Skip to content

Add ComplexDotProduct to SVE microbenchmark #5045

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Add ComplexDotProduct to SVE microbenchmark #5045

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
285 changes: 285 additions & 0 deletions src/benchmarks/micro/sve/ComplexDotProduct.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,285 @@
using System;
using System.Diagnostics;
using System.Numerics;
using System.Runtime.Intrinsics;
using System.Runtime.Intrinsics.Arm;
using BenchmarkDotNet.Attributes;
using BenchmarkDotNet.Extensions;
using BenchmarkDotNet.Configs;
using BenchmarkDotNet.Filters;
using MicroBenchmarks;

namespace SveBenchmarks
{
[BenchmarkCategory(Categories.Runtime)]
[OperatingSystemsArchitectureFilter(allowed: true, System.Runtime.InteropServices.Architecture.Arm64)]
[Config(typeof(Config))]
public class ComplexDotProduct
{
private class Config : ManualConfig
{
public Config()
{
AddFilter(new SimpleFilter(_ => Sve2.IsSupported));
}
}

[Params(15, 127, 527, 10015)]
public int Size;

private sbyte[] _input1;
private sbyte[] _input2;
private int[] _output;

[GlobalSetup]
public virtual void Setup()
{
_input1 = new sbyte[Size * 4];
_input2 = new sbyte[Size * 4];

sbyte[] vals = ValuesGenerator.Array<sbyte>(Size * 8);
for (int i = 0; i < Size * 4; i++)
{
_input1[i] = vals[i];
_input2[i] = vals[Size * 4 + i];
}

_output = new int[Size * 2];
}

[GlobalCleanup]
public virtual void Verify()
{
int[] current = (int[])_output.Clone();
Setup();
Scalar();
int[] scalar = (int[])_output.Clone();

// Check that the result is the same as scalar.
for (int i = 0; i < Size; i++)
{
Debug.Assert(current[i] == scalar[i]);
}
}

// The following algorithms are adapted from the Arm simd-loops repository:
// https://gitlab.arm.com/architecture/simd-loops/-/blob/main/loops/loop_110.c

[Benchmark]
public unsafe void Scalar()
{
fixed (sbyte* a = _input1, b = _input2)
fixed (int* c = _output)
{
for (int i = 0; i < Size; i++)
{
// Real part.
c[i * 2] = (int)(((a[4 * i] * b[4 * i]) - (a[4 * i + 1] * b[4 * i + 1])) +
((a[4 * i + 2] * b[4 * i + 2]) - (a[4 * i + 3] * b[4 * i + 3])));
// Imaginary part.
c[i * 2 + 1] = (int)(((a[4 * i + 1] * b[4 * i]) + (a[4 * i] * b[4 * i + 1])) +
((a[4 * i + 3] * b[4 * i + 2]) + (a[4 * i + 2] * b[4 * i + 3])));
}
}
}

[Benchmark]
public unsafe void Vector128ComplexDotProduct()
{
fixed (sbyte* a = _input1, b = _input2)
fixed (int* c = _output)
{

// Helper function for calculating dot product.
Func<Vector128<int>, Vector128<int>, Vector128<int>, Vector128<int>,
Vector128<int>, Vector128<int>, Vector128<int>, Vector128<int>,
(Vector128<int>, Vector128<int>)> dotProduct = (a0, a1, a2, a3, b0, b1, b2, b3) => {

Vector128<int> cr = Vector128<int>.Zero;
Vector128<int> ci = Vector128<int>.Zero;

// Compute dot product of the real part.
cr = AdvSimd.MultiplyAdd(cr, a0, b0);
cr = AdvSimd.MultiplySubtract(cr, a1, b1);
cr = AdvSimd.MultiplyAdd(cr, a2, b2);
cr = AdvSimd.MultiplySubtract(cr, a3, b3);
// Compute dot product of the imaginary part.
ci = AdvSimd.MultiplyAdd(ci, a0, b1);
ci = AdvSimd.MultiplyAdd(ci, a1, b0);
ci = AdvSimd.MultiplyAdd(ci, a2, b3);
ci = AdvSimd.MultiplyAdd(ci, a3, b2);

return (cr, ci);
};

int i = 0;
int lmt = Size - 16;
for (; i <= lmt; i += 16)
{
// Load inputs as sbytes.
(Vector128<sbyte> a0, Vector128<sbyte> a1, Vector128<sbyte> a2, Vector128<sbyte> a3) = AdvSimd.Arm64.Load4xVector128AndUnzip(a + 4 * i);
(Vector128<sbyte> b0, Vector128<sbyte> b1, Vector128<sbyte> b2, Vector128<sbyte> b3) = AdvSimd.Arm64.Load4xVector128AndUnzip(b + 4 * i);
Vector128<int> cr, ci;

