Speedometer - chrome ~20% faster than content_shell

[Łukasz Bendig]

Hello,

I've built chrome and content_shell from the same code base (with the same gn args: is_official_build=true is_debug=false chrome_pgo_phase=2) and noticed that speedometer test is ~20% slower with content_shell. That's pretty strange, I would rather expect content_shell to be faster. Do you know what's reason for that?

Cheers,

Lucas.

[Łukasz Bendig]

I forgot to mention that there is also ~10% difference between non-PGO chrome and non-PGO content_shell. It seems there is "something" that makes content_shell significantly slower in the test.

[Łukasz Bendig]

The problem solved.

It appears that content_shell builds with different thinlto configuration (thinlto_optimize_default) while chrome (thinlto_optimize_max). Not sure why content_shell has it different but it may be valuable information for embedders.

On Wednesday, September 7, 2022 at 7:52:43 PM UTC+2 Łukasz Bendig wrote:

[Justin Novosad]

There might be runtime flags that affect performance.  In particular there may be feature that are being A/B tested chrome.  Content shell uses the flag values defined in testing/variations/fieldtrial_testing_config.json, while chrome builds get their variation configs from a backend service that manages our A/B testing framework.  Also, some of the command line flags have different default states in content_shell vs chrome.

Since you are noticing a difference in graphics performance (probably), the first thing I would do is compare the contents of the about://gpu pages on the two browsers.  Do you see any difference? The speedometer benchmark uses 2d canvas heavily, so pay special attention to anything related to canvas on the about://gpu page.

--
--
Chromium Developers mailing list: chromi...@chromium.org
View archives, change email options, or unsubscribe:
http://groups.google.com/a/chromium.org/group/chromium-dev
---
You received this message because you are subscribed to the Google Groups "Chromium-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chromium-dev...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/chromium-dev/fea1eed4-1c95-438e-a61c-ed0ed72f0ca0n%40chromium.org.