Firefox Startup Performance Weekly Summary

Current numbers are available on the Performance Snapshot page.

Summary, relative to Firefox 3.5:

• Warm startup: For Mac, 36% better on 3.6 and 35% better on 3.7. For Windows, 5% and 5%. Flat on Linux. Also, Warm startup for Mac on 3.6 is a whopping 13% better than last week, due to the landing of bug 517804.
• Cold startup:  For Mac, 20% better on both 3.6 and 3.7. For Windows, not measuring yet. For Linux, we’re seeing a regression of ~9% across branch and trunk in the snapshot but not on the graphs, so I need to figure out where the discrepancy is.

This week’s activity:

• Dirty-cold-Ts went live this week, thanks to Alice and Lukas. Example: cold startup with a large places.sqlite on Mac.
• Joel is making progress on making a super-static Firefox in bug 525013.
• Ben is making progress on the fastload replacement in bug 520309.
• No updates on Windows cold-startup testing for Talos on bug 522807. I need to test on Vista, and turn off Pre/Superfetch.
• Taras has patches up for service caching (bug 516085) and super-fast-path-ing of Components.* (bug 512584), however the latter he’s hit a wall, passing on to Blake or someone else who knows that code.
• Ted landed rebasing on Windows in bug 484799.
• Jonathan Kew has a new patch in bug 519445 for further reductions in Mac startup time spent in font system initialization, just about there…
• Ryan Flint put a patch to minify JS on bug 524858, not working yet, but significantly reduced the size of shipped JavaScript files.

Projects in a holding pattern:

• JARification: David abandoned moving JS modules into a JAR file, since those files are fastloaded. However, since we want things like post-extension-install restarts to be fast, and those cause fastload cache invalidation, we might want to do things like this anyways. I filed a bug for the same treatment for components. These are lower priority, since they’re not the normal startup case. Follow along with all JAR-ification via the tracker bug.
• Startup Timeline: No updates, still not landed. Add [ft] in the whiteboard of your bug w/ the function names you want timed and David will generate it and update the bug.
• Static Analysis: No progress on bug 506128. David needs to file a bug with the final log of named-yet-uncalled functions.
• Dirty Profile Testing: No progress. Need to list scenarios, file bugs for each, generate Talos config patches and profile data, and then move it into Rel-Eng territory. Also, need to get a separate Tinderbox tree, since it’s going to cause a bazillion new columns.
• Joel Reymont noted in bug 513076 that there are serious drawbacks to getting our libraries in the dyld shared cache on Mac, so has deprioritized that work.
• No updates on Zack’s CSS parser changes in bug 513149.

As usual, more details and links are available on the project wiki, and we’re available to answer questions in #startup on irc.mozilla.org.

Firefox Startup Performance Weekly Summary

Current numbers are available on the Performance Snapshot page. Thanks to Chris Atlee for deploying it to the graph server. The snapshot is updated every 5 minutes with fresh data from Talos.

Summary, relative to Firefox 3.5:

• Warm startup: For Mac, 26% better on 3.6 and 31% better on 3.7. For Windows, 4% and 7%. Flat on Linux.
• Cold startup:  For Mac, 22% better on 3.6, 23% better on 3.7. For Windows, not measuring yet. For Linux, 3% and 9%.

This week’s activity:

• Using the data from bug 524202, Joel is tracking down exactly how and when dynamic library code is loaded. They’re past diagnostics, and are now into implementation, coaxing the linker into ordering functions in the optimal sequence. See this comment for a good summary of the issue. Joel put up a very detailed blog post about the work they’ve been doing.
• Ryan Flint updated the spellcheck bug with some new findings.
• I got a working (so far) patch for Windows cold-startup testing for Talos on bug 522807.
• Taras is in the patch review cycle for service caching in bug 516085.
• Re-enabling rebasing on Windows in bug 484799 has review now, just needs landing.
• John Dagget and Jonathan Kew have a bunch of new findings in bug 519445 for further reductions in Mac startup time spent in font system initialization. Sometimes the patch helps, sometimes it doesn’t.
• Bug 512645, removing the setTimeout 10ms wait in chrome JS has landed.
• Drew’s work on bug 506814, getting rid of Change GetPersistentDescriptor/SetPersistentDescriptor on Mac, has landed.

