Additional cleanup and acceleration of the System.Numerics vector and related types by tannergooding · Pull Request #103527 · dotnet/runtime

tannergooding · 2024-06-15T20:14:49Z

Similar overall improvements to the PR this is continuing from. Most remain identical to what they were previously:

Method	Toolchain	Mean	Error	StdDev	Median	Min	Max	Ratio	RatioSD
Plane.DotCoordinateBenchmark	runtime-base	0.7257 ns	0.0085 ns	0.0075 ns	0.7241 ns	0.7169 ns	0.7398 ns	1.00	0.01
Plane.DotCoordinateBenchmark	runtime	0.5491 ns	0.0197 ns	0.0184 ns	0.5382 ns	0.5358 ns	0.5879 ns	0.76	0.03

Plane.NormalizeBenchmark	runtime-base	1.1018 ns	0.0207 ns	0.0193 ns	1.0922 ns	1.0873 ns	1.1382 ns	1.00	0.02
Plane.NormalizeBenchmark	runtime	1.1102 ns	0.0296 ns	0.0277 ns	1.0938 ns	1.0860 ns	1.1548 ns	1.01	0.03

Plane.TransformByMatrix4x4Benchmark	runtime	11.3975 ns	0.0586 ns	0.0519 ns	11.4152 ns	11.3087 ns	11.4736 ns	1.00	0.01
Plane.TransformByMatrix4x4Benchmark	runtime-base	9.5247 ns	0.0415 ns	0.0368 ns	9.5240 ns	9.4585 ns	9.5803 ns	0.84	0.00

Quaternion.DivisionOperatorBenchmark	runtime-base	4.1745 ns	0.0149 ns	0.0132 ns	4.1728 ns	4.1580 ns	4.2009 ns	1.00	0.00
Quaternion.DivisionOperatorBenchmark	runtime	1.2661 ns	0.0039 ns	0.0030 ns	1.2660 ns	1.2628 ns	1.2734 ns	0.30	0.00

Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime-base	2.2240 ns	0.0162 ns	0.0144 ns	2.2184 ns	2.2091 ns	2.2562 ns	1.00	0.01
Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime	0.5656 ns	0.0232 ns	0.0217 ns	0.5499 ns	0.5478 ns	0.6062 ns	0.25	0.01

Quaternion.ConcatenateBenchmark	runtime-base	3.1269 ns	0.0402 ns	0.0376 ns	3.1064 ns	3.0917 ns	3.1953 ns	1.00	0.02
Quaternion.ConcatenateBenchmark	runtime	0.5663 ns	0.0227 ns	0.0213 ns	0.5633 ns	0.5338 ns	0.6047 ns	0.18	0.01

Quaternion.ConjugateBenchmark	runtime-base	0.1936 ns	0.0124 ns	0.0116 ns	0.1857 ns	0.1818 ns	0.2111 ns	1.00	0.08
Quaternion.ConjugateBenchmark	runtime	0.1968 ns	0.0162 ns	0.0151 ns	0.1878 ns	0.1837 ns	0.2265 ns	1.02	0.10

Quaternion.CreateFromAxisAngleBenchmark	runtime-base	7.7296 ns	0.0161 ns	0.0151 ns	7.7302 ns	7.6947 ns	7.7572 ns	1.00	0.00
Quaternion.CreateFromAxisAngleBenchmark	runtime	7.4446 ns	0.0478 ns	0.0447 ns	7.4217 ns	7.4048 ns	7.5371 ns	0.96	0.01

Quaternion.InverseBenchmark	runtime-base	0.9127 ns	0.0110 ns	0.0103 ns	0.9101 ns	0.9014 ns	0.9351 ns	1.00	0.02
Quaternion.InverseBenchmark	runtime	1.2364 ns	0.1370 ns	0.1577 ns	1.1716 ns	1.0853 ns	1.5973 ns	1.35	0.17

Quaternion.LerpBenchmark	runtime-base	5.0949 ns	0.0332 ns	0.0310 ns	5.0914 ns	5.0590 ns	5.1670 ns	1.00	0.01
Quaternion.LerpBenchmark	runtime	1.6378 ns	0.0297 ns	0.0278 ns	1.6312 ns	1.6031 ns	1.6804 ns	0.32	0.01

Quaternion.MultiplyByQuaternionBenchmark	runtime-base	2.2071 ns	0.0051 ns	0.0042 ns	2.2056 ns	2.2036 ns	2.2187 ns	1.00	0.00
Quaternion.MultiplyByQuaternionBenchmark	runtime	0.5629 ns	0.0293 ns	0.0274 ns	0.5437 ns	0.5400 ns	0.6083 ns	0.26	0.01

Quaternion.SlerpBenchmark	runtime-base	5.1270 ns	0.0144 ns	0.0135 ns	5.1238 ns	5.1027 ns	5.1533 ns	1.00	0.00
Quaternion.SlerpBenchmark	runtime	1.6400 ns	0.0270 ns	0.0252 ns	1.6321 ns	1.6087 ns	1.6817 ns	0.32	0.00

Vector3.Cross(Vector3 vector1, Vector3 vector2) Vector3.Transform(Vector3 value, Quaternion rotation)

added simd paths for quaternion multiplication, division and concatenate added simd path for matrix4x4 multiplication

dotnet-policy-service · 2024-06-15T20:15:25Z

Tagging subscribers to this area: @dotnet/area-system-numerics
See info in area-owners.md if you want to be subscribed.

