) * (long)CosPi16_64; + step1[5] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[6] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + step1[7] = step2[7]; + + // stage 4 + output[0] = HighbdWrapLow(step1[0] + step1[7], bd); + output[1] = HighbdWrapLow(step1[1] + step1[6], bd); + output[2] = HighbdWrapLow(step1[2] + step1[5], bd); + output[3] = HighbdWrapLow(step1[3] + step1[4], bd); + output[4] = HighbdWrapLow(step1[3] - step1[4], bd); + output[5] = HighbdWrapLow(step1[2] - step1[5], bd); + output[6] = HighbdWrapLow(step1[1] - step1[6], bd); + output[7] = HighbdWrapLow(step1[0] - step1[7], bd); + } + + public static void HighbdIdct8x864Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i, j; + Span<int> output = stackalloc int[8 * 8]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[8]; + Span<int> tempOut = stackalloc int[8]; + + // First transform rows + for (i = 0; i < 8; ++i) + { + HighbdIdct8(input, outptr, bd); + input = input.Slice(8); + outptr = outptr.Slice(8); + } + + // Then transform columns + for (i = 0; i < 8; ++i) + { + for (j = 0; j < 8; ++j) + { + tempIn[j] = output[j * 8 + i]; + } + + HighbdIdct8(tempIn, tempOut, bd); + for (j = 0; j < 8; ++j) + { + dest[j * stride + i] = HighbdClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 5), bd); + } + } + } + + public static void HighbdIdct8x812Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i, j; + Span<int> output = stackalloc int[8 * 8]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[8]; + Span<int> tempOut = stackalloc int[8]; + + // First transform rows + // Only first 4 row has non-zero coefs + for (i = 0; i < 4; ++i) + { + HighbdIdct8(input, outptr, bd); + input = input.Slice(8); + outptr = outptr.Slice(8); + } + + // Then transform columns + for (i = 0; i < 8; ++i) + { + for (j = 0; j < 8; ++j) + { + tempIn[j] = output[j * 8 + i]; + } + + HighbdIdct8(tempIn, tempOut, bd); + for (j = 0; j < 8; ++j) + { + dest[j * stride + i] = HighbdClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 5), bd); + } + } + } + + public static void vpx_Highbdidct8x8_1_add_c(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i, j; + long a1; + int output = HighbdWrapLow(DctConstRoundShift(input[0] * (long)CosPi16_64), bd); + + output = HighbdWrapLow(DctConstRoundShift(output * (long)CosPi16_64), bd); + a1 = BitUtils.RoundPowerOfTwo(output, 5); + for (j = 0; j < 8; ++j) + { + for (i = 0; i < 8; ++i) + { + dest[i] = HighbdClipPixelAdd(dest[i], a1, bd); + } + + dest = dest.Slice(stride); + } + } + + public static void HighbdIadst16(ReadOnlySpan<int> input, Span<int> output, int bd) + { + long s0, s1, s2, s3, s4, s5, s6, s7, s8; + long s9, s10, s11, s12, s13, s14, s15; + int x0 = input[15]; + int x1 = input[0]; + int x2 = input[13]; + int x3 = input[2]; + int x4 = input[11]; + int x5 = input[4]; + int x6 = input[9]; + int x7 = input[6]; + int x8 = input[7]; + int x9 = input[8]; + int x10 = input[5]; + int x11 = input[10]; + int x12 = input[3]; + int x13 = input[12]; + int x14 = input[1]; + int x15 = input[14]; + if (DetectInvalidHighbdInput(input, 16) != 0) + { + Debug.Assert(false, "invalid highbd txfm input"); + output.Slice(0, 16).Fill(0); + return; + } + + if ((x0 | x1 | x2 | x3 | x4 | x5 | x6 | x7 | x8 | x9 | x10 | x11 | x12 | x13 | x14 | x15) == 0) + { + output.Slice(0, 16).Fill(0); + return; + } + + // stage 1 + s0 = x0 * (long)CosPi1_64 + x1 * (long)CosPi31_64; + s1 = x0 * (long)CosPi31_64 - x1 * (long)CosPi1_64; + s2 = x2 * (long)CosPi5_64 + x3 * (long)CosPi27_64; + s3 = x2 * (long)CosPi27_64 - x3 * (long)CosPi5_64; + s4 = x4 * (long)CosPi9_64 + x5 * (long)CosPi23_64; + s5 = x4 * (long)CosPi23_64 - x5 * (long)CosPi9_64; + s6 = x6 * (long)CosPi13_64 + x7 * (long)CosPi19_64; + s7 = x6 * (long)CosPi19_64 - x7 * (long)CosPi13_64; + s8 = x8 * (long)CosPi17_64 + x9 * (long)CosPi15_64; + s9 = x8 * (long)CosPi15_64 - x9 * (long)CosPi17_64; + s10 = x10 * (long)CosPi21_64 + x11 * (long)CosPi11_64; + s11 = x10 * (long)CosPi11_64 - x11 * (long)CosPi21_64; + s12 = x12 * (long)CosPi25_64 + x13 * (long)CosPi7_64; + s13 = x12 * (long)CosPi7_64 - x13 * (long)CosPi25_64; + s14 = x14 * (long)CosPi29_64 + x15 * (long)CosPi3_64; + s15 = x14 * (long)CosPi3_64 - x15 * (long)CosPi29_64; + + x0 = HighbdWrapLow(DctConstRoundShift(s0 + s8), bd); + x1 = HighbdWrapLow(DctConstRoundShift(s1 + s9), bd); + x2 = HighbdWrapLow(DctConstRoundShift(s2 + s10), bd); + x3 = HighbdWrapLow(DctConstRoundShift(s3 + s11), bd); + x4 = HighbdWrapLow(DctConstRoundShift(s4 + s12), bd); + x5 = HighbdWrapLow(DctConstRoundShift(s5 + s13), bd); + x6 = HighbdWrapLow(DctConstRoundShift(s6 + s14), bd); + x7 = HighbdWrapLow(DctConstRoundShift(s7 + s15), bd); + x8 = HighbdWrapLow(DctConstRoundShift(s0 - s8), bd); + x9 = HighbdWrapLow(DctConstRoundShift(s1 - s9), bd); + x10 = HighbdWrapLow(DctConstRoundShift(s2 - s10), bd); + x11 = HighbdWrapLow(DctConstRoundShift(s3 - s11), bd); + x12 = HighbdWrapLow(DctConstRoundShift(s4 - s12), bd); + x13 = HighbdWrapLow(DctConstRoundShift(s5 - s13), bd); + x14 = HighbdWrapLow(DctConstRoundShift(s6 - s14), bd); + x15 = HighbdWrapLow(DctConstRoundShift(s7 - s15), bd); + + // stage 2 + s0 = x0; + s1 = x1; + s2 = x2; + s3 = x3; + s4 = x4; + s5 = x5; + s6 = x6; + s7 = x7; + s8 = x8 * (long)CosPi4_64 + x9 * (long)CosPi28_64; + s9 = x8 * (long)CosPi28_64 - x9 * (long)CosPi4_64; + s10 = x10 * (long)CosPi20_64 + x11 * (long)CosPi12_64; + s11 = x10 * (long)CosPi12_64 - x11 * (long)CosPi20_64; + s12 = -x12 * (long)CosPi28_64 + x13 * (long)CosPi4_64; + s13 = x12 * (long)CosPi4_64 + x13 * (long)CosPi28_64; + s14 = -x14 * (long)CosPi12_64 + x15 * (long)CosPi20_64; + s15 = x14 * (long)CosPi20_64 + x15 * (long)CosPi12_64; + + x0 = HighbdWrapLow(s0 + s4, bd); + x1 = HighbdWrapLow(s1 + s5, bd); + x2 = HighbdWrapLow(s2 + s6, bd); + x3 = HighbdWrapLow(s3 + s7, bd); + x4 = HighbdWrapLow(s0 - s4, bd); + x5 = HighbdWrapLow(s1 - s5, bd); + x6 = HighbdWrapLow(s2 - s6, bd); + x7 = HighbdWrapLow(s3 - s7, bd); + x8 = HighbdWrapLow(DctConstRoundShift(s8 + s12), bd); + x9 = HighbdWrapLow(DctConstRoundShift(s9 + s13), bd); + x10 = HighbdWrapLow(DctConstRoundShift(s10 + s14), bd); + x11 = HighbdWrapLow(DctConstRoundShift(s11 + s15), bd); + x12 = HighbdWrapLow(DctConstRoundShift(s8 - s12), bd); + x13 = HighbdWrapLow(DctConstRoundShift(s9 - s13), bd); + x14 = HighbdWrapLow(DctConstRoundShift(s10 - s14), bd); + x15 = HighbdWrapLow(DctConstRoundShift(s11 - s15), bd); + + // stage 3 + s0 = x0; + s1 = x1; + s2 = x2; + s3 = x3; + s4 = x4 * (long)CosPi8_64 + x5 * (long)CosPi24_64; + s5 = x4 * (long)CosPi24_64 - x5 * (long)CosPi8_64; + s6 = -x6 * (long)CosPi24_64 + x7 * (long)CosPi8_64; + s7 = x6 * (long)CosPi8_64 + x7 * (long)CosPi24_64; + s8 = x8; + s9 = x9; + s10 = x10; + s11 = x11; + s12 = x12 * (long)CosPi8_64 + x13 * (long)CosPi24_64; + s13 = x12 * (long)CosPi24_64 - x13 * (long)CosPi8_64; + s14 = -x14 * (long)CosPi24_64 + x15 * (long)CosPi8_64; + s15 = x14 * (long)CosPi8_64 + x15 * (long)CosPi24_64; + + x0 = HighbdWrapLow(s0 + s2, bd); + x1 = HighbdWrapLow(s1 + s3, bd); + x2 = HighbdWrapLow(s0 - s2, bd); + x3 = HighbdWrapLow(s1 - s3, bd); + x4 = HighbdWrapLow(DctConstRoundShift(s4 + s6), bd); + x5 = HighbdWrapLow(DctConstRoundShift(s5 + s7), bd); + x6 = HighbdWrapLow(DctConstRoundShift(s4 - s6), bd); + x7 = HighbdWrapLow(DctConstRoundShift(s5 - s7), bd); + x8 = HighbdWrapLow(s8 + s10, bd); + x9 = HighbdWrapLow(s9 + s11, bd); + x10 = HighbdWrapLow(s8 - s10, bd); + x11 = HighbdWrapLow(s9 - s11, bd); + x12 = HighbdWrapLow(DctConstRoundShift(s12 + s14), bd); + x13 = HighbdWrapLow(DctConstRoundShift(s13 + s15), bd); + x14 = HighbdWrapLow(DctConstRoundShift(s12 - s14), bd); + x15 = HighbdWrapLow(DctConstRoundShift(s13 - s15), bd); + + // stage 4 + s2 = (long)(-CosPi16_64) * (x2 + x3); + s3 = (long)CosPi16_64 * (x2 - x3); + s6 = (long)CosPi16_64 * (x6 + x7); + s7 = (long)CosPi16_64 * (-x6 + x7); + s10 = (long)CosPi16_64 * (x10 + x11); + s11 = (long)CosPi16_64 * (-x10 + x11); + s14 = (long)(-CosPi16_64) * (x14 + x15); + s15 = (long)CosPi16_64 * (x14 - x15); + + x2 = HighbdWrapLow(DctConstRoundShift(s2), bd); + x3 = HighbdWrapLow(DctConstRoundShift(s3), bd); + x6 = HighbdWrapLow(DctConstRoundShift(s6), bd); + x7 = HighbdWrapLow(DctConstRoundShift(s7), bd); + x10 = HighbdWrapLow(DctConstRoundShift(s10), bd); + x11 = HighbdWrapLow(DctConstRoundShift(s11), bd); + x14 = HighbdWrapLow(DctConstRoundShift(s14), bd); + x15 = HighbdWrapLow(DctConstRoundShift(s15), bd); + + output[0] = HighbdWrapLow(x0, bd); + output[1] = HighbdWrapLow(-x8, bd); + output[2] = HighbdWrapLow(x12, bd); + output[3] = HighbdWrapLow(-x4, bd); + output[4] = HighbdWrapLow(x6, bd); + output[5] = HighbdWrapLow(x14, bd); + output[6] = HighbdWrapLow(x10, bd); + output[7] = HighbdWrapLow(x2, bd); + output[8] = HighbdWrapLow(x3, bd); + output[9] = HighbdWrapLow(x11, bd); + output[10] = HighbdWrapLow(x15, bd); + output[11] = HighbdWrapLow(x7, bd); + output[12] = HighbdWrapLow(x5, bd); + output[13] = HighbdWrapLow(-x13, bd); + output[14] = HighbdWrapLow(x9, bd); + output[15] = HighbdWrapLow(-x1, bd); + } + + public static void HighbdIdct16(ReadOnlySpan<int> input, Span<int> output, int bd) + { + Span<int> step1 = stackalloc int[16]; + Span<int> step2 = stackalloc int[16]; + long temp1, temp2; + + if (DetectInvalidHighbdInput(input, 16) != 0) + { + Debug.Assert(false, "invalid highbd txfm input"); + output.Slice(0, 16).Fill(0); + return; + } + + // stage 1 + step1[0] = input[0 / 2]; + step1[1] = input[16 / 2]; + step1[2] = input[8 / 2]; + step1[3] = input[24 / 2]; + step1[4] = input[4 / 2]; + step1[5] = input[20 / 2]; + step1[6] = input[12 / 2]; + step1[7] = input[28 / 2]; + step1[8] = input[2 / 2]; + step1[9] = input[18 / 2]; + step1[10] = input[10 / 2]; + step1[11] = input[26 / 2]; + step1[12] = input[6 / 2]; + step1[13] = input[22 / 2]; + step1[14] = input[14 / 2]; + step1[15] = input[30 / 2]; + + // stage 2 + step2[0] = step1[0]; + step2[1] = step1[1]; + step2[2] = step1[2]; + step2[3] = step1[3]; + step2[4] = step1[4]; + step2[5] = step1[5]; + step2[6] = step1[6]; + step2[7] = step1[7]; + + temp1 = step1[8] * (long)CosPi30_64 - step1[15] * (long)CosPi2_64; + temp2 = step1[8] * (long)CosPi2_64 + step1[15] * (long)CosPi30_64; + step2[8] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[15] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + temp1 = step1[9] * (long)CosPi14_64 - step1[14] * (long)CosPi18_64; + temp2 = step1[9] * (long)CosPi18_64 + step1[14] * (long)CosPi14_64; + step2[9] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[14] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + temp1 = step1[10] * (long)CosPi22_64 - step1[13] * (long)CosPi10_64; + temp2 = step1[10] * (long)CosPi10_64 + step1[13] * (long)CosPi22_64; + step2[10] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[13] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + temp1 = step1[11] * (long)CosPi6_64 - step1[12] * (long)CosPi26_64; + temp2 = step1[11] * (long)CosPi26_64 + step1[12] * (long)CosPi6_64; + step2[11] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[12] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + // stage 3 + step1[0] = step2[0]; + step1[1] = step2[1]; + step1[2] = step2[2]; + step1[3] = step2[3]; + + temp1 = step2[4] * (long)CosPi28_64 - step2[7] * (long)CosPi4_64; + temp2 = step2[4] * (long)CosPi4_64 + step2[7] * (long)CosPi28_64; + step1[4] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[7] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = step2[5] * (long)CosPi12_64 - step2[6] * (long)CosPi20_64; + temp2 = step2[5] * (long)CosPi20_64 + step2[6] * (long)CosPi12_64; + step1[5] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[6] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + step1[8] = HighbdWrapLow(step2[8] + step2[9], bd); + step1[9] = HighbdWrapLow(step2[8] - step2[9], bd); + step1[10] = HighbdWrapLow(-step2[10] + step2[11], bd); + step1[11] = HighbdWrapLow(step2[10] + step2[11], bd); + step1[12] = HighbdWrapLow(step2[12] + step2[13], bd); + step1[13] = HighbdWrapLow(step2[12] - step2[13], bd); + step1[14] = HighbdWrapLow(-step2[14] + step2[15], bd); + step1[15] = HighbdWrapLow(step2[14] + step2[15], bd); + + // stage 4 + temp1 = (step1[0] + step1[1]) * (long)CosPi16_64; + temp2 = (step1[0] - step1[1]) * (long)CosPi16_64; + step2[0] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[1] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = step1[2] * (long)CosPi24_64 - step1[3] * (long)CosPi8_64; + temp2 = step1[2] * (long)CosPi8_64 + step1[3] * (long)CosPi24_64; + step2[2] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[3] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + step2[4] = HighbdWrapLow(step1[4] + step1[5], bd); + step2[5] = HighbdWrapLow(step1[4] - step1[5], bd); + step2[6] = HighbdWrapLow(-step1[6] + step1[7], bd); + step2[7] = HighbdWrapLow(step1[6] + step1[7], bd); + + step2[8] = step1[8]; + step2[15] = step1[15]; + temp1 = -step1[9] * (long)CosPi8_64 + step1[14] * (long)CosPi24_64; + temp2 = step1[9] * (long)CosPi24_64 + step1[14] * (long)CosPi8_64; + step2[9] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[14] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = -step1[10] * (long)CosPi24_64 - step1[13] * (long)CosPi8_64; + temp2 = -step1[10] * (long)CosPi8_64 + step1[13] * (long)CosPi24_64; + step2[10] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[13] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + step2[11] = step1[11]; + step2[12] = step1[12]; + + // stage 5 + step1[0] = HighbdWrapLow(step2[0] + step2[3], bd); + step1[1] = HighbdWrapLow(step2[1] + step2[2], bd); + step1[2] = HighbdWrapLow(step2[1] - step2[2], bd); + step1[3] = HighbdWrapLow(step2[0] - step2[3], bd); + step1[4] = step2[4]; + temp1 = (step2[6] - step2[5]) * (long)CosPi16_64; + temp2 = (step2[5] + step2[6]) * (long)CosPi16_64; + step1[5] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[6] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + step1[7] = step2[7]; + + step1[8] = HighbdWrapLow(step2[8] + step2[11], bd); + step1[9] = HighbdWrapLow(step2[9] + step2[10], bd); + step1[10] = HighbdWrapLow(step2[9] - step2[10], bd); + step1[11] = HighbdWrapLow(step2[8] - step2[11], bd); + step1[12] = HighbdWrapLow(-step2[12] + step2[15], bd); + step1[13] = HighbdWrapLow(-step2[13] + step2[14], bd); + step1[14] = HighbdWrapLow(step2[13] + step2[14], bd); + step1[15] = HighbdWrapLow(step2[12] + step2[15], bd); + + // stage 6 + step2[0] = HighbdWrapLow(step1[0] + step1[7], bd); + step2[1] = HighbdWrapLow(step1[1] + step1[6], bd); + step2[2] = HighbdWrapLow(step1[2] + step1[5], bd); + step2[3] = HighbdWrapLow(step1[3] + step1[4], bd); + step2[4] = HighbdWrapLow(step1[3] - step1[4], bd); + step2[5] = HighbdWrapLow(step1[2] - step1[5], bd); + step2[6] = HighbdWrapLow(step1[1] - step1[6], bd); + step2[7] = HighbdWrapLow(step1[0] - step1[7], bd); + step2[8] = step1[8]; + step2[9] = step1[9]; + temp1 = (-step1[10] + step1[13]) * (long)CosPi16_64; + temp2 = (step1[10] + step1[13]) * (long)CosPi16_64; + step2[10] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[13] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = (-step1[11] + step1[12]) * (long)CosPi16_64; + temp2 = (step1[11] + step1[12]) * (long)CosPi16_64; + step2[11] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[12] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + step2[14] = step1[14]; + step2[15] = step1[15]; + + // stage 7 + output[0] = HighbdWrapLow(step2[0] + step2[15], bd); + output[1] = HighbdWrapLow(step2[1] + step2[14], bd); + output[2] = HighbdWrapLow(step2[2] + step2[13], bd); + output[3] = HighbdWrapLow(step2[3] + step2[12], bd); + output[4] = HighbdWrapLow(step2[4] + step2[11], bd); + output[5] = HighbdWrapLow(step2[5] + step2[10], bd); + output[6] = HighbdWrapLow(step2[6] + step2[9], bd); + output[7] = HighbdWrapLow(step2[7] + step2[8], bd); + output[8] = HighbdWrapLow(step2[7] - step2[8], bd); + output[9] = HighbdWrapLow(step2[6] - step2[9], bd); + output[10] = HighbdWrapLow(step2[5] - step2[10], bd); + output[11] = HighbdWrapLow(step2[4] - step2[11], bd); + output[12] = HighbdWrapLow(step2[3] - step2[12], bd); + output[13] = HighbdWrapLow(step2[2] - step2[13], bd); + output[14] = HighbdWrapLow(step2[1] - step2[14], bd); + output[15] = HighbdWrapLow(step2[0] - step2[15], bd); + } + + public static void HighbdIdct16x16256Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i, j; + Span<int> output = stackalloc int[16 * 16]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[16]; + Span<int> tempOut = stackalloc int[16]; + + // First transform rows + for (i = 0; i < 16; ++i) + { + HighbdIdct16(input, outptr, bd); + input = input.Slice(16); + outptr = outptr.Slice(16); + } + + // Then transform columns + for (i = 0; i < 16; ++i) + { + for (j = 0; j < 16; ++j) + { + tempIn[j] = output[j * 16 + i]; + } + + HighbdIdct16(tempIn, tempOut, bd); + for (j = 0; j < 16; ++j) + { + dest[j * stride + i] = HighbdClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 6), bd); + } + } + } + + public static void HighbdIdct16x1638Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i, j; + Span<int> output = stackalloc int[16 * 16]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[16]; + Span<int> tempOut = stackalloc int[16]; + + // First transform rows. Since all non-zero dct coefficients are in + // upper-left 8x8 area, we only need to calculate first 8 rows here. + for (i = 0; i < 8; ++i) + { + HighbdIdct16(input, outptr, bd); + input = input.Slice(16); + outptr = outptr.Slice(16); + } + + // Then transform columns + for (i = 0; i < 16; ++i) + { + Span<ushort> destT = dest; + for (j = 0; j < 16; ++j) + { + tempIn[j] = output[j * 16 + i]; + } + + HighbdIdct16(tempIn, tempOut, bd); + for (j = 0; j < 16; ++j) + { + destT[i] = HighbdClipPixelAdd(destT[i], BitUtils.RoundPowerOfTwo(tempOut[j], 6), bd); + destT = destT.Slice(stride); + } + } + } + + public static void HighbdIdct16x1610Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i, j; + Span<int> output = stackalloc int[16 * 16]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[16]; + Span<int> tempOut = stackalloc int[16]; + + // First transform rows. Since all non-zero dct coefficients are in + // upper-left 4x4 area, we only need to calculate first 4 rows here. + for (i = 0; i < 4; ++i) + { + HighbdIdct16(input, outptr, bd); + input = input.Slice(16); + outptr = outptr.Slice(16); + } + + // Then transform columns + for (i = 0; i < 16; ++i) + { + for (j = 0; j < 16; ++j) + { + tempIn[j] = output[j * 16 + i]; + } + + HighbdIdct16(tempIn, tempOut, bd); + for (j = 0; j < 16; ++j) + { + dest[j * stride + i] = HighbdClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 6), bd); + } + } + } + + public static void HighbdIdct16x161Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i, j; + long a1; + int output = HighbdWrapLow(DctConstRoundShift(input[0] * (long)CosPi16_64), bd); + + output = HighbdWrapLow(DctConstRoundShift(output * (long)CosPi16_64), bd); + a1 = BitUtils.RoundPowerOfTwo(output, 6); + for (j = 0; j < 16; ++j) + { + for (i = 0; i < 16; ++i) + { + dest[i] = HighbdClipPixelAdd(dest[i], a1, bd); + } + + dest = dest.Slice(stride); + } + } + + public static void HighbdIdct32(ReadOnlySpan<int> input, Span<int> output, int bd) + { + Span<int> step1 = stackalloc int[32]; + Span<int> step2 = stackalloc int[32]; + long temp1, temp2; + + if (DetectInvalidHighbdInput(input, 32) != 0) + { + Debug.Assert(false, "invalid highbd txfm input"); + output.Slice(0, 32).Fill(0); + return; + } + + // stage 1 + step1[0] = input[0]; + step1[1] = input[16]; + step1[2] = input[8]; + step1[3] = input[24]; + step1[4] = input[4]; + step1[5] = input[20]; + step1[6] = input[12]; + step1[7] = input[28]; + step1[8] = input[2]; + step1[9] = input[18]; + step1[10] = input[10]; + step1[11] = input[26]; + step1[12] = input[6]; + step1[13] = input[22]; + step1[14] = input[14]; + step1[15] = input[30]; + + temp1 = input[1] * (long)CosPi31_64 - input[31] * (long)CosPi1_64; + temp2 = input[1] * (long)CosPi1_64 + input[31] * (long)CosPi31_64; + step1[16] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[31] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + temp1 = input[17] * (long)CosPi15_64 - input[15] * (long)CosPi17_64; + temp2 = input[17] * (long)CosPi17_64 + input[15] * (long)CosPi15_64; + step1[17] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[30] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + temp1 = input[9] * (long)CosPi23_64 - input[23] * (long)CosPi9_64; + temp2 = input[9] * (long)CosPi9_64 + input[23] * (long)CosPi23_64; + step1[18] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[29] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + temp1 = input[25] * (long)CosPi7_64 - input[7] * (long)CosPi25_64; + temp2 = input[25] * (long)CosPi25_64 + input[7] * (long)CosPi7_64; + step1[19] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[28] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + temp1 = input[5] * (long)CosPi27_64 - input[27] * (long)CosPi5_64; + temp2 = input[5] * (long)CosPi5_64 + input[27] * (long)CosPi27_64; + step1[20] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[27] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + temp1 = input[21] * (long)CosPi11_64 - input[11] * (long)CosPi21_64; + temp2 = input[21] * (long)CosPi21_64 + input[11] * (long)CosPi11_64; + step1[21] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[26] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + temp1 = input[13] * (long)CosPi19_64 - input[19] * (long)CosPi13_64; + temp2 = input[13] * (long)CosPi13_64 + input[19] * (long)CosPi19_64; + step1[22] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[25] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + temp1 = input[29] * (long)CosPi3_64 - input[3] * (long)CosPi29_64; + temp2 = input[29] * (long)CosPi29_64 + input[3] * (long)CosPi3_64; + step1[23] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[24] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + // stage 2 + step2[0] = step1[0]; + step2[1] = step1[1]; + step2[2] = step1[2]; + step2[3] = step1[3]; + step2[4] = step1[4]; + step2[5] = step1[5]; + step2[6] = step1[6]; + step2[7] = step1[7]; + + temp1 = step1[8] * (long)CosPi30_64 - step1[15] * (long)CosPi2_64; + temp2 = step1[8] * (long)CosPi2_64 + step1[15] * (long)CosPi30_64; + step2[8] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[15] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + temp1 = step1[9] * (long)CosPi14_64 - step1[14] * (long)CosPi18_64; + temp2 = step1[9] * (long)CosPi18_64 + step1[14] * (long)CosPi14_64; + step2[9] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[14] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + temp1 = step1[10] * (long)CosPi22_64 - step1[13] * (long)CosPi10_64; + temp2 = step1[10] * (long)CosPi10_64 + step1[13] * (long)CosPi22_64; + step2[10] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[13] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + temp1 = step1[11] * (long)CosPi6_64 - step1[12] * (long)CosPi26_64; + temp2 = step1[11] * (long)CosPi26_64 + step1[12] * (long)CosPi6_64; + step2[11] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[12] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + step2[16] = HighbdWrapLow(step1[16] + step1[17], bd); + step2[17] = HighbdWrapLow(step1[16] - step1[17], bd); + step2[18] = HighbdWrapLow(-step1[18] + step1[19], bd); + step2[19] = HighbdWrapLow(step1[18] + step1[19], bd); + step2[20] = HighbdWrapLow(step1[20] + step1[21], bd); + step2[21] = HighbdWrapLow(step1[20] - step1[21], bd); + step2[22] = HighbdWrapLow(-step1[22] + step1[23], bd); + step2[23] = HighbdWrapLow(step1[22] + step1[23], bd); + step2[24] = HighbdWrapLow(step1[24] + step1[25], bd); + step2[25] = HighbdWrapLow(step1[24] - step1[25], bd); + step2[26] = HighbdWrapLow(-step1[26] + step1[27], bd); + step2[27] = HighbdWrapLow(step1[26] + step1[27], bd); + step2[28] = HighbdWrapLow(step1[28] + step1[29], bd); + step2[29] = HighbdWrapLow(step1[28] - step1[29], bd); + step2[30] = HighbdWrapLow(-step1[30] + step1[31], bd); + step2[31] = HighbdWrapLow(step1[30] + step1[31], bd); + + // stage 3 + step1[0] = step2[0]; + step1[1] = step2[1]; + step1[2] = step2[2]; + step1[3] = step2[3]; + + temp1 = step2[4] * (long)CosPi28_64 - step2[7] * (long)CosPi4_64; + temp2 = step2[4] * (long)CosPi4_64 + step2[7] * (long)CosPi28_64; + step1[4] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[7] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = step2[5] * (long)CosPi12_64 - step2[6] * (long)CosPi20_64; + temp2 = step2[5] * (long)CosPi20_64 + step2[6] * (long)CosPi12_64; + step1[5] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[6] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + + step1[8] = HighbdWrapLow(step2[8] + step2[9], bd); + step1[9] = HighbdWrapLow(step2[8] - step2[9], bd); + step1[10] = HighbdWrapLow(-step2[10] + step2[11], bd); + step1[11] = HighbdWrapLow(step2[10] + step2[11], bd); + step1[12] = HighbdWrapLow(step2[12] + step2[13], bd); + step1[13] = HighbdWrapLow(step2[12] - step2[13], bd); + step1[14] = HighbdWrapLow(-step2[14] + step2[15], bd); + step1[15] = HighbdWrapLow(step2[14] + step2[15], bd); + + step1[16] = step2[16]; + step1[31] = step2[31]; + temp1 = -step2[17] * (long)CosPi4_64 + step2[30] * (long)CosPi28_64; + temp2 = step2[17] * (long)CosPi28_64 + step2[30] * (long)CosPi4_64; + step1[17] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[30] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = -step2[18] * (long)CosPi28_64 - step2[29] * (long)CosPi4_64; + temp2 = -step2[18] * (long)CosPi4_64 + step2[29] * (long)CosPi28_64; + step1[18] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[29] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + step1[19] = step2[19]; + step1[20] = step2[20]; + temp1 = -step2[21] * (long)CosPi20_64 + step2[26] * (long)CosPi12_64; + temp2 = step2[21] * (long)CosPi12_64 + step2[26] * (long)CosPi20_64; + step1[21] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[26] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = -step2[22] * (long)CosPi12_64 - step2[25] * (long)CosPi20_64; + temp2 = -step2[22] * (long)CosPi20_64 + step2[25] * (long)CosPi12_64; + step1[22] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[25] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + step1[23] = step2[23]; + step1[24] = step2[24]; + step1[27] = step2[27]; + step1[28] = step2[28]; + + // stage 4 + temp1 = (step1[0] + step1[1]) * (long)CosPi16_64; + temp2 = (step1[0] - step1[1]) * (long)CosPi16_64; + step2[0] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[1] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = step1[2] * (long)CosPi24_64 - step1[3] * (long)CosPi8_64; + temp2 = step1[2] * (long)CosPi8_64 + step1[3] * (long)CosPi24_64; + step2[2] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[3] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + step2[4] = HighbdWrapLow(step1[4] + step1[5], bd); + step2[5] = HighbdWrapLow(step1[4] - step1[5], bd); + step2[6] = HighbdWrapLow(-step1[6] + step1[7], bd); + step2[7] = HighbdWrapLow(step1[6] + step1[7], bd); + + step2[8] = step1[8]; + step2[15] = step1[15]; + temp1 = -step1[9] * (long)CosPi8_64 + step1[14] * (long)CosPi24_64; + temp2 = step1[9] * (long)CosPi24_64 + step1[14] * (long)CosPi8_64; + step2[9] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[14] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = -step1[10] * (long)CosPi24_64 - step1[13] * (long)CosPi8_64; + temp2 = -step1[10] * (long)CosPi8_64 + step1[13] * (long)CosPi24_64; + step2[10] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[13] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + step2[11] = step1[11]; + step2[12] = step1[12]; + + step2[16] = HighbdWrapLow(step1[16] + step1[19], bd); + step2[17] = HighbdWrapLow(step1[17] + step1[18], bd); + step2[18] = HighbdWrapLow(step1[17] - step1[18], bd); + step2[19] = HighbdWrapLow(step1[16] - step1[19], bd); + step2[20] = HighbdWrapLow(-step1[20] + step1[23], bd); + step2[21] = HighbdWrapLow(-step1[21] + step1[22], bd); + step2[22] = HighbdWrapLow(step1[21] + step1[22], bd); + step2[23] = HighbdWrapLow(step1[20] + step1[23], bd); + + step2[24] = HighbdWrapLow(step1[24] + step1[27], bd); + step2[25] = HighbdWrapLow(step1[25] + step1[26], bd); + step2[26] = HighbdWrapLow(step1[25] - step1[26], bd); + step2[27] = HighbdWrapLow(step1[24] - step1[27], bd); + step2[28] = HighbdWrapLow(-step1[28] + step1[31], bd); + step2[29] = HighbdWrapLow(-step1[29] + step1[30], bd); + step2[30] = HighbdWrapLow(step1[29] + step1[30], bd); + step2[31] = HighbdWrapLow(step1[28] + step1[31], bd); + + // stage 5 + step1[0] = HighbdWrapLow(step2[0] + step2[3], bd); + step1[1] = HighbdWrapLow(step2[1] + step2[2], bd); + step1[2] = HighbdWrapLow(step2[1] - step2[2], bd); + step1[3] = HighbdWrapLow(step2[0] - step2[3], bd); + step1[4] = step2[4]; + temp1 = (step2[6] - step2[5]) * (long)CosPi16_64; + temp2 = (step2[5] + step2[6]) * (long)CosPi16_64; + step1[5] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[6] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + step1[7] = step2[7]; + + step1[8] = HighbdWrapLow(step2[8] + step2[11], bd); + step1[9] = HighbdWrapLow(step2[9] + step2[10], bd); + step1[10] = HighbdWrapLow(step2[9] - step2[10], bd); + step1[11] = HighbdWrapLow(step2[8] - step2[11], bd); + step1[12] = HighbdWrapLow(-step2[12] + step2[15], bd); + step1[13] = HighbdWrapLow(-step2[13] + step2[14], bd); + step1[14] = HighbdWrapLow(step2[13] + step2[14], bd); + step1[15] = HighbdWrapLow(step2[12] + step2[15], bd); + + step1[16] = step2[16]; + step1[17] = step2[17]; + temp1 = -step2[18] * (long)CosPi8_64 + step2[29] * (long)CosPi24_64; + temp2 = step2[18] * (long)CosPi24_64 + step2[29] * (long)CosPi8_64; + step1[18] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[29] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = -step2[19] * (long)CosPi8_64 + step2[28] * (long)CosPi24_64; + temp2 = step2[19] * (long)CosPi24_64 + step2[28] * (long)CosPi8_64; + step1[19] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[28] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = -step2[20] * (long)CosPi24_64 - step2[27] * (long)CosPi8_64; + temp2 = -step2[20] * (long)CosPi8_64 + step2[27] * (long)CosPi24_64; + step1[20] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[27] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = -step2[21] * (long)CosPi24_64 - step2[26] * (long)CosPi8_64; + temp2 = -step2[21] * (long)CosPi8_64 + step2[26] * (long)CosPi24_64; + step1[21] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[26] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + step1[22] = step2[22]; + step1[23] = step2[23]; + step1[24] = step2[24]; + step1[25] = step2[25]; + step1[30] = step2[30]; + step1[31] = step2[31]; + + // stage 6 + step2[0] = HighbdWrapLow(step1[0] + step1[7], bd); + step2[1] = HighbdWrapLow(step1[1] + step1[6], bd); + step2[2] = HighbdWrapLow(step1[2] + step1[5], bd); + step2[3] = HighbdWrapLow(step1[3] + step1[4], bd); + step2[4] = HighbdWrapLow(step1[3] - step1[4], bd); + step2[5] = HighbdWrapLow(step1[2] - step1[5], bd); + step2[6] = HighbdWrapLow(step1[1] - step1[6], bd); + step2[7] = HighbdWrapLow(step1[0] - step1[7], bd); + step2[8] = step1[8]; + step2[9] = step1[9]; + temp1 = (-step1[10] + step1[13]) * (long)CosPi16_64; + temp2 = (step1[10] + step1[13]) * (long)CosPi16_64; + step2[10] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[13] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = (-step1[11] + step1[12]) * (long)CosPi16_64; + temp2 = (step1[11] + step1[12]) * (long)CosPi16_64; + step2[11] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step2[12] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + step2[14] = step1[14]; + step2[15] = step1[15]; + + step2[16] = HighbdWrapLow(step1[16] + step1[23], bd); + step2[17] = HighbdWrapLow(step1[17] + step1[22], bd); + step2[18] = HighbdWrapLow(step1[18] + step1[21], bd); + step2[19] = HighbdWrapLow(step1[19] + step1[20], bd); + step2[20] = HighbdWrapLow(step1[19] - step1[20], bd); + step2[21] = HighbdWrapLow(step1[18] - step1[21], bd); + step2[22] = HighbdWrapLow(step1[17] - step1[22], bd); + step2[23] = HighbdWrapLow(step1[16] - step1[23], bd); + + step2[24] = HighbdWrapLow(-step1[24] + step1[31], bd); + step2[25] = HighbdWrapLow(-step1[25] + step1[30], bd); + step2[26] = HighbdWrapLow(-step1[26] + step1[29], bd); + step2[27] = HighbdWrapLow(-step1[27] + step1[28], bd); + step2[28] = HighbdWrapLow(step1[27] + step1[28], bd); + step2[29] = HighbdWrapLow(step1[26] + step1[29], bd); + step2[30] = HighbdWrapLow(step1[25] + step1[30], bd); + step2[31] = HighbdWrapLow(step1[24] + step1[31], bd); + + // stage 7 + step1[0] = HighbdWrapLow(step2[0] + step2[15], bd); + step1[1] = HighbdWrapLow(step2[1] + step2[14], bd); + step1[2] = HighbdWrapLow(step2[2] + step2[13], bd); + step1[3] = HighbdWrapLow(step2[3] + step2[12], bd); + step1[4] = HighbdWrapLow(step2[4] + step2[11], bd); + step1[5] = HighbdWrapLow(step2[5] + step2[10], bd); + step1[6] = HighbdWrapLow(step2[6] + step2[9], bd); + step1[7] = HighbdWrapLow(step2[7] + step2[8], bd); + step1[8] = HighbdWrapLow(step2[7] - step2[8], bd); + step1[9] = HighbdWrapLow(step2[6] - step2[9], bd); + step1[10] = HighbdWrapLow(step2[5] - step2[10], bd); + step1[11] = HighbdWrapLow(step2[4] - step2[11], bd); + step1[12] = HighbdWrapLow(step2[3] - step2[12], bd); + step1[13] = HighbdWrapLow(step2[2] - step2[13], bd); + step1[14] = HighbdWrapLow(step2[1] - step2[14], bd); + step1[15] = HighbdWrapLow(step2[0] - step2[15], bd); + + step1[16] = step2[16]; + step1[17] = step2[17]; + step1[18] = step2[18]; + step1[19] = step2[19]; + temp1 = (-step2[20] + step2[27]) * (long)CosPi16_64; + temp2 = (step2[20] + step2[27]) * (long)CosPi16_64; + step1[20] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[27] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = (-step2[21] + step2[26]) * (long)CosPi16_64; + temp2 = (step2[21] + step2[26]) * (long)CosPi16_64; + step1[21] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[26] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = (-step2[22] + step2[25]) * (long)CosPi16_64; + temp2 = (step2[22] + step2[25]) * (long)CosPi16_64; + step1[22] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[25] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + temp1 = (-step2[23] + step2[24]) * (long)CosPi16_64; + temp2 = (step2[23] + step2[24]) * (long)CosPi16_64; + step1[23] = HighbdWrapLow(DctConstRoundShift(temp1), bd); + step1[24] = HighbdWrapLow(DctConstRoundShift(temp2), bd); + step1[28] = step2[28]; + step1[29] = step2[29]; + step1[30] = step2[30]; + step1[31] = step2[31]; + + // final stage + output[0] = HighbdWrapLow(step1[0] + step1[31], bd); + output[1] = HighbdWrapLow(step1[1] + step1[30], bd); + output[2] = HighbdWrapLow(step1[2] + step1[29], bd); + output[3] = HighbdWrapLow(step1[3] + step1[28], bd); + output[4] = HighbdWrapLow(step1[4] + step1[27], bd); + output[5] = HighbdWrapLow(step1[5] + step1[26], bd); + output[6] = HighbdWrapLow(step1[6] + step1[25], bd); + output[7] = HighbdWrapLow(step1[7] + step1[24], bd); + output[8] = HighbdWrapLow(step1[8] + step1[23], bd); + output[9] = HighbdWrapLow(step1[9] + step1[22], bd); + output[10] = HighbdWrapLow(step1[10] + step1[21], bd); + output[11] = HighbdWrapLow(step1[11] + step1[20], bd); + output[12] = HighbdWrapLow(step1[12] + step1[19], bd); + output[13] = HighbdWrapLow(step1[13] + step1[18], bd); + output[14] = HighbdWrapLow(step1[14] + step1[17], bd); + output[15] = HighbdWrapLow(step1[15] + step1[16], bd); + output[16] = HighbdWrapLow(step1[15] - step1[16], bd); + output[17] = HighbdWrapLow(step1[14] - step1[17], bd); + output[18] = HighbdWrapLow(step1[13] - step1[18], bd); + output[19] = HighbdWrapLow(step1[12] - step1[19], bd); + output[20] = HighbdWrapLow(step1[11] - step1[20], bd); + output[21] = HighbdWrapLow(step1[10] - step1[21], bd); + output[22] = HighbdWrapLow(step1[9] - step1[22], bd); + output[23] = HighbdWrapLow(step1[8] - step1[23], bd); + output[24] = HighbdWrapLow(step1[7] - step1[24], bd); + output[25] = HighbdWrapLow(step1[6] - step1[25], bd); + output[26] = HighbdWrapLow(step1[5] - step1[26], bd); + output[27] = HighbdWrapLow(step1[4] - step1[27], bd); + output[28] = HighbdWrapLow(step1[3] - step1[28], bd); + output[29] = HighbdWrapLow(step1[2] - step1[29], bd); + output[30] = HighbdWrapLow(step1[1] - step1[30], bd); + output[31] = HighbdWrapLow(step1[0] - step1[31], bd); + } + + public static void HighbdIdct32x321024Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i, j; + Span<int> output = stackalloc int[32 * 32]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[32]; + Span<int> tempOut = stackalloc int[32]; + + // Rows + for (i = 0; i < 32; ++i) + { + int zeroCoeff = 0; + for (j = 0; j < 32; ++j) + { + zeroCoeff |= input[j]; + } + + if (zeroCoeff != 0) + { + HighbdIdct32(input, outptr, bd); + } + else + { + outptr.Slice(0, 32).Fill(0); + } + + input = input.Slice(32); + outptr = outptr.Slice(32); + } + + // Columns + for (i = 0; i < 32; ++i) + { + for (j = 0; j < 32; ++j) + { + tempIn[j] = output[j * 32 + i]; + } + + HighbdIdct32(tempIn, tempOut, bd); + for (j = 0; j < 32; ++j) + { + dest[j * stride + i] = HighbdClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 6), bd); + } + } + } + + public static void HighbdIdct32x32135Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i, j; + Span<int> output = stackalloc int[32 * 32]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[32]; + Span<int> tempOut = stackalloc int[32]; + + // Rows + // Only upper-left 16x16 has non-zero coeff + for (i = 0; i < 16; ++i) + { + HighbdIdct32(input, outptr, bd); + input = input.Slice(32); + outptr = outptr.Slice(32); + } + + // Columns + for (i = 0; i < 32; ++i) + { + Span<ushort> destT = dest; + for (j = 0; j < 32; ++j) + { + tempIn[j] = output[j * 32 + i]; + } + + HighbdIdct32(tempIn, tempOut, bd); + for (j = 0; j < 32; ++j) + { + destT[i] = HighbdClipPixelAdd(destT[i], BitUtils.RoundPowerOfTwo(tempOut[j], 6), bd); + destT = destT.Slice(stride); + } + } + } + + public static void HighbdIdct32x3234Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i, j; + Span<int> output = stackalloc int[32 * 32]; + Span<int> outptr = output; + Span<int> tempIn = stackalloc int[32]; + Span<int> tempOut = stackalloc int[32]; + + // Rows + // Only upper-left 8x8 has non-zero coeff + for (i = 0; i < 8; ++i) + { + HighbdIdct32(input, outptr, bd); + input = input.Slice(32); + outptr = outptr.Slice(32); + } + + // Columns + for (i = 0; i < 32; ++i) + { + for (j = 0; j < 32; ++j) + { + tempIn[j] = output[j * 32 + i]; + } + + HighbdIdct32(tempIn, tempOut, bd); + for (j = 0; j < 32; ++j) + { + dest[j * stride + i] = HighbdClipPixelAdd(dest[j * stride + i], BitUtils.RoundPowerOfTwo(tempOut[j], 6), bd); + } + } + } + + public static void HighbdIdct32x321Add(ReadOnlySpan<int> input, Span<ushort> dest, int stride, int bd) + { + int i, j; + int a1; + int output = HighbdWrapLow(DctConstRoundShift(input[0] * (long)CosPi16_64), bd); + + output = HighbdWrapLow(DctConstRoundShift(output * (long)CosPi16_64), bd); + a1 = BitUtils.RoundPowerOfTwo(output, 6); + + for (j = 0; j < 32; ++j) + { + for (i = 0; i < 32; ++i) + { + dest[i] = HighbdClipPixelAdd(dest[i], a1, bd); + } + + dest = dest.Slice(stride); + } + } + } +} |