Projects in a holding pattern:

• Ben Hsieh has been prototyping a whole Fastload cache replacement in bug 520309.
• JARification: David abandoned moving JS modules into a JAR file, since those files are fastloaded. However, since we want things like post-extension-install restarts to be fast, and those cause fastload cache invalidation, we might want to do things like this anyways. I filed a bug for the same treatment for components. These are lower priority, since they’re not the normal startup case. Follow along with all JAR-ification via the tracker bug.
• Startup Timeline: No updates, still not landed. Add [ft] in the whiteboard of your bug w/ the function names you want timed and David will generate it and update the bug.
• Static Analysis: No progress on bug 506128. David needs to file a bug with the final log of named-yet-uncalled functions.
• Dirty Profile Testing: No progress. Need to list scenarios, file bugs for each, generate Talos config patches and profile data, and then move it into Rel-Eng territory. Also, need to get a separate Tinderbox tree, since it’s going to cause a bazillion new columns.
• Joel Reymont noted in bug 513076 that there are serious drawbacks to getting our libraries in the dyld shared cache on Mac, so has deprioritized that work.
• No updates on Zack’s CSS parser changes in bug 513149.

As usual, more details and links are available on the project wiki, and we’re available to answer questions in #startup on irc.mozilla.org.

Firefox Startup Performance Weekly Summary

This week brings a boost in visibility of results, not just for startup, but for all the performance testing we’re doing on all branches and platforms. As I mentioned last week, I was working on an automated method of generating the cross-branch startup results. Luckily Johnath and Chris Atlee had done the hard work when making the performance dashboard. It has a JSON file that contains a roll-up of the previous 7 days performance data, which is updated every 5 minutes with new Talos results. Using that as the datasource, I wrote  a script summarizes the results for each test+branch+OS combination across all the boxes that returned results, as well as stable branch and previous week differences. The final product gives a snapshot view of how each branch compares to the stable branch.

This is useful for a few different reasons. First, we (and the press and our users and our managers and everyone really) are able to know at-a-glance how any branch compares to the stable release branch. An example of another use is that before the JS team does a Tracemonkey merge, they can quickly see if any major performance effects can be expected.

The table shows red or green for any differences that are outside of a 2% threshold, to take test noise into account. This is quite liberal, as I’ve already calculated the numbers to take into account whether the difference is within the standard deviation. Reducing the noise in the tests would be a big help – perhaps some researcher will take up Roc’s challenge. A second point of trust is my math I’m not a statistician, so please view source and let me know where I’ve miscalculated.

An example of the full report is here. I’m going to file a bug and work with release-engineering to get it pushed out to the graph server, where the dashboard lives. The data from this week’s startup table is copied below (though sans the color-coding of the real thing. Actually, now WordPress keeps stripping out my styles, so you get an ugly table this week).

Ts

	Firefox3.5	Firefox3.6	Firefox	TraceMonkey
Leopard	median: 1469deviation: 110mean: 1481 from last week: 2%	median: 1115 deviation: 59 mean: 1096 from last week: 3% from 3.5: 26%	median: 1036 deviation: 48 mean: 1006 from last week: 1% from 3.5: 32%	median: 1014 deviation: 38 mean: 1000 from last week: 0% from 3.5: 32%
Linux	median: 625 deviation: 7 mean: 626 from last week: 0%	median: 632 deviation: 7 mean: 633 from last week: 0% from 3.5: -1%	median: 619 deviation: 10 mean: 623 from last week: 1%	median: 628 deviation: 7 mean: 628 from last week: 0%
Vista	median: 538 deviation: 8 mean: 540 from last week: 0%	median: 533 deviation: 13 mean: 537 from last week: 1% from 3.5: 1%	median: 503 deviation: 23 mean: 509 from last week: 0% from 3.5: 6%	median: 511 deviation: 41 mean: 531 from last week: 11% from 3.5: 2%
XP	median: 461 deviation: 6 mean: 461 from last week: 0%	median: 464 deviation: 6 mean: 464 from last week: 1% from 3.5: -1%	median: 448 deviation: 7 mean: 448 from last week: 0% from 3.5: 3%	median: 530 deviation: 38 mean: 501 from last week: 10% from 3.5: -9%