…rix types

tannergooding · 2024-06-15T20:14:49Z

Similar overall improvements to the PR this is continuing from. Most remain identical to what they were previously:

Method	Toolchain	Mean	Error	StdDev	Median	Min	Max	Ratio	RatioSD
Plane.DotCoordinateBenchmark	runtime-base	0.7257 ns	0.0085 ns	0.0075 ns	0.7241 ns	0.7169 ns	0.7398 ns	1.00	0.01
Plane.DotCoordinateBenchmark	runtime	0.5491 ns	0.0197 ns	0.0184 ns	0.5382 ns	0.5358 ns	0.5879 ns	0.76	0.03

Plane.NormalizeBenchmark	runtime-base	1.1018 ns	0.0207 ns	0.0193 ns	1.0922 ns	1.0873 ns	1.1382 ns	1.00	0.02
Plane.NormalizeBenchmark	runtime	1.1102 ns	0.0296 ns	0.0277 ns	1.0938 ns	1.0860 ns	1.1548 ns	1.01	0.03

Plane.TransformByMatrix4x4Benchmark	runtime	11.3975 ns	0.0586 ns	0.0519 ns	11.4152 ns	11.3087 ns	11.4736 ns	1.00	0.01
Plane.TransformByMatrix4x4Benchmark	runtime-base	9.5247 ns	0.0415 ns	0.0368 ns	9.5240 ns	9.4585 ns	9.5803 ns	0.84	0.00

Quaternion.DivisionOperatorBenchmark	runtime-base	4.1745 ns	0.0149 ns	0.0132 ns	4.1728 ns	4.1580 ns	4.2009 ns	1.00	0.00
Quaternion.DivisionOperatorBenchmark	runtime	1.2661 ns	0.0039 ns	0.0030 ns	1.2660 ns	1.2628 ns	1.2734 ns	0.30	0.00

Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime-base	2.2240 ns	0.0162 ns	0.0144 ns	2.2184 ns	2.2091 ns	2.2562 ns	1.00	0.01
Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime	0.5656 ns	0.0232 ns	0.0217 ns	0.5499 ns	0.5478 ns	0.6062 ns	0.25	0.01

Quaternion.ConcatenateBenchmark	runtime-base	3.1269 ns	0.0402 ns	0.0376 ns	3.1064 ns	3.0917 ns	3.1953 ns	1.00	0.02
Quaternion.ConcatenateBenchmark	runtime	0.5663 ns	0.0227 ns	0.0213 ns	0.5633 ns	0.5338 ns	0.6047 ns	0.18	0.01

Quaternion.ConjugateBenchmark	runtime-base	0.1936 ns	0.0124 ns	0.0116 ns	0.1857 ns	0.1818 ns	0.2111 ns	1.00	0.08
Quaternion.ConjugateBenchmark	runtime	0.1968 ns	0.0162 ns	0.0151 ns	0.1878 ns	0.1837 ns	0.2265 ns	1.02	0.10

Quaternion.CreateFromAxisAngleBenchmark	runtime-base	7.7296 ns	0.0161 ns	0.0151 ns	7.7302 ns	7.6947 ns	7.7572 ns	1.00	0.00
Quaternion.CreateFromAxisAngleBenchmark	runtime	7.4446 ns	0.0478 ns	0.0447 ns	7.4217 ns	7.4048 ns	7.5371 ns	0.96	0.01

Quaternion.InverseBenchmark	runtime-base	0.9127 ns	0.0110 ns	0.0103 ns	0.9101 ns	0.9014 ns	0.9351 ns	1.00	0.02
Quaternion.InverseBenchmark	runtime	1.2364 ns	0.1370 ns	0.1577 ns	1.1716 ns	1.0853 ns	1.5973 ns	1.35	0.17

Quaternion.LerpBenchmark	runtime-base	5.0949 ns	0.0332 ns	0.0310 ns	5.0914 ns	5.0590 ns	5.1670 ns	1.00	0.01
Quaternion.LerpBenchmark	runtime	1.6378 ns	0.0297 ns	0.0278 ns	1.6312 ns	1.6031 ns	1.6804 ns	0.32	0.01

Quaternion.MultiplyByQuaternionBenchmark	runtime-base	2.2071 ns	0.0051 ns	0.0042 ns	2.2056 ns	2.2036 ns	2.2187 ns	1.00	0.00
Quaternion.MultiplyByQuaternionBenchmark	runtime	0.5629 ns	0.0293 ns	0.0274 ns	0.5437 ns	0.5400 ns	0.6083 ns	0.26	0.01

Quaternion.SlerpBenchmark	runtime-base	5.1270 ns	0.0144 ns	0.0135 ns	5.1238 ns	5.1027 ns	5.1533 ns	1.00	0.00
Quaternion.SlerpBenchmark	runtime	1.6400 ns	0.0270 ns	0.0252 ns	1.6321 ns	1.6087 ns	1.6817 ns	0.32	0.00

Vector3.Cross(Vector3 vector1, Vector3 vector2) Vector3.Transform(Vector3 value, Quaternion rotation)

added simd paths for quaternion multiplication, division and concatenate added simd path for matrix4x4 multiplication

dotnet-policy-service · 2024-06-15T20:15:25Z

Tagging subscribers to this area: @dotnet/area-system-numerics
See info in area-owners.md if you want to be subscribed.