// Extend sbyte to short for each input component.
Vector128<short> a0l = AdvSimd.SignExtendWideningLower(a0.GetLower());
Vector128<short> a0h = AdvSimd.SignExtendWideningUpper(a0);
Vector128<short> a1l = AdvSimd.SignExtendWideningLower(a1.GetLower());
Vector128<short> a1h = AdvSimd.SignExtendWideningUpper(a1);
Vector128<short> a2l = AdvSimd.SignExtendWideningLower(a2.GetLower());
Vector128<short> a2h = AdvSimd.SignExtendWideningUpper(a2);
Vector128<short> a3l = AdvSimd.SignExtendWideningLower(a3.GetLower());
Vector128<short> a3h = AdvSimd.SignExtendWideningUpper(a3);
Vector128<short> b0l = AdvSimd.SignExtendWideningLower(b0.GetLower());
Vector128<short> b0h = AdvSimd.SignExtendWideningUpper(b0);
Vector128<short> b1l = AdvSimd.SignExtendWideningLower(b1.GetLower());
Vector128<short> b1h = AdvSimd.SignExtendWideningUpper(b1);
Vector128<short> b2l = AdvSimd.SignExtendWideningLower(b2.GetLower());
Vector128<short> b2h = AdvSimd.SignExtendWideningUpper(b2);
Vector128<short> b3l = AdvSimd.SignExtendWideningLower(b3.GetLower());
Vector128<short> b3h = AdvSimd.SignExtendWideningUpper(b3);

// Compute the lower half of the lower half.
Vector128<int> a0ll = AdvSimd.SignExtendWideningLower(a0l.GetLower());
Vector128<int> a1ll = AdvSimd.SignExtendWideningLower(a1l.GetLower());
Vector128<int> a2ll = AdvSimd.SignExtendWideningLower(a2l.GetLower());
Vector128<int> a3ll = AdvSimd.SignExtendWideningLower(a3l.GetLower());
Vector128<int> b0ll = AdvSimd.SignExtendWideningLower(b0l.GetLower());
Vector128<int> b1ll = AdvSimd.SignExtendWideningLower(b1l.GetLower());
Vector128<int> b2ll = AdvSimd.SignExtendWideningLower(b2l.GetLower());
Vector128<int> b3ll = AdvSimd.SignExtendWideningLower(b3l.GetLower());
(cr, ci) = dotProduct(a0ll, a1ll, a2ll, a3ll, b0ll, b1ll, b2ll, b3ll);
AdvSimd.Arm64.StoreVectorAndZip(c + 2 * i, (cr, ci));

// Compute the upper half of the lower half.
Vector128<int> a0lh = AdvSimd.SignExtendWideningUpper(a0l);
Vector128<int> a1lh = AdvSimd.SignExtendWideningUpper(a1l);
Vector128<int> a2lh = AdvSimd.SignExtendWideningUpper(a2l);
Vector128<int> a3lh = AdvSimd.SignExtendWideningUpper(a3l);
Vector128<int> b0lh = AdvSimd.SignExtendWideningUpper(b0l);
Vector128<int> b1lh = AdvSimd.SignExtendWideningUpper(b1l);
Vector128<int> b2lh = AdvSimd.SignExtendWideningUpper(b2l);
Vector128<int> b3lh = AdvSimd.SignExtendWideningUpper(b3l);
(cr, ci) = dotProduct(a0lh, a1lh, a2lh, a3lh, b0lh, b1lh, b2lh, b3lh);
AdvSimd.Arm64.StoreVectorAndZip(c + 2 * i + 8, (cr, ci));

// Compute the lower half of the upper half.
Vector128<int> a0hl = AdvSimd.SignExtendWideningLower(a0h.GetLower());
Vector128<int> a1hl = AdvSimd.SignExtendWideningLower(a1h.GetLower());
Vector128<int> a2hl = AdvSimd.SignExtendWideningLower(a2h.GetLower());
Vector128<int> a3hl = AdvSimd.SignExtendWideningLower(a3h.GetLower());
Vector128<int> b0hl = AdvSimd.SignExtendWideningLower(b0h.GetLower());
Vector128<int> b1hl = AdvSimd.SignExtendWideningLower(b1h.GetLower());
Vector128<int> b2hl = AdvSimd.SignExtendWideningLower(b2h.GetLower());
Vector128<int> b3hl = AdvSimd.SignExtendWideningLower(b3h.GetLower());
(cr, ci) = dotProduct(a0hl, a1hl, a2hl, a3hl, b0hl, b1hl, b2hl, b3hl);
AdvSimd.Arm64.StoreVectorAndZip(c + 2 * i + 16, (cr, ci));

// Compute the upper half of the upper half.
Vector128<int> a0hh = AdvSimd.SignExtendWideningUpper(a0h);
Vector128<int> a1hh = AdvSimd.SignExtendWideningUpper(a1h);
Vector128<int> a2hh = AdvSimd.SignExtendWideningUpper(a2h);
Vector128<int> a3hh = AdvSimd.SignExtendWideningUpper(a3h);
Vector128<int> b0hh = AdvSimd.SignExtendWideningUpper(b0h);
Vector128<int> b1hh = AdvSimd.SignExtendWideningUpper(b1h);
Vector128<int> b2hh = AdvSimd.SignExtendWideningUpper(b2h);
Vector128<int> b3hh = AdvSimd.SignExtendWideningUpper(b3h);
(cr, ci) = dotProduct(a0hh, a1hh, a2hh, a3hh, b0hh, b1hh, b2hh, b3hh);
AdvSimd.Arm64.StoreVectorAndZip(c + 2 * i + 24, (cr, ci));