This week’s activity:

• Taras and Joel are still working on bug 524202, tracking down exactly how and when dynamic library code is loaded. They’re past diagnostics, and are now into implementation, coaxing the linker into ordering functions in the optimal sequence. See this comment for a good summary of the issue. Joel put up a very detailed blog post about the work they’ve been doing.
• Ryan Flint posted an update on his startup bug activity this week.
• Have a patch enabling Windows cold-startup testing for Talos for bug 522807, but it’s causing the whole OS to freeze, only recoverable via reboot. Fun! I also added some details and links about how Prefetch/SuperFetch work on Windows to the wiki.
• Rob Strong pushed bug 311965 to mozilla-central, comm-central and 1.9.2 while also ensuring not to break all the toolkit apps that depend on this code. Truly a gentleman of the Mozilla ecosystem.
• Taras put a new patch up for service caching in bug 516085.
• Everything is about ready for re-enabling rebasing on Windows in bug 484799, just needs landing.
• John Dagget posted some test times in bug 519445 for yet further reductions in Mac startup time spent in font system initialization, just needs review.
• Bug 512645, removing the setTimeout 10ms wait in chrome JS, is ready to land. I’ll try to land this weekend if the bug owner doesn’t get around to it first (hint hint).

Projects in a holding pattern:

• Drew has a patch up for bug 506814, getting rid of Change GetPersistentDescriptor/SetPersistentDescriptor on Mac, just needs review from Josh.
• Ben Hsieh has been prototyping a whole Fastload cache replacement in bug 520309.
• JARification: David abandoned moving JS modules into a JAR file, since those files are fastloaded. However, since we want things like post-extension-install restarts to be fast, and those cause fastload cache invalidation, we might want to do things like this anyways. I filed a bug for the same treatment for components. These are lower priority, since they’re not the normal startup case. Follow along with all JAR-ification via the tracker bug.
• Startup Timeline: No updates, still not landed. Add [ft] in the whiteboard of your bug w/ the function names you want timed and David will generate it and update the bug.
• Static Analysis: No progress on bug 506128. David needs to file a bug with the final log of named-yet-uncalled functions.
• Dirty Profile Testing: No progress. Need to list scenarios, file bugs for each, generate Talos config patches and profile data, and then move it into Rel-Eng territory. Also, need to get a separate Tinderbox tree, since it’s going to cause a bazillion new columns.
• Joel Reymont noted in bug 513076 that there are serious drawbacks to getting our libraries in the dyld shared cache on Mac, so has deprioritized that work.
• No updates on Zack’s CSS parser changes in bug 513149.

As usual, more details and links are available on the project wiki, and we’re available to answer questions in #startup on irc.mozilla.org.

UPDATE: Fixed link to full report.

Firefox Startup Performance Weekly Summary

No numbers summary this week, as 1) there weren’t any major landings and 2) I’m working on a script to automate the generation of these summarized numbers. Updates on the week’s activity is below. As usual, more details and links are available on the project wiki, and we’re available to answer questions in #startup on irc.mozilla.org.

• Taras and Joel are working on bug 524202, tracking down exactly how and when dynamic library code is loaded. See this comment for a good summary of the issue.
• I think I’ve found a scenario for stable cold startup numbers on Windows, which is basically: 1) Run consume.exe for N seconds where N is how long it takes to consume all physical RAM (and which also pegs the CPU), 2) sleep for 30 seconds (or however long it takes for the system to quiet down) and 3) measure startup time as usual. Next step is to get a Talos patch up, and work with Rel/Eng to get it deployed into testing so we can see numbers on real Talos boxes.
• Rob Strong split up the update service in bug 311965, has most reviews (affects all toolkit apps) and is about ready to land. This showed a significant win on WinCE startup.