…rix types

tannergooding · 2024-06-15T20:14:49Z

Similar overall improvements to the PR this is continuing from. Most remain identical to what they were previously:

Method	Toolchain	Mean	Error	StdDev	Median	Min	Max	Ratio	RatioSD
Plane.DotCoordinateBenchmark	runtime-base	0.7257 ns	0.0085 ns	0.0075 ns	0.7241 ns	0.7169 ns	0.7398 ns	1.00	0.01
Plane.DotCoordinateBenchmark	runtime	0.5491 ns	0.0197 ns	0.0184 ns	0.5382 ns	0.5358 ns	0.5879 ns	0.76	0.03

Plane.NormalizeBenchmark	runtime-base	1.1018 ns	0.0207 ns	0.0193 ns	1.0922 ns	1.0873 ns	1.1382 ns	1.00	0.02
Plane.NormalizeBenchmark	runtime	1.1102 ns	0.0296 ns	0.0277 ns	1.0938 ns	1.0860 ns	1.1548 ns	1.01	0.03

Plane.TransformByMatrix4x4Benchmark	runtime	11.3975 ns	0.0586 ns	0.0519 ns	11.4152 ns	11.3087 ns	11.4736 ns	1.00	0.01
Plane.TransformByMatrix4x4Benchmark	runtime-base	9.5247 ns	0.0415 ns	0.0368 ns	9.5240 ns	9.4585 ns	9.5803 ns	0.84	0.00

Quaternion.DivisionOperatorBenchmark	runtime-base	4.1745 ns	0.0149 ns	0.0132 ns	4.1728 ns	4.1580 ns	4.2009 ns	1.00	0.00
Quaternion.DivisionOperatorBenchmark	runtime	1.2661 ns	0.0039 ns	0.0030 ns	1.2660 ns	1.2628 ns	1.2734 ns	0.30	0.00

Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime-base	2.2240 ns	0.0162 ns	0.0144 ns	2.2184 ns	2.2091 ns	2.2562 ns	1.00	0.01
Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime	0.5656 ns	0.0232 ns	0.0217 ns	0.5499 ns	0.5478 ns	0.6062 ns	0.25	0.01

Quaternion.ConcatenateBenchmark	runtime-base	3.1269 ns	0.0402 ns	0.0376 ns	3.1064 ns	3.0917 ns	3.1953 ns	1.00	0.02
Quaternion.ConcatenateBenchmark	runtime	0.5663 ns	0.0227 ns	0.0213 ns	0.5633 ns	0.5338 ns	0.6047 ns	0.18	0.01

Quaternion.ConjugateBenchmark	runtime-base	0.1936 ns	0.0124 ns	0.0116 ns	0.1857 ns	0.1818 ns	0.2111 ns	1.00	0.08
Quaternion.ConjugateBenchmark	runtime	0.1968 ns	0.0162 ns	0.0151 ns	0.1878 ns	0.1837 ns	0.2265 ns	1.02	0.10

Quaternion.CreateFromAxisAngleBenchmark	runtime-base	7.7296 ns	0.0161 ns	0.0151 ns	7.7302 ns	7.6947 ns	7.7572 ns	1.00	0.00
Quaternion.CreateFromAxisAngleBenchmark	runtime	7.4446 ns	0.0478 ns	0.0447 ns	7.4217 ns	7.4048 ns	7.5371 ns	0.96	0.01

Quaternion.InverseBenchmark	runtime-base	0.9127 ns	0.0110 ns	0.0103 ns	0.9101 ns	0.9014 ns	0.9351 ns	1.00	0.02
Quaternion.InverseBenchmark	runtime	1.2364 ns	0.1370 ns	0.1577 ns	1.1716 ns	1.0853 ns	1.5973 ns	1.35	0.17

Quaternion.LerpBenchmark	runtime-base	5.0949 ns	0.0332 ns	0.0310 ns	5.0914 ns	5.0590 ns	5.1670 ns	1.00	0.01
Quaternion.LerpBenchmark	runtime	1.6378 ns	0.0297 ns	0.0278 ns	1.6312 ns	1.6031 ns	1.6804 ns	0.32	0.01

Quaternion.MultiplyByQuaternionBenchmark	runtime-base	2.2071 ns	0.0051 ns	0.0042 ns	2.2056 ns	2.2036 ns	2.2187 ns	1.00	0.00
Quaternion.MultiplyByQuaternionBenchmark	runtime	0.5629 ns	0.0293 ns	0.0274 ns	0.5437 ns	0.5400 ns	0.6083 ns	0.26	0.01

Quaternion.SlerpBenchmark	runtime-base	5.1270 ns	0.0144 ns	0.0135 ns	5.1238 ns	5.1027 ns	5.1533 ns	1.00	0.00
Quaternion.SlerpBenchmark	runtime	1.6400 ns	0.0270 ns	0.0252 ns	1.6321 ns	1.6087 ns	1.6817 ns	0.32	0.00

Vector3.Cross(Vector3 vector1, Vector3 vector2) Vector3.Transform(Vector3 value, Quaternion rotation)

added simd paths for quaternion multiplication, division and concatenate added simd path for matrix4x4 multiplication

dotnet-policy-service · 2024-06-15T20:15:25Z

Tagging subscribers to this area: @dotnet/area-system-numerics
See info in area-owners.md if you want to be subscribed.

…rix types

tannergooding · 2024-06-15T20:14:49Z

Similar overall improvements to the PR this is continuing from. Most remain identical to what they were previously:

Method	Toolchain	Mean	Error	StdDev	Median	Min	Max	Ratio	RatioSD
Plane.DotCoordinateBenchmark	runtime-base	0.7257 ns	0.0085 ns	0.0075 ns	0.7241 ns	0.7169 ns	0.7398 ns	1.00	0.01
Plane.DotCoordinateBenchmark	runtime	0.5491 ns	0.0197 ns	0.0184 ns	0.5382 ns	0.5358 ns	0.5879 ns	0.76	0.03

Plane.NormalizeBenchmark	runtime-base	1.1018 ns	0.0207 ns	0.0193 ns	1.0922 ns	1.0873 ns	1.1382 ns	1.00	0.02
Plane.NormalizeBenchmark	runtime	1.1102 ns	0.0296 ns	0.0277 ns	1.0938 ns	1.0860 ns	1.1548 ns	1.01	0.03

