Add benchmarks

[ptoffy]

No description provided.

[MahdiBM]

IMO we should use static thresholds, like SwiftNIO does.

package-benchmark relatively recently added a thresholds commands for this as well, and there is a bit of docs for it here.

I think using something like this should work to compare static thresholds against current branch:

            swift package -c release --disable-sandbox \
              benchmark thresholds check \
              --path $PWD/Benchmarks/Thresholds/ \
              --format markdown

[MahdiBM]

Benchmark accepts a setUp closure IIRC. I'd personally just use a lazy variable to make sure the whole thing is only done once, regardless of how many iterations the benchmark is going to have.

[ptoffy]

IMO we should use static thresholds, like SwiftNIO does.

The problem with this is we don't have any baseline thresholds (yet)

[github-actions]

PR benchmarks run at
(main branch benchmarks not available)

[github-actions]

Benchmark Report

❌ Pull request has significant performance differences 📊

Click to expand comparison result

Benchmark check running at 2025-01-10 16:48:19 UTC

Could not find any matching absolute thresholds at path [/__w/multipart-kit/multipart-kit/Benchmarks/Thresholds/], failing threshold check.

Click to expand benchmark result

Baseline benchmarks

Host dc75ffde3938 with 2 aarch64 processors with 3 GB memory, running:
#23-Ubuntu SMP Mon Dec  9 23:51:16 UTC 2024

Parser

Collating Parser Allocations

Metric	p0	p25	p50	p75	p90	p99	p100	Samples
Malloc (total) (K) *	153	153	153	153	153	153	153	1

Collating Parser Throughput

Metric	p0	p25	p50	p75	p90	p99	p100	Samples
Throughput (# / s) (#)	1	1	1	1	1	1	1	1
Time (total CPU) (ms) *	1430	1430	1430	1430	1430	1430	1430	1
Time (wall clock) (ms) *	1430	1430	1430	1430	1430	1430	1430	1

Streaming Parser Allocations

Metric	p0	p25	p50	p75	p90	p99	p100	Samples
Malloc (total) (K) *	153	153	153	153	153	153	153	1

Streaming Parser Throughput

Metric	p0	p25	p50	p75	p90	p99	p100	Samples
Throughput (# / s) (#)	1	1	1	1	1	1	1	1
Time (total CPU) (ms) *	1060	1060	1060	1060	1060	1060	1060	1
Time (wall clock) (ms) *	1068	1068	1068	1068	1068	1068	1068	1

Serializer

Serializer Allocations

Metric	p0	p25	p50	p75	p90	p99	p100	Samples
Malloc (total) *	8	8	8	8	8	8	8	10000

Serializer Throughput

Metric	p0	p25	p50	p75	p90	p99	p100	Samples
Throughput (# / s) (#)	386	384	383	382	381	365	330	371
Time (total CPU) (μs) *	0	0	0	10000	10000	10000	10000	371
Time (wall clock) (μs) *	2592	2603	2609	2617	2628	2742	3033	371