Projects in a holding pattern:

• Drew has a patch up for bug 506814, getting rid of Change GetPersistentDescriptor/SetPersistentDescriptor on Mac, just needs review from Josh.
• Bug 512645, removing the setTimeout 10ms wait in chrome JS, is just waiting on review just needs to be landed (thanks jesse).
• Ben Hsieh has been prototyping a whole Fastload cache replacement in bug 520309.
• Ted has been looking at re-enabling rebasing on Windows in bug 484799 for a potential performance boost there.
• Service caching work in bug 516085 still needs to be pushed to the Places branch for testing.
• Moving font-loading out of the startup path on Mac: Jonathan Kew filed bug 519445 with a WIP patch for yet further reductions in Mac startup time spent in font system initialization.
• JARification: David abandoned moving JS modules into a JAR file, since those files are fastloaded. However, since we want things like post-extension-install restarts to be fast, and those cause fastload cache invalidation, we might want to do things like this anyways. I filed a bug for the same treatment for components. These are lower priority, since they’re not the normal startup case. Follow along with all JAR-ification via the tracker bug.
• Startup Timeline: No updates, still not landed. Add [ft] in the whiteboard of your bug w/ the function names you want timed and David will generate it and update the bug.
• Static Analysis: No progress on bug 506128. David needs to file a bug with the final log of named-yet-uncalled functions.
• Dirty Profile Testing: No progress. Need to list scenarios, file bugs for each, generate Talos config patches and profile data, and then move it into Rel-Eng territory. Also, need to get a separate Tinderbox tree, since it’s going to cause a bazillion new columns.
• Joel Reymont noted in bug 513076 that there are serious drawbacks to getting our libraries in the dyld shared cache on Mac, so has deprioritized that work.
• No updates on Zack’s CSS parser changes in bug 513149.

Firefox Startup Performance Weekly Summary

The numbers aren’t updated yet, as they were oddly counter-intuitive. Firefox 3.5 showed a massive week-over-week improvement in cold startup on Windows, which from what I can tell is not expected. On Leopard, cold startup shows a couple of percentage point improvement over last week, which is expected. Warm startup for Mac shows a 20% improvement over 3.5, but a 10% *regression* from last week, contradicting the numbers from Ts. Again, counter-intuitive numbers given that we landed a bunch of changes focused squarely on startup performance during that period.

Fortuitously, Alice and the release engineering team deployed automated cold startup testing for Mac and Linux. This gives us per-checkin visibility of cold startup times, and removes the need for error-prone local measurements for those platforms. On Windows we haven’t yet figured out how to emulate cold startup reliably, so that’s the next step.

As usual, more details and links are available on the project wiki, and we’re available to answer questions in #startup on irc.mozilla.org.

Recent activity:

• Bug 511761 landed, Ben Hsieh’s work to optimize fastload cache invalidation, with a 3% win on warm startup of WinXP.
• Alfred Kayser put a patch up on bug 511754, which improves JAR file reading efficiency.
• Rob Strong closed bug 521956 and started some cleanup of nsUpdateService, further reducing the size of it for bug 311965.
• Drew has a patch up for bug 506814, getting rid of Change GetPersistentDescriptor/SetPersistentDescriptor on Mac.
• Bug 504858 pushes back the population of the bookmarks toolbar until after the browser window comes up. Dao put up a new patch. Measuring the wall-clock effect of this on startup, and determining what exactly is “gaming” the Ts test is part of the work here.
• Service caching work in bug 516085 still needs to be pushed to the Places branch for testing.

Projects in a holding pattern:

• Moving font-loading out of the startup path on Mac: Jonathan Kew filed bug 519445 with a WIP patch for yet further reductions in Mac startup time spent in font system initialization.
• JARification: David abandoned moving JS modules into a JAR file, since those files are fastloaded. However, since we want things like post-extension-install restarts to be fast, and those cause fastload cache invalidation, we might want to do things like this anyways. I filed a bug for the same treatment for components. These are lower priority, since they’re not the normal startup case. Follow along with all JAR-ification via the tracker bug.
• Startup Timeline: No updates, still not landed. Add [ft] in the whiteboard of your bug w/ the function names you want timed and David will generate it and update the bug.
• Static Analysis: No progress on bug 506128. David needs to file a bug with the final log of named-yet-uncalled functions.
• Dirty Profile Testing: No progress. Need to list scenarios, file bugs for each, generate Talos config patches and profile data, and then move it into Rel-Eng territory. Also, need to get a separate Tinderbox tree, since it’s going to cause a bazillion new columns.
• Joel Reymont noted in bug 513076 that there are serious drawbacks to getting our libraries in the dyld shared cache on Mac, so has deprioritized that work.
• No updates on Zack’s CSS parser changes in bug 513149.