Plane.TransformByMatrix4x4Benchmark	runtime	11.3975 ns	0.0586 ns	0.0519 ns	11.4152 ns	11.3087 ns	11.4736 ns	1.00	0.01
Plane.TransformByMatrix4x4Benchmark	runtime-base	9.5247 ns	0.0415 ns	0.0368 ns	9.5240 ns	9.4585 ns	9.5803 ns	0.84	0.00

Quaternion.DivisionOperatorBenchmark	runtime-base	4.1745 ns	0.0149 ns	0.0132 ns	4.1728 ns	4.1580 ns	4.2009 ns	1.00	0.00
Quaternion.DivisionOperatorBenchmark	runtime	1.2661 ns	0.0039 ns	0.0030 ns	1.2660 ns	1.2628 ns	1.2734 ns	0.30	0.00

Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime-base	2.2240 ns	0.0162 ns	0.0144 ns	2.2184 ns	2.2091 ns	2.2562 ns	1.00	0.01
Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime	0.5656 ns	0.0232 ns	0.0217 ns	0.5499 ns	0.5478 ns	0.6062 ns	0.25	0.01

Quaternion.ConcatenateBenchmark	runtime-base	3.1269 ns	0.0402 ns	0.0376 ns	3.1064 ns	3.0917 ns	3.1953 ns	1.00	0.02
Quaternion.ConcatenateBenchmark	runtime	0.5663 ns	0.0227 ns	0.0213 ns	0.5633 ns	0.5338 ns	0.6047 ns	0.18	0.01

Quaternion.ConjugateBenchmark	runtime-base	0.1936 ns	0.0124 ns	0.0116 ns	0.1857 ns	0.1818 ns	0.2111 ns	1.00	0.08
Quaternion.ConjugateBenchmark	runtime	0.1968 ns	0.0162 ns	0.0151 ns	0.1878 ns	0.1837 ns	0.2265 ns	1.02	0.10

Quaternion.CreateFromAxisAngleBenchmark	runtime-base	7.7296 ns	0.0161 ns	0.0151 ns	7.7302 ns	7.6947 ns	7.7572 ns	1.00	0.00
Quaternion.CreateFromAxisAngleBenchmark	runtime	7.4446 ns	0.0478 ns	0.0447 ns	7.4217 ns	7.4048 ns	7.5371 ns	0.96	0.01

Quaternion.InverseBenchmark	runtime-base	0.9127 ns	0.0110 ns	0.0103 ns	0.9101 ns	0.9014 ns	0.9351 ns	1.00	0.02
Quaternion.InverseBenchmark	runtime	1.2364 ns	0.1370 ns	0.1577 ns	1.1716 ns	1.0853 ns	1.5973 ns	1.35	0.17

Quaternion.LerpBenchmark	runtime-base	5.0949 ns	0.0332 ns	0.0310 ns	5.0914 ns	5.0590 ns	5.1670 ns	1.00	0.01
Quaternion.LerpBenchmark	runtime	1.6378 ns	0.0297 ns	0.0278 ns	1.6312 ns	1.6031 ns	1.6804 ns	0.32	0.01

Quaternion.MultiplyByQuaternionBenchmark	runtime-base	2.2071 ns	0.0051 ns	0.0042 ns	2.2056 ns	2.2036 ns	2.2187 ns	1.00	0.00
Quaternion.MultiplyByQuaternionBenchmark	runtime	0.5629 ns	0.0293 ns	0.0274 ns	0.5437 ns	0.5400 ns	0.6083 ns	0.26	0.01

Quaternion.SlerpBenchmark	runtime-base	5.1270 ns	0.0144 ns	0.0135 ns	5.1238 ns	5.1027 ns	5.1533 ns	1.00	0.00
Quaternion.SlerpBenchmark	runtime	1.6400 ns	0.0270 ns	0.0252 ns	1.6321 ns	1.6087 ns	1.6817 ns	0.32	0.00

Vector3.Cross(Vector3 vector1, Vector3 vector2) Vector3.Transform(Vector3 value, Quaternion rotation)

added simd paths for quaternion multiplication, division and concatenate added simd path for matrix4x4 multiplication

dotnet-policy-service · 2024-06-15T20:15:25Z

Tagging subscribers to this area: @dotnet/area-system-numerics
See info in area-owners.md if you want to be subscribed.

…rix types

tannergooding · 2024-06-15T20:14:49Z

Similar overall improvements to the PR this is continuing from. Most remain identical to what they were previously:

Method	Toolchain	Mean	Error	StdDev	Median	Min	Max	Ratio	RatioSD
Plane.DotCoordinateBenchmark	runtime-base	0.7257 ns	0.0085 ns	0.0075 ns	0.7241 ns	0.7169 ns	0.7398 ns	1.00	0.01
Plane.DotCoordinateBenchmark	runtime	0.5491 ns	0.0197 ns	0.0184 ns	0.5382 ns	0.5358 ns	0.5879 ns	0.76	0.03

Plane.NormalizeBenchmark	runtime-base	1.1018 ns	0.0207 ns	0.0193 ns	1.0922 ns	1.0873 ns	1.1382 ns	1.00	0.02
Plane.NormalizeBenchmark	runtime	1.1102 ns	0.0296 ns	0.0277 ns	1.0938 ns	1.0860 ns	1.1548 ns	1.01	0.03

Plane.TransformByMatrix4x4Benchmark	runtime	11.3975 ns	0.0586 ns	0.0519 ns	11.4152 ns	11.3087 ns	11.4736 ns	1.00	0.01
Plane.TransformByMatrix4x4Benchmark	runtime-base	9.5247 ns	0.0415 ns	0.0368 ns	9.5240 ns	9.4585 ns	9.5803 ns	0.84	0.00

Quaternion.DivisionOperatorBenchmark	runtime-base	4.1745 ns	0.0149 ns	0.0132 ns	4.1728 ns	4.1580 ns	4.2009 ns	1.00	0.00
Quaternion.DivisionOperatorBenchmark	runtime	1.2661 ns	0.0039 ns	0.0030 ns	1.2660 ns	1.2628 ns	1.2734 ns	0.30	0.00

Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime-base	2.2240 ns	0.0162 ns	0.0144 ns	2.2184 ns	2.2091 ns	2.2562 ns	1.00	0.01
Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime	0.5656 ns	0.0232 ns	0.0217 ns	0.5499 ns	0.5478 ns	0.6062 ns	0.25	0.01

Quaternion.ConcatenateBenchmark	runtime-base	3.1269 ns	0.0402 ns	0.0376 ns	3.1064 ns	3.0917 ns	3.1953 ns	1.00	0.02
Quaternion.ConcatenateBenchmark	runtime	0.5663 ns	0.0227 ns	0.0213 ns	0.5633 ns	0.5338 ns	0.6047 ns	0.18	0.01

Quaternion.ConjugateBenchmark	runtime-base	0.1936 ns	0.0124 ns	0.0116 ns	0.1857 ns	0.1818 ns	0.2111 ns	1.00	0.08
Quaternion.ConjugateBenchmark	runtime	0.1968 ns	0.0162 ns	0.0151 ns	0.1878 ns	0.1837 ns	0.2265 ns	1.02	0.10

Quaternion.CreateFromAxisAngleBenchmark	runtime-base	7.7296 ns	0.0161 ns	0.0151 ns	7.7302 ns	7.6947 ns	7.7572 ns	1.00	0.00
Quaternion.CreateFromAxisAngleBenchmark	runtime	7.4446 ns	0.0478 ns	0.0447 ns	7.4217 ns	7.4048 ns	7.5371 ns	0.96	0.01

Quaternion.InverseBenchmark	runtime-base	0.9127 ns	0.0110 ns	0.0103 ns	0.9101 ns	0.9014 ns	0.9351 ns	1.00	0.02
Quaternion.InverseBenchmark	runtime	1.2364 ns	0.1370 ns	0.1577 ns	1.1716 ns	1.0853 ns	1.5973 ns	1.35	0.17

Quaternion.LerpBenchmark	runtime-base	5.0949 ns	0.0332 ns	0.0310 ns	5.0914 ns	5.0590 ns	5.1670 ns	1.00	0.01
Quaternion.LerpBenchmark	runtime	1.6378 ns	0.0297 ns	0.0278 ns	1.6312 ns	1.6031 ns	1.6804 ns	0.32	0.01

Quaternion.MultiplyByQuaternionBenchmark	runtime-base	2.2071 ns	0.0051 ns	0.0042 ns	2.2056 ns	2.2036 ns	2.2187 ns	1.00	0.00
Quaternion.MultiplyByQuaternionBenchmark	runtime	0.5629 ns	0.0293 ns	0.0274 ns	0.5437 ns	0.5400 ns	0.6083 ns	0.26	0.01

Quaternion.SlerpBenchmark	runtime-base	5.1270 ns	0.0144 ns	0.0135 ns	5.1238 ns	5.1027 ns	5.1533 ns	1.00	0.00
Quaternion.SlerpBenchmark	runtime	1.6400 ns	0.0270 ns	0.0252 ns	1.6321 ns	1.6087 ns	1.6817 ns	0.32	0.00

Vector3.Cross(Vector3 vector1, Vector3 vector2) Vector3.Transform(Vector3 value, Quaternion rotation)

added simd paths for quaternion multiplication, division and concatenate added simd path for matrix4x4 multiplication

dotnet-policy-service · 2024-06-15T20:15:25Z

Tagging subscribers to this area: @dotnet/area-system-numerics
See info in area-owners.md if you want to be subscribed.

…rix types

tannergooding · 2024-06-15T20:14:49Z

Similar overall improvements to the PR this is continuing from. Most remain identical to what they were previously:

Method	Toolchain	Mean	Error	StdDev	Median	Min	Max	Ratio	RatioSD
Plane.DotCoordinateBenchmark	runtime-base	0.7257 ns	0.0085 ns	0.0075 ns	0.7241 ns	0.7169 ns	0.7398 ns	1.00	0.01
Plane.DotCoordinateBenchmark	runtime	0.5491 ns	0.0197 ns	0.0184 ns	0.5382 ns	0.5358 ns	0.5879 ns	0.76	0.03

Plane.NormalizeBenchmark	runtime-base	1.1018 ns	0.0207 ns	0.0193 ns	1.0922 ns	1.0873 ns	1.1382 ns	1.00	0.02
Plane.NormalizeBenchmark	runtime	1.1102 ns	0.0296 ns	0.0277 ns	1.0938 ns	1.0860 ns	1.1548 ns	1.01	0.03

Plane.TransformByMatrix4x4Benchmark	runtime	11.3975 ns	0.0586 ns	0.0519 ns	11.4152 ns	11.3087 ns	11.4736 ns	1.00	0.01
Plane.TransformByMatrix4x4Benchmark	runtime-base	9.5247 ns	0.0415 ns	0.0368 ns	9.5240 ns	9.4585 ns	9.5803 ns	0.84	0.00

Quaternion.DivisionOperatorBenchmark	runtime-base	4.1745 ns	0.0149 ns	0.0132 ns	4.1728 ns	4.1580 ns	4.2009 ns	1.00	0.00
Quaternion.DivisionOperatorBenchmark	runtime	1.2661 ns	0.0039 ns	0.0030 ns	1.2660 ns	1.2628 ns	1.2734 ns	0.30	0.00

Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime-base	2.2240 ns	0.0162 ns	0.0144 ns	2.2184 ns	2.2091 ns	2.2562 ns	1.00	0.01
Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime	0.5656 ns	0.0232 ns	0.0217 ns	0.5499 ns	0.5478 ns	0.6062 ns	0.25	0.01

