Add VectorMax and Clamp to SVE microbenchmark

[ylpoonlg]

Performance Results

Run on Neoverse-V2

Method	Size	Mean	Error	StdDev	Median	Min	Max	Allocated
Scalar	15	5.381 ns	0.0042 ns	0.0035 ns	5.380 ns	5.378 ns	5.391 ns	-
Vector128VectorMax	15	2.279 ns	0.0055 ns	0.0043 ns	2.278 ns	2.276 ns	2.292 ns	-
SveVectorMax	15	1.741 ns	0.0021 ns	0.0019 ns	1.742 ns	1.739 ns	1.745 ns	-
Scalar	127	68.210 ns	0.1491 ns	0.1245 ns	68.165 ns	68.126 ns	68.531 ns	-
Vector128VectorMax	127	8.865 ns	0.0280 ns	0.0262 ns	8.868 ns	8.790 ns	8.901 ns	-
SveVectorMax	127	19.277 ns	0.0137 ns	0.0128 ns	19.280 ns	19.256 ns	19.296 ns	-
Scalar	527	321.773 ns	0.0619 ns	0.0517 ns	321.769 ns	321.699 ns	321.891 ns	-
Vector128VectorMax	527	39.748 ns	0.0437 ns	0.0409 ns	39.744 ns	39.684 ns	39.820 ns	-
SveVectorMax	527	78.429 ns	0.0336 ns	0.0298 ns	78.424 ns	78.392 ns	78.489 ns	-
Scalar	10015	6,092.076 ns	1.1249 ns	0.9394 ns	6,091.902 ns	6,090.969 ns	6,094.097 ns	-
Vector128VectorMax	10015	859.076 ns	13.3252 ns	12.4644 ns	862.904 ns	827.475 ns	871.356 ns	-
SveVectorMax	10015	1,535.044 ns	4.8796 ns	4.5643 ns	1,532.708 ns	1,530.924 ns	1,542.698 ns	-

Method	Size	Mean	Error	StdDev	Median	Min	Max	Allocated
Scalar	15	7.3506 ns	0.0053 ns	0.0047 ns	7.3501 ns	7.3441 ns	7.3589 ns	-
Vector128Clamp	15	1.3256 ns	0.1418 ns	0.1633 ns	1.2917 ns	1.1610 ns	1.5025 ns	-
SveClamp	15	0.9035 ns	0.0032 ns	0.0025 ns	0.9031 ns	0.9001 ns	0.9076 ns	-
Scalar	127	73.3619 ns	0.8254 ns	0.6444 ns	73.1360 ns	72.9714 ns	75.2351 ns	-
Vector128Clamp	127	22.5408 ns	0.0095 ns	0.0080 ns	22.5431 ns	22.5205 ns	22.5506 ns	-
SveClamp	127	23.8960 ns	0.0040 ns	0.0031 ns	23.8965 ns	23.8900 ns	23.9001 ns	-
Scalar	527	313.9545 ns	0.2512 ns	0.2350 ns	313.9443 ns	313.5304 ns	314.3282 ns	-
Vector128Clamp	527	112.9809 ns	0.0761 ns	0.0674 ns	112.9729 ns	112.8673 ns	113.0953 ns	-
SveClamp	527	115.3293 ns	0.0392 ns	0.0347 ns	115.3185 ns	115.2728 ns	115.3961 ns	-
Scalar	10015	7,691.7844 ns	517.1365 ns	595.5352 ns	7,280.6021 ns	7,273.3083 ns	8,619.5838 ns	-
Vector128Clamp	10015	2,265.9251 ns	0.6277 ns	0.5564 ns	2,265.8004 ns	2,265.2121 ns	2,266.9799 ns	-
SveClamp	10015	2,269.6220 ns	0.9217 ns	0.8621 ns	2,269.2198 ns	2,268.6624 ns	2,271.5569 ns	-

cc @dotnet/arm64-contrib @SwapnilGaikwad @LoopedBard3

[Copilot]

Pull request overview

Adds new SVE-focused microbenchmarks to the src/benchmarks/micro/sve suite to measure VectorMax and Clamp implementations on Arm64 + SVE-capable machines, enabling comparison between scalar, Vector128 (AdvSimd), and SVE approaches.

Changes:

• Introduce VectorMax benchmark with Scalar, Vector128, and SVE implementations plus correctness verification.
• Introduce Clamp benchmark with Scalar, Vector128, and two SVE loop variants plus correctness verification.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 4 comments.

File	Description
src/benchmarks/micro/sve/VectorMax.cs	New SVE benchmark for max value + index reduction across an array, with Scalar/Vector128/SVE variants.
src/benchmarks/micro/sve/Clamp.cs	New clamp-style benchmark comparing Scalar vs Vector128 and SVE implementations.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review. Take the survey.