}
// Handle remaining elements.
for (; i < Size; i++)
{
// Real part.
c[i * 2] = (int)(((a[4 * i] * b[4 * i]) - (a[4 * i + 1] * b[4 * i + 1])) +
((a[4 * i + 2] * b[4 * i + 2]) - (a[4 * i + 3] * b[4 * i + 3])));
// Imaginary part.
c[i * 2 + 1] = (int)(((a[4 * i + 1] * b[4 * i]) + (a[4 * i] * b[4 * i + 1])) +
((a[4 * i + 3] * b[4 * i + 2]) + (a[4 * i + 2] * b[4 * i + 3])));
}
}
}

[Benchmark]
public unsafe void SveComplexDotProduct()
{
fixed (sbyte* a = _input1, b = _input2)
fixed (int* c = _output)
{
int i = 0;
int cntw = (int)Sve.Count32BitElements();

// Create mask for the imaginary half of a word.
Vector<short> imMask = (Vector<short>)(new Vector<uint>(0xFFFF0000u));
Vector<int> pTrue = Sve.CreateTrueMaskInt32();
Vector<int> pLoop = (Vector<int>)Sve.CreateWhileLessThanMask32Bit(0, Size);
while (Sve.TestFirstTrue(pTrue, pLoop))
{
// Load inputs.
Vector<sbyte> a1 = (Vector<sbyte>)Sve.LoadVector(pLoop, (int*)(a + 4 * i));
Vector<sbyte> b1 = (Vector<sbyte>)Sve.LoadVector(pLoop, (int*)(b + 4 * i));

// Unpack and extend the lower and upper halves to short.
Vector<short> a1l = Sve.SignExtendWideningLower(a1);
Vector<short> a1h = Sve.SignExtendWideningUpper(a1);
Vector<short> b1l = Sve.SignExtendWideningLower(b1);
Vector<short> b1h = Sve.SignExtendWideningUpper(b1);

// Negate the imaginary components.
Vector<short> b1l_re = Sve.ConditionalSelect(imMask, Sve.Negate(b1l), b1l);
Vector<short> b1h_re = Sve.ConditionalSelect(imMask, Sve.Negate(b1h), b1h);
// Swap the real and imaginary components from each word.
Vector<short> b1l_im = (Vector<short>)Sve.ReverseElement16((Vector<int>)b1l);
Vector<short> b1h_im = (Vector<short>)Sve.ReverseElement16((Vector<int>)b1h);

// Compute dot products (real and imaginary, low and high bits).
Vector<long> c1l_re = Sve.DotProduct(Vector<long>.Zero, a1l, b1l_re);
Vector<long> c1h_re = Sve.DotProduct(Vector<long>.Zero, a1h, b1h_re);
Vector<long> c1l_im = Sve.DotProduct(Vector<long>.Zero, a1l, b1l_im);
Vector<long> c1h_im = Sve.DotProduct(Vector<long>.Zero, a1h, b1h_im);

// Combine low and high parts back together.
Vector<int> cr = Sve.UnzipEven((Vector<int>)c1l_re, (Vector<int>)c1h_re);
Vector<int> ci = Sve.UnzipEven((Vector<int>)c1l_im, (Vector<int>)c1h_im);

// Interleaved store real and imaginary parts of the result.
Sve.StoreAndZip(pLoop, c + 2 * i, (cr, ci));

// Handle loop.
i += cntw;
pLoop = (Vector<int>)Sve.CreateWhileLessThanMask32Bit(i, Size);
}
}
}

[Benchmark]
public unsafe void Sve2ComplexDotProduct()
{
fixed (sbyte* a = _input1, b = _input2)
fixed (int* c = _output)
{
int i = 0;
int cntw = (int)Sve.Count32BitElements();

Vector<int> pTrue = Sve.CreateTrueMaskInt32();
Vector<int> pLoop = (Vector<int>)Sve.CreateWhileLessThanMask32Bit(0, Size);
while (Sve.TestFirstTrue(pTrue, pLoop))
{
Vector<sbyte> a1 = (Vector<sbyte>)Sve.LoadVector(pLoop, (int*)(a + 4 * i));
Vector<sbyte> b1 = (Vector<sbyte>)Sve.LoadVector(pLoop, (int*)(b + 4 * i));

Vector<int> cr = Sve2.DotProductRotateComplex(Vector<int>.Zero, a1, b1, 0);
Vector<int> ci = Sve2.DotProductRotateComplex(Vector<int>.Zero, a1, b1, 1);

Sve.StoreAndZip(pLoop, c + 2 * i, (cr, ci));

// Handle loop.
i += cntw;
pLoop = (Vector<int>)Sve.CreateWhileLessThanMask32Bit(i, Size);
}
}
}

}
}
Loading