Quaternion.ConcatenateBenchmark	runtime-base	3.1269 ns	0.0402 ns	0.0376 ns	3.1064 ns	3.0917 ns	3.1953 ns	1.00	0.02
Quaternion.ConcatenateBenchmark	runtime	0.5663 ns	0.0227 ns	0.0213 ns	0.5633 ns	0.5338 ns	0.6047 ns	0.18	0.01

Quaternion.ConjugateBenchmark	runtime-base	0.1936 ns	0.0124 ns	0.0116 ns	0.1857 ns	0.1818 ns	0.2111 ns	1.00	0.08
Quaternion.ConjugateBenchmark	runtime	0.1968 ns	0.0162 ns	0.0151 ns	0.1878 ns	0.1837 ns	0.2265 ns	1.02	0.10

Quaternion.CreateFromAxisAngleBenchmark	runtime-base	7.7296 ns	0.0161 ns	0.0151 ns	7.7302 ns	7.6947 ns	7.7572 ns	1.00	0.00
Quaternion.CreateFromAxisAngleBenchmark	runtime	7.4446 ns	0.0478 ns	0.0447 ns	7.4217 ns	7.4048 ns	7.5371 ns	0.96	0.01

Quaternion.InverseBenchmark	runtime-base	0.9127 ns	0.0110 ns	0.0103 ns	0.9101 ns	0.9014 ns	0.9351 ns	1.00	0.02
Quaternion.InverseBenchmark	runtime	1.2364 ns	0.1370 ns	0.1577 ns	1.1716 ns	1.0853 ns	1.5973 ns	1.35	0.17

Quaternion.LerpBenchmark	runtime-base	5.0949 ns	0.0332 ns	0.0310 ns	5.0914 ns	5.0590 ns	5.1670 ns	1.00	0.01
Quaternion.LerpBenchmark	runtime	1.6378 ns	0.0297 ns	0.0278 ns	1.6312 ns	1.6031 ns	1.6804 ns	0.32	0.01

Quaternion.MultiplyByQuaternionBenchmark	runtime-base	2.2071 ns	0.0051 ns	0.0042 ns	2.2056 ns	2.2036 ns	2.2187 ns	1.00	0.00
Quaternion.MultiplyByQuaternionBenchmark	runtime	0.5629 ns	0.0293 ns	0.0274 ns	0.5437 ns	0.5400 ns	0.6083 ns	0.26	0.01

Quaternion.SlerpBenchmark	runtime-base	5.1270 ns	0.0144 ns	0.0135 ns	5.1238 ns	5.1027 ns	5.1533 ns	1.00	0.00
Quaternion.SlerpBenchmark	runtime	1.6400 ns	0.0270 ns	0.0252 ns	1.6321 ns	1.6087 ns	1.6817 ns	0.32	0.00

Vector3.Cross(Vector3 vector1, Vector3 vector2) Vector3.Transform(Vector3 value, Quaternion rotation)

added simd paths for quaternion multiplication, division and concatenate added simd path for matrix4x4 multiplication

dotnet-policy-service · 2024-06-15T20:15:25Z

Tagging subscribers to this area: @dotnet/area-system-numerics
See info in area-owners.md if you want to be subscribed.

…rix types

tannergooding · 2024-06-15T20:14:49Z

Similar overall improvements to the PR this is continuing from. Most remain identical to what they were previously:

Method	Toolchain	Mean	Error	StdDev	Median	Min	Max	Ratio	RatioSD
Plane.DotCoordinateBenchmark	runtime-base	0.7257 ns	0.0085 ns	0.0075 ns	0.7241 ns	0.7169 ns	0.7398 ns	1.00	0.01
Plane.DotCoordinateBenchmark	runtime	0.5491 ns	0.0197 ns	0.0184 ns	0.5382 ns	0.5358 ns	0.5879 ns	0.76	0.03

Plane.NormalizeBenchmark	runtime-base	1.1018 ns	0.0207 ns	0.0193 ns	1.0922 ns	1.0873 ns	1.1382 ns	1.00	0.02
Plane.NormalizeBenchmark	runtime	1.1102 ns	0.0296 ns	0.0277 ns	1.0938 ns	1.0860 ns	1.1548 ns	1.01	0.03

Plane.TransformByMatrix4x4Benchmark	runtime	11.3975 ns	0.0586 ns	0.0519 ns	11.4152 ns	11.3087 ns	11.4736 ns	1.00	0.01
Plane.TransformByMatrix4x4Benchmark	runtime-base	9.5247 ns	0.0415 ns	0.0368 ns	9.5240 ns	9.4585 ns	9.5803 ns	0.84	0.00

Quaternion.DivisionOperatorBenchmark	runtime-base	4.1745 ns	0.0149 ns	0.0132 ns	4.1728 ns	4.1580 ns	4.2009 ns	1.00	0.00
Quaternion.DivisionOperatorBenchmark	runtime	1.2661 ns	0.0039 ns	0.0030 ns	1.2660 ns	1.2628 ns	1.2734 ns	0.30	0.00

Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime-base	2.2240 ns	0.0162 ns	0.0144 ns	2.2184 ns	2.2091 ns	2.2562 ns	1.00	0.01
Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime	0.5656 ns	0.0232 ns	0.0217 ns	0.5499 ns	0.5478 ns	0.6062 ns	0.25	0.01

Quaternion.ConcatenateBenchmark	runtime-base	3.1269 ns	0.0402 ns	0.0376 ns	3.1064 ns	3.0917 ns	3.1953 ns	1.00	0.02
Quaternion.ConcatenateBenchmark	runtime	0.5663 ns	0.0227 ns	0.0213 ns	0.5633 ns	0.5338 ns	0.6047 ns	0.18	0.01