Firefox Startup Performance Weekly Summary

Work on blockers for 3.6 stole some time last week, but some startup work advanced nonetheless, including a bunch of landings. I’ve posted the numbers from early last week below. The percentages are the difference from 3.5. Given the spate of landings later in the week after we measured,  I’ll have to update the numbers again to get the full effect of everything listed below. As usual, more details and links are available on the project wiki, and we’re available to answer questions in #startup on irc.mozilla.org.

Cold Startup (median results from Juan and myself)

Platform	3.5	3.6	Trunk
Mac Leopard (10/6)	12161	10440 (-14%)	10175 (-16%)
Win XP (10/4)	5610	5640 (0%)	4050 (-27.8%)

Warm Startup (median of results from Talos, updated 10/6)

Platform	3.5	3.6	Trunk
Mac Leopard	1500ms	1190ms (-20%)	1040ms (-30%)
Win XP	458ms	470ms (+2%)	468ms (+2%)
Win Vista	545ms	544ms (0%)	545ms (0%)
Linux	624ms	634ms (0%)	634ms (0%)

Recent landings:

• Ryan landed bug 475289 to lazily initialize the engineMetadataService, bug 520342 to consolidate service initialization in the microsummary service, and pushed bug 499123, coalescing browser about: pages, to the 3.6 branch.
• Taras landed a bunch of optimizations to the fastload system as well as his work on combining the different preference files.
• I turned on PGO for Places, mozStorage and SQLite.
• Alfred Kayser’s patch on bug 510844 landed, which should increase throughput of JAR file reading.
• Ben Hsieh got review on bug 510991, which removes some stats() from the startup path.

Recent activity:

• Ryan put up a patch for bug 520284, which utilizes an important change that happened recently in bug 471219 which allows components to register for periodic timer notifications *without* instantiating the service or component. An example is in Ryan’s patch, and the docs are in the source.
• Bug 504858 pushes back the population of the bookmarks toolbar until after the browser window comes up. Dao put up a new patch. Measuring the wall-clock effect of this on startup, and determining what exactly is “gaming” the Ts test is part of the work here.
• Ben Hsieh is in review cycles for bug 511761 which optimizes the fastload cache invalidation, and consolidates some of the invalidation triggers.
• Service caching work in bug 516085, still made no progress. This week I’ll try pushing this to the Places branch.
• No updates on Zack’s CSS parser changes in bug 513149.
• Cold Startup Testing: Alice put more patches on bug 510587, getting us ever closer to automated testing of cold startup.

Projects in a holding pattern:

• Moving font-loading out of the startup path on Mac: Jonathan Kew filed bug 519445 with a WIP patch for yet further reductions in Mac startup time spent in font system initialization.
• JARification: David abandoned moving JS modules into a JAR file, since those files are fastloaded. However, since we want things like post-extension-install restarts to be fast, and those cause fastload cache invalidation, we might want to do things like this anyways. I filed a bug for the same treatment for components. These are lower priority, since they’re not the normal startup case. Follow along with all JAR-ification via the tracker bug.
• Startup Timeline: No updates, still not landed. Add [ft] in the whiteboard of your bug w/ the function names you want timed and David will generate it and update the bug.
• Static Analysis: No progress on bug 506128. David needs to file a bug with the final log of named-yet-uncalled functions.
• Dirty Profile Testing: No progress. Need to list scenarios, file bugs for each, generate Talos config patches and profile data, and then move it into Rel-Eng territory. Also, need to get a separate Tinderbox tree, since it’s going to cause a bazillion new columns.
• Joel Reymont noted in bug 513076 that there are serious drawbacks to getting our libraries in the dyld shared cache on Mac, so has deprioritized that work.

