I tested on my mac (10.13.3, AMD Radeon R9) with Chrome 64.0.3282.140 and I can't reproduce this issue.

It's probably GPU specific. Reporter, you will need to provide the about:gpu content on the machine where you can reproduce this bug.

Attaching the about://gpu log.

Deleted: gpu.htm 72.7 KB

gpu.htm 72.7 KB View Download

This is the Intel Integrated GPU in the dual-GPU NVIDIA MacBook Pros. We have one of these in house and could use gfxCardStatus to force it onto the integrated GPU.

I have checked with another macbook. Attaching the about://gpu log of that macbook.

I checked in another macbook where the issue is reproducible. Attaching about://gpu log of that macbook.

Deleted: gpu.html 64.3 KB

gpu.html 64.3 KB View Download

Thanks for updating chrome://gpu info.

Unable to reproduce this issue on 64.0.3282.140 using Mac 10.13.3 with steps mentioned below.

1. Installed 63.0.3239.132 and navigated to  https://threejs.org/examples/webgl_lights_rectarealight.html and observed FPS as 60.
2. Upgraded to 64.0.3282.140 and navigated to  https://threejs.org/examples/webgl_lights_rectarealight.html and observed FPS as 60. No FPS drop is seen. Attaching screencast for reference.

@zmo: Please check the attached GPU contents of reporter.

Thanks!

Deleted: 809422.mp4 3.5 MB

	809422.mp4 3.5 MB View Download

I can reproduce when launching Chrome with --force_integrated_gpu=1 on my macbook (dual GPU, AMD Radeon R9 / Intel Iris Pro). This is with Stable and Canary.

Can someone do a bisect on a MacBook Pro (Retina, 15-inch, Mid 2015)? Run Chrome with --force_integrated_gpu=1 (and made sure you don't have any other apps running that might trigger discrete GPU, say, another Chrome running WebGL).

Rajshree, could you please give a try using the flag mentioned in #11. Looks like you have a Retina in the inventory, but not sure about the GPU card. 

By any chance if you cannot repro, please route to Inhouse/ MTV.

Able to reproduce the issue on mac 10.12.6 MacBook Pro (Retina, 15-inch, Mid 2015) using chrome reported version #64.0.3282.140 and latest canary #66.0.3342.0.

Bisect Information:
=====================
Good build: 64.0.3271.0
Bad Build : 64.0.3272.0

Change Log URL: 
https://chromium.googlesource.com/chromium/src/+log/8e638519a02830035b58a02dd23dc5bb86488d34..6eda055fb80e52bd7808e4e5c94755c0bd6e4f9c

https://chromium.googlesource.com/angle/angle.git/+log/bb27c3a1483c..2c7f34c833b4

From the above change log suspecting below change
Change-Id: Iec727821d8137db56b440ddbe007879b1b55f61f
Reviewed-on: https://chromium-review.googlesource.com/707195

oetuaho@ - Could you please check whether this is caused with respect to your change, if not please help us in assigning it to the right owner.
Note: Adding stable blocker for M-64 as it seems to be a recent regression. Please feel free to remove the same if not appropriate

Thanks...!!

I can conform by running Chrome on my Retina with switches --force_integrated_gpu, we see this bug; running with --force_integrated_gpu --dont_use_loops_to_initialize_variables, the bug no longer manifests.

Olli, should we just add this driver bug workaround for Mac Intel?

By the way, thank you for the bisecting!

+ Abdul

Interesting that the loops to initialize variables affects performance in a noticable way. I'd say turning on dont_use_loops_to_initialize_variables on Mac Intel is a reasonable solution. Maybe revisiting the loop initialization in general would be a good idea, it might trigger in some cases where it's not reasonable to use it.

+  ellyjones@ (Mac TL)to keep her in loop.

I'll put in a driver bug list entry.

Friendly ping to get an update on update on this issue as it is marked as stable blocker for M65.

