Closed Bug 1342797 Opened 8 years ago Closed 8 years ago

Javascript benchmark 3.5X slower than Chrome

Categories

(Core :: JavaScript Engine, defect)

51 Branch
defect
Not set
normal

Tracking

()

RESOLVED FIXED

People

(Reporter: ianopolous, Unassigned)

References

(Depends on 2 open bugs)

Details

User Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Ubuntu Chromium/55.0.2883.87 Chrome/55.0.2883.87 Safari/537.36 Steps to reproduce: @zbraniecki recommended I submit my real world JS benchmark here. The benchmark is an encrypted file storage app (Peergos), which downloads the cipher text of a 15 MiB file (The Sintel trailer), decrypts it in the browser, and then triggers a download to the host file system. In Chrome this take 24s beginning to end. In Firefox it takes 85s. To reproduce, follow this link and wait for the file to download: https://demo.peergos.net/#zFoxfTAUGfEbcS9NJCgKPhxuEKdibFWdjWZmR6UkfSigStRyc/7a7GJeJeiXpd9NeSBE56X4PPkitdLyMZYRSYpUfUFS7H/5Pf7SvuUU89z6i5HJHBKuNZDHJ1hA6p9YzkkMZAMAveV4AAHgmH?download=true
Flags: needinfo?(jdemooij)
Thank you for reporting this! Can you confirm this is faster with a Firefox Nightly build? I get the slow script dialog with Firefox 51 but not with Nightly builds. It's still not super fast, I'll file some bugs to improve this.
Status: UNCONFIRMED → NEW
Ever confirmed: true
Flags: needinfo?(jdemooij)
Depends on: 890329
Depends on: 1344173
Depends on: 1344177
Depends on: 1344463
Thank you so much for your help, Jan. I can confirm on a Nightly v54, it takes 48 seconds, or down to 2X slower than chrome, which is a big improvement.
Depends on: 1337773
(In reply to ianopolous from comment #2) > Thank you so much for your help, Jan. I can confirm on a Nightly v54, it > takes 48 seconds, or down to 2X slower than chrome, which is a big > improvement. Great, we've done a lot of work recently to eliminate performance cliffs and it's good to see it's paying off already. I pushed 2 patches today to improve this more, they should be in the next Nightly build after these bugs get closed. Probably tomorrow or on Monday depending on when these changes get merged.
ianopolous: Did you have a chance to test this again?
G'day Tom, Jan, Yes I tried it again on Monday evening, but the timing was the same. However, the patches mustn't have made it, as I've just tried again now and it is down to 27s, which is almost within measurement error of chrome. Fantastic! Is it the August public release I should expect these changes? Thank you so much for all your awesome work!
(In reply to ianopolous from comment #5) > Is it the August public release I should expect these changes? These changes should be in FF 54 (June 13th) but FF 53 might also be faster than 51 and 52. Your website has a message suggesting Firefox is 4x slower than Chrome. Maybe you can remove that message then or at least check the version number.
Yes, I've just updated that number to 3X (based on the just released firefox 52), which will make our next deployment. I will reevaluate when 53 is out, and update accordingly. Thanks again for being so awesome!
ianopolous: the demo website in comment 0 has an invalid certificate (it expired yesterday). I was wondering if bug 1337773 made a difference.
jandem: Thanks! I've just renewed it. Is that fix in the nightly yet?
(In reply to ianopolous from comment #9) > jandem: Thanks! I've just renewed it. Is that fix in the nightly yet? Yes. How's performance for you nowadays?
jandem: The latest nightly seems to have fully caught up to chrome (24s each). Well done (and thank you)! I look forward to removing the warning in a month.
Thanks for testing again! Great to hear the the performance issue is fixed now. Firefox 55 is going to be released 2017-08-08.
Status: NEW → RESOLVED
Closed: 8 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.