Firefox Startup Performance Weekly Summary

This week brought a big win on Mac, and a slew of changes that are about to land.

Recent activity:

• Bug 517045 is fixed and brings another win on Mac. Found by Joel Reymont and fixed by John Daggett, it improves cold startup by ~20% and warm startup by ~4%. It gets better: Jonathan Kew filed bug 519445 with a WIP patch for yet further reductions in Mac startup time spent in font system initialization.
• Taras Glek’s work on combining the different preference files into a single one is ready to land.
• We now have unit-tested optimized builds, thanks to Release Engineering’s work on bug 486783. This means we can now turn on PGO for Places, mozStorage and SQLite, which I’ll land presently.
• Bug 504858 pushes back the population of the bookmarks toolbar until after the browser window comes up. The patch has review, is ready to land.
• Alfred Kayser’s patch on bug 510844 which should increase throughput of JAR file reading, is ready to land.
• Ben Hsieh got review on bug 510991, which removes some stats() from the startup path.
• I’m no longer able to reproduce the cold startup regression from 3.5 to 3.6. I turned of SuperFetch, which resulted in more stable numbers, but there was no visible regression, even in builds going back a couple of months. Juan is going to test on Windows XP.
• Ben Hsieh has a patch up for review on bug 511761, which removes needless stat()s of component files, and improves and clarifies current cache invalidation behavior.
• Service caching work in bug 516085, made no progress. Next week I’ll try pushing this to the Places branch to get a better idea of the total performance impact of the change.
• Zack is working on major CSS parser changes in bug 513149.

Projects in a holding pattern:

• Cold Startup Testing: Drew and Alice made a bunch of progress on bug 510587, to create a new Ts that measures cold startup. Alice is working on mobile Talos though, so this is sidelined for a bit. The only issue left is reliable Windows measurement, but we’re not going to block on it, can live with Mac and Linux to start off.
• JARification: David abandoned moving JS modules into a JAR file, since those files are fastloaded. However, since we want things like post-extension-install restarts to be fast, and those cause fastload cache invalidation, we might want to do things like this anyways. I filed a bug for the same treatment for components. These are lower priority, since they’re not the normal startup case. Follow along with all JAR-ification via the tracker bug.
• Startup Timeline: No updates, still not landed. Add [ft] in the whiteboard of your bug w/ the function names you want timed and David will generate it and update the bug.
• Static Analysis: No progress on bug 506128. David needs to file a bug with the final log of named-yet-uncalled functions.
• Dirty Profile Testing: No progress. Need to list scenarios, file bugs for each, generate Talos config patches and profile data, and then move it into Rel-Eng territory. Also, need to get a separate Tinderbox tree, since it’s going to cause a bazillion new columns.
• Joel Reymont noted in bug 513076 that there are serious drawbacks to getting our libraries in the dyld shared cache on Mac, so has deprioritized that work.

Firefox Startup Performance Weekly Summary

Recent activity:

• An improvement in startup time on Mac Leopard of around ~16% (aka: Very Large) was found by Joel Reymont and fixed by Masayuki Nakano in bug 517549. Mac users rejoice!
• The Windows cold startup regression from 3.5 to 3.6 being tracked in bug 517741. Help finding the window for this would be awesome.
• JAR Performance: Alfred Kayser has a new patch on bug 510844 which should increase throughput of JAR file reading, about ready to land.
• Ben Hsieh is making progress removing the needless stats of already-fastloaded components and other stat removals in bug 511761.
• Service caching work is still in progress in bug 516085. Drew’s tests showed no significant win on Tp, but the patch did reduce IO service retrievals by 58%. It might be worth pushing this to the Places branch to get a better idea of the total performance impact of the change.
• Ryan Flint landed bug 499123, combining a bunch of about: pages into a single component, on 1.9.2 branch for Firefox 3.6.
• Bookmarks toolbar: I put a patch on bug 504858 that pushes back the population of the toolbar until after the browser window comes up. Checking into the Places branch showed a 1.5% improvement to warm startup on Windows. Just waiting on review.
• CSS Parsing Time: Zack is working on major CSS parser changes in bug 513149.

