Investigate why sunspider overhead is higher in-browser vs. jsshell
Nobody has claimed this yet.
- Dominant language
- C++
- Stars
- 10.7k
- Forks
- 662
- Avg merge
- 2d 3h
- Merged PRs (30d)
- 2
Description
We saw in #10 that, before meliorizations, rr record overhead was about 2.2x for sunspider in-browser, and about 1.9x in jsshell. After meliorizations, we got it down to 1.2x for jsshell but I didn't re-measure in-browser.
On my bare metal x64 machine, I'm seeing ~1.4x overhead in-browser. It's more complicated to measure jsshell overhead, but I expect it to be no worse than 1.2x. So we're seeing ballpark the same difference in overhead.
It'd be interesting to see why the difference is this big. 1.4x is palatable for 1.0, so my inclination is to punt this for now. (This also might end up requiring analysis from someone with serious gecko or SM chops, which I don't have.)
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reproducing and comparing rr record overhead for the SunSpider benchmark in-browser and in jsshell, using the reported x64 measurements as a baseline. Investigate the Gecko or SpiderMonkey paths involved in each environment; done means identifying a credible cause for the difference and explaining it with measurements.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp, javascript
- Domain
- devtools, performance
- Issue type
- Bug
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100