Hi Andrew,
I’m reaching out to you because my message to binary-size@ got a “Delivery Status Notification (Failure)” and your name was suggested in Chromium Slack as a point of contact.
I have a couple of draft CLs that are failing the android-binary-size and Chromium Binary Size bots due to increases in "Normalized APK Size (arm64)," but I'm stumped as to why because the CLs are just moving code from one place to another. I looked at the bot outputs but didn't gain any insights from them. I also tried running `tools/binary_size/diagnose_bloat.py --arm64` as suggested in //docs/speed/binary_size/android_binary_size_trybot.md, but the build for that appears to depend on //chrome/app/theme/google_chrome/BRANDING which I don’t have.
Can you help me understand these increases? The CLs are:
https://chromium-review.googlesource.com/c/chromium/src/+/5651218
https://chromium-review.googlesource.com/c/chromium/src/+/5651570
Thanks!
Kevin
Sounds good, thank you Andrew!
From: Andrew Grieve <agr...@chromium.org>
Sent: Thursday, June 27, 2024 8:41 AM
To: Kevin Babbitt <kbab...@microsoft.com>
Cc: agr...@chromium.org; binary-size <binar...@chromium.org>
Subject: Re: Android binary size increase on refactor CLs
You don't often get email from agr...@chromium.org. Learn why this is important
Thanks for the quick turnaround on the tooling fix!
If blink::RuleFeatureSet::CollectFeaturesFromSelector got bigger due to the compiler changing its inlining decisions, then I’m a little hesitant to second-guess it given that this is a performance sensitive codepath. I admit though I don’t have a good sense for this tradeoff on Android specifically. Adding Rune for his thoughts as well.
Another thought: If the bot includes dependent changes in its analysis… the third change in the stack is https://chromium-review.googlesource.com/c/chromium/src/+/5651670 and that CL passed the binary size check with a delta of +3,668 bytes on arm64. Should I interpret that to mean that the compiler changed its inlining decision again and thus the size increase is a temporary effect?
--
You received this message because you are subscribed to the Google Groups "binary-size" group.
To unsubscribe from this group and stop receiving emails from it, send an email to binary-size...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/binary-size/MW4PR00MB1146F44B964A76C49CC70C5CC0D72%40MW4PR00MB1146.namprd00.prod.outlook.com.
Does the binary-size bot build using PGO? I have some vague recollection that it does, and that someone else who encountered similar issues noticed that it went away after the profiles were regenerated.
DanielOn Thu, 27 Jun 2024 at 17:42, 'Kevin Babbitt' via binary-size <binar...@chromium.org> wrote:Thanks for the quick turnaround on the tooling fix!
If blink::RuleFeatureSet::CollectFeaturesFromSelector got bigger due to the compiler changing its inlining decisions, then I’m a little hesitant to second-guess it given that this is a performance sensitive codepath. I admit though I don’t have a good sense for this tradeoff on Android specifically. Adding Rune for his thoughts as well.
Another thought: If the bot includes dependent changes in its analysis… the third change in the stack is https://chromium-review.googlesource.com/c/chromium/src/+/5651670 and that CL passed the binary size check with a delta of +3,668 bytes on arm64. Should I interpret that to mean that the compiler changed its inlining decision again and thus the size increase is a temporary effect?