Projects in a holding pattern:

• Preference Files: Taras Glek has been working on combining the different preference files into a single one at build time, in bug 507288. The patch is there, and is just trying to find a way to land cleanly.
• Cold Startup Testing: Drew and Alice made a bunch of progress on bug 510587, to create a new Ts that measures cold startup. Alice is working on mobile Talos though, so this is sidelined for a bit. The only issue left is reliable Windows measurement, but we’re not going to block on it, can live with Mac and Linux to start off.
• JARification: David abandoned moving JS modules into a JAR file, since those files are fastloaded. However, since we want things like post-extension-install restarts to be fast, and those cause fastload cache invalidation, we might want to do things like this anyways. I filed a bug for the same treatment for components. These are lower priority, since they’re not the normal startup case. Follow along with all JAR-ification via the tracker bug.
• Startup Timeline: No updates, still not landed. Add [ft] in the whiteboard of your bug w/ the function names you want timed and David will generate it and update the bug.
• PGO for Places, mozStorage and SQLite: Still blocking on Rel-Eng fixing bug 486783, which now has patches!
• Static Analysis: No progress on bug 506128. David needs to file a bug with the final log of named-yet-uncalled functions.
• Dirty Profile Testing: No progress. Need to list scenarios, file bugs for each, generate Talos config patches and profile data, and then move it into Rel-Eng territory. Also, need to get a separate Tinderbox tree, since it’s going to cause a bazillion new columns.
• Joel Reymont noted in bug 513076 that there are serious drawbacks to getting our libraries in the dyld shared cache on Mac, so has deprioritized that work.

Firefox Startup Performance Weekly Summary

Apologies for not posting last week, but I was on vacation. Recent activity:

• I’ve updated the project wiki page with a table of summarized numbers for at-a-glance tracking of progress. I need to disable superfetch and try again on Windows.
• Our measurements of cold startup showed a huge regression on Windows from 3.5 to 3.6. Tracking this in bug 517741.
• JARification: David abandoned moving JS modules into a JAR file, since those files are fastloaded. However, since we want things like post-extension-install restarts to be fast, and those cause fastload cache invalidation, we might want to do things like this anyways. I filed a bug for the same treatment for components. These are lower priority, since they’re not the normal startup case. Follow along with all JAR-ification via the tracker bug.
• JAR Performance: I’m going to test-land bug 510844 on the Places branch to figure out what’s going wrong on the tinderbox. It should increase throughput of JAR file reading, possibly affecting Ts and Tp.
• Removing the needless stats of already-fastloaded components is being looked at more broadly in bug 511761 by Ben Hsieh with help from Taras and others. There are a bunch of dirty ways to trigger component re-registration and fastload cache invalidation, so I filed bug 517515 to consolidate these in an explicit manner.
• Cold Startup Testing: Drew and Alice made a bunch of progress on bug 510587, to create a new Ts that measures cold startup. Alice is working on mobile Talos though, so this is sidelined for a bit. The only issue left is reliable Windows measurement, but we’re not going to block on it, can live with Mac and Linux to start off.
• Preference Files: Taras Glek has been working on combining the different preference files into a single one at build time, in bug 507288. The patch is there, and is just trying to find a way to land cleanly.
• Service Caching: Vlad, Drew and Taras are working on caching some services globally for C++ code. For example, 50% of the GetService calls for the IO service originate from NS_NewURI usage. This work is in bug 516085.

Projects in a holding pattern:

• Startup Timeline: No updates, still not landed. Add [ft] in the whiteboard of your bug w/ the function names you want timed and David will generate it and update the bug.
• PGO for Places, mozStorage and SQLite: Still blocking on Rel-Eng fixing bug 486783, which now has patches!
• Static Analysis: No progress on bug 506128. David needs to file a bug with the final log of named-yet-uncalled functions.
• Dirty Profile Testing: No progress. Need to list scenarios, file bugs for each, generate Talos config patches and profile data, and then move it into Rel-Eng territory. Also, need to get a separate Tinderbox tree, since it’s going to cause a bazillion new columns.
• Joel Reymont noted in bug 513076 that there are serious drawbacks to getting our libraries in the dyld shared cache on Mac, so has deprioritized that work.
• Ryan Flint blogged about various front-end improvements he’s working on that affect startup time.
• CSS Parsing Time: Zack says his work in bug 513149 might improve CSS parsing time by 50%, and it’s next on his list.