Quaternion.ConjugateBenchmark	runtime-base	0.1936 ns	0.0124 ns	0.0116 ns	0.1857 ns	0.1818 ns	0.2111 ns	1.00	0.08
Quaternion.ConjugateBenchmark	runtime	0.1968 ns	0.0162 ns	0.0151 ns	0.1878 ns	0.1837 ns	0.2265 ns	1.02	0.10

Quaternion.CreateFromAxisAngleBenchmark	runtime-base	7.7296 ns	0.0161 ns	0.0151 ns	7.7302 ns	7.6947 ns	7.7572 ns	1.00	0.00
Quaternion.CreateFromAxisAngleBenchmark	runtime	7.4446 ns	0.0478 ns	0.0447 ns	7.4217 ns	7.4048 ns	7.5371 ns	0.96	0.01

Quaternion.InverseBenchmark	runtime-base	0.9127 ns	0.0110 ns	0.0103 ns	0.9101 ns	0.9014 ns	0.9351 ns	1.00	0.02
Quaternion.InverseBenchmark	runtime	1.2364 ns	0.1370 ns	0.1577 ns	1.1716 ns	1.0853 ns	1.5973 ns	1.35	0.17

Quaternion.LerpBenchmark	runtime-base	5.0949 ns	0.0332 ns	0.0310 ns	5.0914 ns	5.0590 ns	5.1670 ns	1.00	0.01
Quaternion.LerpBenchmark	runtime	1.6378 ns	0.0297 ns	0.0278 ns	1.6312 ns	1.6031 ns	1.6804 ns	0.32	0.01

Quaternion.MultiplyByQuaternionBenchmark	runtime-base	2.2071 ns	0.0051 ns	0.0042 ns	2.2056 ns	2.2036 ns	2.2187 ns	1.00	0.00
Quaternion.MultiplyByQuaternionBenchmark	runtime	0.5629 ns	0.0293 ns	0.0274 ns	0.5437 ns	0.5400 ns	0.6083 ns	0.26	0.01

Quaternion.SlerpBenchmark	runtime-base	5.1270 ns	0.0144 ns	0.0135 ns	5.1238 ns	5.1027 ns	5.1533 ns	1.00	0.00
Quaternion.SlerpBenchmark	runtime	1.6400 ns	0.0270 ns	0.0252 ns	1.6321 ns	1.6087 ns	1.6817 ns	0.32	0.00

Vector3.Cross(Vector3 vector1, Vector3 vector2) Vector3.Transform(Vector3 value, Quaternion rotation)

added simd paths for quaternion multiplication, division and concatenate added simd path for matrix4x4 multiplication

dotnet-policy-service · 2024-06-15T20:15:25Z

Tagging subscribers to this area: @dotnet/area-system-numerics
See info in area-owners.md if you want to be subscribed.

…rix types

tannergooding · 2024-06-15T20:14:49Z

Similar overall improvements to the PR this is continuing from. Most remain identical to what they were previously:

Method	Toolchain	Mean	Error	StdDev	Median	Min	Max	Ratio	RatioSD
Plane.DotCoordinateBenchmark	runtime-base	0.7257 ns	0.0085 ns	0.0075 ns	0.7241 ns	0.7169 ns	0.7398 ns	1.00	0.01
Plane.DotCoordinateBenchmark	runtime	0.5491 ns	0.0197 ns	0.0184 ns	0.5382 ns	0.5358 ns	0.5879 ns	0.76	0.03

Plane.NormalizeBenchmark	runtime-base	1.1018 ns	0.0207 ns	0.0193 ns	1.0922 ns	1.0873 ns	1.1382 ns	1.00	0.02
Plane.NormalizeBenchmark	runtime	1.1102 ns	0.0296 ns	0.0277 ns	1.0938 ns	1.0860 ns	1.1548 ns	1.01	0.03

Plane.TransformByMatrix4x4Benchmark	runtime	11.3975 ns	0.0586 ns	0.0519 ns	11.4152 ns	11.3087 ns	11.4736 ns	1.00	0.01
Plane.TransformByMatrix4x4Benchmark	runtime-base	9.5247 ns	0.0415 ns	0.0368 ns	9.5240 ns	9.4585 ns	9.5803 ns	0.84	0.00

Quaternion.DivisionOperatorBenchmark	runtime-base	4.1745 ns	0.0149 ns	0.0132 ns	4.1728 ns	4.1580 ns	4.2009 ns	1.00	0.00
Quaternion.DivisionOperatorBenchmark	runtime	1.2661 ns	0.0039 ns	0.0030 ns	1.2660 ns	1.2628 ns	1.2734 ns	0.30	0.00

Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime-base	2.2240 ns	0.0162 ns	0.0144 ns	2.2184 ns	2.2091 ns	2.2562 ns	1.00	0.01
Quaternion.MultiplyByQuaternionOperatorBenchmark	runtime	0.5656 ns	0.0232 ns	0.0217 ns	0.5499 ns	0.5478 ns	0.6062 ns	0.25	0.01

Quaternion.ConcatenateBenchmark	runtime-base	3.1269 ns	0.0402 ns	0.0376 ns	3.1064 ns	3.0917 ns	3.1953 ns	1.00	0.02
Quaternion.ConcatenateBenchmark	runtime	0.5663 ns	0.0227 ns	0.0213 ns	0.5633 ns	0.5338 ns	0.6047 ns	0.18	0.01

Quaternion.ConjugateBenchmark	runtime-base	0.1936 ns	0.0124 ns	0.0116 ns	0.1857 ns	0.1818 ns	0.2111 ns	1.00	0.08
Quaternion.ConjugateBenchmark	runtime	0.1968 ns	0.0162 ns	0.0151 ns	0.1878 ns	0.1837 ns	0.2265 ns	1.02	0.10

