perf(sourcemap): shorten main loop encoding VLQ

[overlookmotel]

Reduce number of operations in main loop in source map VLQ encoding.

#4583 made pushing a byte to output only 2 instructions, so that makes it workable to repeat push_byte_unchecked inside and outside the loop.

On a local benchmark of just VLQ encoding shows this increases performance by 16% (on top of the 11% from #4583).

Probably main gain is it makes a fast path for encoding 0, which is common.

[graphite-app]

Your org has enabled the Graphite merge queue for merging into main

Add the label “merge” to the PR and Graphite will automatically add it to the merge queue when it’s ready to merge. Or use the label “hotfix” to add to the merge queue as a hot fix.

You must have a Graphite account and log in to Graphite in order to use the merge queue. Sign up using this link.

[overlookmotel]

• perf(sourcemap): shorten main loop encoding VLQ #4586 👈
• perf(sourcemap): elide bounds checks in VLQ encoding #4583
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

Join @overlookmotel and the rest of your teammates on Graphite

[codspeed-hq]

CodSpeed Performance Report

Merging #4586 will improve performances by 3.3%

_{Comparing 07-31-perf_sourcemap_shorten_main_loop_encoding_vlq (590d795) with main (e02bac3)}

Summary

⚡ 1 improvements
✅ 31 untouched benchmarks

Benchmarks breakdown

	Benchmark	main	07-31-perf_sourcemap_shorten_main_loop_encoding_vlq	Change
⚡	sourcemap[cal.com.tsx]	74.6 ms	72.2 ms	+3.3%

[overlookmotel]

One thing I'm confused by. On a local benchmark that purely covers VLQ encoding (excluding the rest of source map generation):

• #4583 shows an 11% speed-up.
• This PR shows a 16% speed-up.

But #4583 barely registers on our CodSpeed sourcemap benchmark, whereas this PR gets +3%.

Maybe:

• #4583 only has an effect on aarch64 processors (I benchmarked locally on a Macbook Pro M1) or
• The gain of #4583 is all in something that CodSpeed doesn't measure e.g. better branch prediction.

[graphite-app]

Merge activity

• Aug 1, 8:20 AM EDT: The merge label 'merge' was detected. This PR will be added to the Graphite merge queue once it meets the requirements.
• Aug 1, 8:20 AM EDT: Boshen added this pull request to the Graphite merge queue.
• Aug 1, 8:24 AM EDT: Boshen merged this pull request with the Graphite merge queue.

[overlookmotel]

Concerning previous comment about CodSpeed benchmarks, I replicated the results of sourcemap benchmark that CodSpeed produced locally on MacBook Pro M1.

#4583 hardly made any change to benchmark (+0.16%), but this PR gave an approx +2% speed-up. So broadly in line with CodSpeed.

I can't explain that, as both changes each give roughly same speed-up on a benchmark which purely measures VLQ encoding. But, whatever the reason, it doesn't seem to be particularly related to running benchmark on CodSpeed vs locally.