Thanks..!

M65 Stable promotion is coming VERY soon. Your bug is labelled as Stable ReleaseBlock, pls make sure to land the fix and request a merge  into the release branch ASAP. Thank you.

CL is in queue: https://chromium-review.googlesource.com/c/chromium/src/+/917018

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/04dc3fe0ebce31ddfe0249be05e7658177235605

commit 04dc3fe0ebce31ddfe0249be05e7658177235605
Author: Zhenyao Mo <zmo@chromium.org>
Date: Tue Feb 13 22:25:50 2018

Turn off "shader var init in a loop" on Mac Intel to avoid perf regression.

BUG= 809422 
TEST=manual
R=kbr@chromium.org,oetuaho@nvidia.com

Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
Change-Id: I665ae361cc9485c120ca2fdb5d174213e083042b
Reviewed-on: https://chromium-review.googlesource.com/917018
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Zhenyao Mo <zmo@chromium.org>
Cr-Commit-Position: refs/heads/master@{#536514}
[modify] https://crrev.com/04dc3fe0ebce31ddfe0249be05e7658177235605/gpu/config/gpu_driver_bug_list.json

CL listed at #23 just to landed in trunk. Pls update the bug with canary result tomorrow.

The NextAction date has arrived: 2018-02-14

Able to reproduce the issue on mac 10.12.6(retina) using chrome reported version #64.0.3282.140 as per comment #0 and #11
Verified the fix on Mac 10.12.6(retina) using Chrome version #66.0.3347.0 as per the comment #0 and #11.
Attaching screen cast for reference.
Observed that FPS did not drop and remained at 33-60.
Hence, the fix is working as expected. 
Adding the verified labels.

Thanks...!!

Deleted: 809422.mp4 3.0 MB

	809422.mp4 3.0 MB View Download

How safe is the change listed at #23 to merge to M65?

It's only a driver bug workaround data entry, so I'd say pretty safe.

Approving merge to M65 branch 3325 based on comments #27 and #29. Please merge ASAP. Thank you.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/75d633748b03e556edc8443ee7f9c81e5adfb9d6

commit 75d633748b03e556edc8443ee7f9c81e5adfb9d6
Author: Zhenyao Mo <zmo@chromium.org>
Date: Wed Feb 14 21:06:00 2018

Turn off "shader var init in a loop" on Mac Intel to avoid perf regression.

BUG= 809422 
TEST=manual
R=kbr@chromium.org, oetuaho@nvidia.com
TBR=zmo@chromium.org

(cherry picked from commit 04dc3fe0ebce31ddfe0249be05e7658177235605)

Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
Change-Id: I665ae361cc9485c120ca2fdb5d174213e083042b
Reviewed-on: https://chromium-review.googlesource.com/917018
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Zhenyao Mo <zmo@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#536514}
Reviewed-on: https://chromium-review.googlesource.com/919424
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Cr-Commit-Position: refs/branch-heads/3325@{#469}
Cr-Branched-From: bc084a8b5afa3744a74927344e304c02ae54189f-refs/heads/master@{#530369}
[modify] https://crrev.com/75d633748b03e556edc8443ee7f9c81e5adfb9d6/gpu/config/gpu_driver_bug_list.json

Tested this issue on Mac OS 10.13.3 (Retina, 15-inch, Mid 2015) on the reported version 64.0.3282.140 and latest M-65 Chrome build 65.0.3325.88 by following the steps mentioned in the original comment and #11.
Able to reproduce this issue on the reported version 64.0.3282.140 and the issue is fixed on the latest M-65 build 65.0.3325.88.

On launching Chrome with the flag '--force_integrated_gpu=1' and navigating to the given html page, can observe no drop in FPS.
Attached is the screen cast for reference.

Hence adding TE verified labels as the fix is working as intended.

Thanks..

Deleted: 809422_M65.mp4 1.6 MB

	809422_M65.mp4 1.6 MB View Download