Firefox Startup Performance Weekly Summary

Let’s start with the numbers. Vlad spent some time measuring warm startup on the Mac, and it looks very nice: Firefox 3.6 starts up about 10% faster than 3.5, and 3.7 starts up about 20% faster than 3.5. I measured both cold and warm startup on Windows 7, and the picture is not so rosy there. Warm startup is basically flat across 3.5, 3.6 and 3.7, in my tests. Cold startup is… well, weird. Allow me to explain…

My cold startup methodology is simple: On a basically clean install of Windows 7, I’d create a new profile, start the browser up, close it and then reboot. Then I’d measure Firefox startup time after the OS settled. Then I’d do that reboot-and-measure step a bunch more times. Yes, so much fun. I measured Firefox startup time using the method Vlad posted to his blog. This doesn’t do anything to avoid OS optimizations that might persist across reboots, which is a potential flaw, as you’ll see below.

Firefox 3, 3.6 and 3.7 all had consistent numbers across their runs. However, Firefox 3.5 displayed a unique behavior: The first two cold starts were appropriately slow, similar to the cold startup times of the other versions. The subsequent runs were *very* fast for a cold start! I rebooted a few times, and started other versions a few times, then tried a series of 3.5 runs again… and saw the same thing: the first two runs were in the ballpark of other Firefox versions, but subsequent runs were crazy fast. Vlad was talking earlier in the week about a predictive optimization feature that Windows has, learning from an application’s behavior to optimize it’s performance. Possibly that feature works fantastically with Firefox 3.5, and for whatever reason does not work nearly as well with previous or subsequent versions of the app? I’ve clearly crossed into the realm of fantasy here, so alternate theories are welcome. Perhaps I should round-robin through all versions instead of doing serial runs of the same version.

Firefox 3.5 aside, when measuring cold startup I also see the regression that Drew blogged about. We’ll get a bug filed and help from QA to get a regression range for this.

Status updates:

• Startup Timeline: Still not landed, and David found that the timeline instrumentation if put in certain places causes major stability issues. As before, he can provide timeline info for you, just point him at the code by putting [ft] in the whiteboard of your bug.
• PGO for Places, mozStorage and SQLite: Still blocking on Rel-Eng fixing bug 486783. I still need to connect with John O’Duinn and see if we can move this up in the schedule.
• JARification: David started working on moving JS modules into a JAR file. Follow along with all JAR-ification via the tracker bug. Alfred Kayser is working on getting bug 510844 landed, which should increase throughput of JAR file reading.
• JS Component Combining: Further effort is probably not worth it, since these components are fastloaded. Removing the needless file stats when these are first instantiated is what we need to do. I need to talk to Damon and find a XPConnect/Component-loader person for this.
• Static Analysis: No progress on Bug 506128. David needs to file a bug with the final log of named-yet-uncalled functions.
• Dirty Profile Testing: No progress. Need to list scenarios, file bugs for each, generate Talos config patches and profile data, and then move it into Rel-Eng territory. Also, need to get a separate Tinderbox tree, since it’s going to cause a bazillion new columns.
• Cold Startup Testing: Drew and Alice made a bunch of progress on bug 510587, to create a new Ts that measures cold startup. They’re starting with Mac and Linux, and will tackle Windows later. Drew blogged about their work on this.
• Joel Reymont noted in bug 513076 that there are serious drawbacks to getting our libraries in the dyld shared cache on Mac, so is deprioritizing that work for now.
• Ryan Flint blogged about various front-end improvements he’s working on that affect startup time.
• CSS Parsing Time: Zack says his work in bug 513149 might improve CSS parsing time by 50%, and it’s next on his list.