Quaternion.CreateFromAxisAngleBenchmark	runtime-base	7.7296 ns	0.0161 ns	0.0151 ns	7.7302 ns	7.6947 ns	7.7572 ns	1.00	0.00
Quaternion.CreateFromAxisAngleBenchmark	runtime	7.4446 ns	0.0478 ns	0.0447 ns	7.4217 ns	7.4048 ns	7.5371 ns	0.96	0.01

Quaternion.InverseBenchmark	runtime-base	0.9127 ns	0.0110 ns	0.0103 ns	0.9101 ns	0.9014 ns	0.9351 ns	1.00	0.02
Quaternion.InverseBenchmark	runtime	1.2364 ns	0.1370 ns	0.1577 ns	1.1716 ns	1.0853 ns	1.5973 ns	1.35	0.17

Quaternion.LerpBenchmark	runtime-base	5.0949 ns	0.0332 ns	0.0310 ns	5.0914 ns	5.0590 ns	5.1670 ns	1.00	0.01
Quaternion.LerpBenchmark	runtime	1.6378 ns	0.0297 ns	0.0278 ns	1.6312 ns	1.6031 ns	1.6804 ns	0.32	0.01

Quaternion.MultiplyByQuaternionBenchmark	runtime-base	2.2071 ns	0.0051 ns	0.0042 ns	2.2056 ns	2.2036 ns	2.2187 ns	1.00	0.00
Quaternion.MultiplyByQuaternionBenchmark	runtime	0.5629 ns	0.0293 ns	0.0274 ns	0.5437 ns	0.5400 ns	0.6083 ns	0.26	0.01

Quaternion.SlerpBenchmark	runtime-base	5.1270 ns	0.0144 ns	0.0135 ns	5.1238 ns	5.1027 ns	5.1533 ns	1.00	0.00
Quaternion.SlerpBenchmark	runtime	1.6400 ns	0.0270 ns	0.0252 ns	1.6321 ns	1.6087 ns	1.6817 ns	0.32	0.00

Vector3.Cross(Vector3 vector1, Vector3 vector2) Vector3.Transform(Vector3 value, Quaternion rotation)

added simd paths for quaternion multiplication, division and concatenate added simd path for matrix4x4 multiplication

dotnet-policy-service · 2024-06-15T20:15:25Z

Tagging subscribers to this area: @dotnet/area-system-numerics
See info in area-owners.md if you want to be subscribed.

…rix types

martenfand others added 8 commits March 11, 2024 17:20

Added SIMD paths for:
38bb586
Vector3.Cross(Vector3 vector1, Vector3 vector2) Vector3.Transform(Vector3 value, Quaternion rotation)

Fixed typo
0e14da8

fixed missing conversion back to Vector3
cfa24d9
added simd paths for quaternion multiplication, division and concatenate added simd path for matrix4x4 multiplication

Implemented feedback
8d71e34

fixed non static field
27bd42c

removed whitespace
1f35b22

Use Vector4.Transform
00199d2

Merge remote-tracking branch 'martenf/vectorsimd' into numerics-cleanup
ba252ac

ghost added the area-System.Numerics label Jun 15, 2024

dotnet-policy-serviceBot assigned tannergooding Jun 15, 2024

tannergooding force-pushed the numerics-cleanup branch from d40954b to 3e2abe9Compare June 15, 2024 20:48

tannergooding force-pushed the numerics-cleanup branch from 3e2abe9 to 3f30ff4Compare June 16, 2024 06:48

Perform some additional cleanup of the System.Numerics.Vector and Mat…
d8e29a6
…rix types

tannergooding force-pushed the numerics-cleanup branch from 3f30ff4 to d8e29a6Compare June 16, 2024 16:18

tannergooding marked this pull request as ready for review June 16, 2024 19:23

build-analysisBot mentioned this pull request Jun 16, 2024
FileSystemWatcher_InternalBufferSize_SynchronizingObject test failed in CI #103495
Closed

Merge remote-tracking branch 'dotnet/main' into numerics-cleanup
70013c0

This was referenced Jun 19, 2024
System.Data.OleDb.Tests timeout in net48 x86 Release leg #87783
Closed
NativeAOT legs timing out in CI #102239
Closed
The Operation will be canceled. The next steps may not contain expected logs. dotnet/dnceng#3008
Open
LibraryTests (mostly) timing out #103674
Closed

michaelgsharp approved these changes Jun 19, 2024
View reviewed changes

build-analysisBot mentioned this pull request Jun 20, 2024
[browser] Unable to evaluate script: tab crashed #103623
Closed

tannergooding merged commit 8ab80b9 into dotnet:mainJun 20, 2024

tannergooding deleted the numerics-cleanup branch June 20, 2024 04:56

DrewScoggins mentioned this pull request Jun 25, 2024
[Perf] Windows/x64: 6 Regressions in Perf_Vector on 6/20/2024 7:04:47 AM #103971
Closed

matouskozak mentioned this pull request Jun 26, 2024
[Perf] Linux/x64: 20 Regressions on 6/20/2024 7:04:47 AM #104022
Closed

AndyAyersMS mentioned this pull request Jun 27, 2024
[Perf] Linux/arm64: 17 Improvements on 6/20/2024 7:04:47 AM dotnet/perf-autofiling-issues#37175
Closed

This was referenced Jul 3, 2024
[Perf] Linux/x64: 21 Regressions on 6/20/2024 7:04:47 AM dotnet/perf-autofiling-issues#37140
Open
[wasm][perf] Tracking #96444
Open

tannergooding mentioned this pull request Jul 8, 2024
Godot's .NET math types are way slower than System.Numerics godotengine/godot#69490
Open

github-actionsBot locked and limited conversation to collaborators Jul 20, 2024