Adding label as "TE-Hardware-Dependency" as cjrome-hyd team do not have 4k monitor to triage the issue.

Thanks!!

With most recent version of canary, the problem reproduces in canary as well as standard build.

Safari does not exhibit this behavior.

To help reproduce the issue, please post the contents of your chrome://gpu

Here you go:

Graphics Feature Status
Canvas: Hardware accelerated
Flash: Hardware accelerated
Flash Stage3D: Hardware accelerated
Flash Stage3D Baseline profile: Hardware accelerated
Compositing: Hardware accelerated
Multiple Raster Threads: Enabled
Native GpuMemoryBuffers: Hardware accelerated
Rasterization: Hardware accelerated
Video Decode: Hardware accelerated
Video Encode: Hardware accelerated
VPx Video Decode: Hardware accelerated
WebGL: Hardware accelerated
WebGL2: Hardware accelerated
Driver Bug Workarounds
adjust_src_dst_region_for_blitframebuffer
decode_encode_srgb_for_generatemipmap
disable_framebuffer_cmaa
disable_multimonitor_multisampling
get_frag_data_info_bug
pack_parameters_workaround_with_pack_buffer
regenerate_struct_names
remove_invariant_and_centroid_for_essl3
reset_base_mipmap_level_before_texstorage
scalarize_vec_and_mat_constructor_args
set_zero_level_before_generating_mipmap
unfold_short_circuit_as_ternary_operation
unpack_alignment_workaround_with_unpack_buffer
use_intermediary_for_copy_texture_image
use_shadowed_tex_level_params
use_unused_standard_shared_blocks
Problems Detected
Multisampling is buggy on OSX when multiple monitors are connected: 237931
Applied Workarounds: disable_multimonitor_multisampling
Unfold short circuit on Mac OS X: 307751
Applied Workarounds: unfold_short_circuit_as_ternary_operation
Always rewrite vec/mat constructors to be consistent: 398694
Applied Workarounds: scalarize_vec_and_mat_constructor_args
Mac drivers handle struct scopes incorrectly: 403957
Applied Workarounds: regenerate_struct_names
glGenerateMipmap fails if the zero texture level is not set on some Mac drivers: 560499
Applied Workarounds: set_zero_level_before_generating_mipmap
Pack parameters work incorrectly with pack buffer bound: 563714
Applied Workarounds: pack_parameters_workaround_with_pack_buffer
Alignment works incorrectly with unpack buffer bound: 563714
Applied Workarounds: unpack_alignment_workaround_with_unpack_buffer
copyTexImage2D fails when reading from IOSurface on multiple GPU types.: 581777
Applied Workarounds: use_intermediary_for_copy_texture_image
Mac Drivers store texture level parameters on int16_t that overflow: 610153
Applied Workarounds: use_shadowed_tex_level_params
Limited enabling of Chromium GL_INTEL_framebuffer_CMAA: 535198
Applied Workarounds: disable_framebuffer_cmaa
glGetFragData{Location|Index} works incorrectly on Max: 638340
Applied Workarounds: get_frag_data_info_bug
glTexStorage* are buggy when base mipmap level is not 0: 640506
Applied Workarounds: reset_base_mipmap_level_before_texstorage
Decode and encode before generateMipmap for srgb format textures on macosx: 634519
Applied Workarounds: decode_encode_srgb_for_generatemipmap
Insert statements to reference all members in unused std140/shared blocks on Mac: 618464
Applied Workarounds: use_unused_standard_shared_blocks
adjust src/dst region if blitting pixels outside read framebuffer on Mac: 644740
Applied Workarounds: adjust_src_dst_region_for_blitframebuffer
Mac driver GL 4.1 requires invariant and centroid to match between shaders: 639760, 641129
Applied Workarounds: remove_invariant_and_centroid_for_essl3
Version Information
Data exported	3/3/2017, 8:47:15 AM
Chrome version	Chrome/56.0.2924.87
Operating system	Mac OS X 10.12.3
Software rendering list version	12.06
Driver bug list version	9.24
ANGLE commit id	a4aaa2de57dc
2D graphics backend	Skia/56 bf2d9e02d58ea01f1c239f7e2fc024cba140ccb1
Command Line Args	Chrome.app/Contents/MacOS/Google Chrome --flag-switches-begin --secondary-ui-md --flag-switches-end
Driver Information
Initialization time	41
In-process GPU	false
Sandboxed	true
GPU0	VENDOR = 0x1002, DEVICE= 0x67ef *ACTIVE*
GPU1	VENDOR = 0x8086, DEVICE= 0x191b
Optimus	false
AMD switchable	true
Driver vendor	
Driver version	1.48.21
Driver date	
Pixel shader version	4.10
Vertex shader version	4.10
Max. MSAA samples	8
Machine model name	MacBookPro
Machine model version	13.3
GL_VENDOR	ATI Technologies Inc.
GL_RENDERER	AMD Radeon Pro 455 OpenGL Engine
GL_VERSION	4.1 ATI-1.48.21
GL_EXTENSIONS	GL_ARB_blend_func_extended GL_ARB_draw_buffers_blend GL_ARB_draw_indirect GL_ARB_ES2_compatibility GL_ARB_explicit_attrib_location GL_ARB_gpu_shader_fp64 GL_ARB_gpu_shader5 GL_ARB_instanced_arrays GL_ARB_internalformat_query GL_ARB_occlusion_query2 GL_ARB_sample_shading GL_ARB_sampler_objects GL_ARB_separate_shader_objects GL_ARB_shader_bit_encoding GL_ARB_shader_subroutine GL_ARB_shading_language_include GL_ARB_tessellation_shader GL_ARB_texture_buffer_object_rgb32 GL_ARB_texture_cube_map_array GL_ARB_texture_gather GL_ARB_texture_query_lod GL_ARB_texture_rgb10_a2ui GL_ARB_texture_storage GL_ARB_texture_swizzle GL_ARB_timer_query GL_ARB_transform_feedback2 GL_ARB_transform_feedback3 GL_ARB_vertex_attrib_64bit GL_ARB_vertex_type_2_10_10_10_rev GL_ARB_viewport_array GL_EXT_debug_label GL_EXT_debug_marker GL_EXT_depth_bounds_test GL_EXT_texture_compression_s3tc GL_EXT_texture_filter_anisotropic GL_EXT_texture_mirror_clamp GL_EXT_texture_sRGB_decode GL_APPLE_client_storage GL_APPLE_container_object_shareable GL_APPLE_flush_render GL_APPLE_object_purgeable GL_APPLE_rgb_422 GL_APPLE_row_bytes GL_APPLE_texture_range GL_ATI_texture_mirror_once GL_NV_texture_barrier
Disabled Extensions	
Window system binding vendor	
Window system binding version	
Window system binding extensions	
Direct rendering	Yes
Reset notification strategy	0x0000
GPU process crash count	0
Compositor Information
Tile Update Mode	Zero-copy
Partial Raster	Enabled
GpuMemoryBuffers Status
ATC	Software only
ATCIA	Software only
DXT1	Software only
DXT5	Software only
ETC1	Software only
R_8	GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
RG_88	Software only
BGR_565	Software only
RGBA_4444	Software only
RGBX_8888	Software only
RGBA_8888	GPU_READ, SCANOUT
BGRX_8888	GPU_READ, SCANOUT
BGRA_8888	GPU_READ, SCANOUT, GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
YVU_420	Software only
YUV_420_BIPLANAR	GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
UYVY_422	GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT

Is it possible we are discarding texture under memory pressure?

Does someone have a 4k monitor to reproduce?

rutt@ can you take a trace of this happening, with the cc, cc.debug, cc.debug.picture, cc.debug.quads categories enabled? https://www.chromium.org/developers/how-tos/trace-event-profiling-tool

rutt@ Chrome 58.0.3021.0 is Dev channel.  Is this what you expected?

> rutt@ Chrome 58.0.3021.0 is Dev channel.  Is this what you expected?

58.0.3021.0 seems to not be an official released Dev verion.  Could you please check latest Dev release 58.0.3026.3?  Thanks.

Attaching a trace

Deleted: trace_Tue_Mar_07_2017_6.55.15_AM.json.gz 14.5 MB

trace_Tue_Mar_07_2017_6.55.15_AM.json.gz 14.5 MB Download

Thank you for providing more feedback. Adding requester "danakj@chromium.org" to the cc list and removing "Needs-Feedback" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Please note this was done with the stable release version, not a dev version.  I do not know how to go back and forth without deleting my profile an starting from scratch.

About says: Version 56.0.2924.87 (64-bit)

Could you attach a cellphone video of the flickering, so we know what to look for?

Attaching cellpone video.

Deleted: IMG_6354.MOV 13.2 MB

IMG_6354.MOV 13.2 MB Download

Thank you for providing more feedback. Adding requester "ccameron@chromium.org" to the cc list and removing "Needs-Feedback" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Assigning to ccameron@ to decide on the next steps on this.

vmpstr@ can you take a look at the video and check if this is a tiling issue? (doesn't look like it because the flickering happens for half screen width but gpu raster tiles are full width right?)

M58 stable is approaching soon, can we have a fix before M59 hits stable.

Interesting -- I'll test on the 4k that we have in SFO when I get a chance.

Re #14, does this reproduce with Canary (Canary doesn't break side-by-side installs).

I'd guess this to be an out-of-memory situation, but I'm not sure.

Also of note is that it looks like we're using 256x256 tiles on this machine, which probably doesn't help performance.

Friendly ping!!
ccameron@, Could you please check & update the thread as per comment#20.
Thank you!

Using a pop out window for message composition is a reasonable workaround.  I wish there were a way to make that the default.

We may just need to bump memory limits on these systems - if I understand correctly, retina scaling on Mac always renders to 2x scales (confirmed on my MBP). So if you run at a "non-native" scale, it doubles that scale (even if bigger than the screen), then does nice downsampling.

This means that with retina scaling at 3360 x 1980, we are rendering to a virtual 6720x3960 screen. This lines up with the tile sizes in the trace.

This means that a single screen of tiles uses 106MB of memory.

I think that on systems where we are rendering to a screen > 3k pixels (or some large threshold), we should probably bump our tile memory limits by a factor of 2...

That sounds reasonable to me

I can test this if you guys like, but not until next week.  I am traveling
until then.

 Issue 709875  has been merged into this issue.

I can reproduce this - it's definitely just an out of memory case described in 23 (I can get my gmail to match the uploaded video almost exactly).

I'm looking at various ways to improve memory usage (both by increasing the limits as well as cutting back our allocations) - will have an update in a bit.

 Issue 719938  has been merged into this issue.

Reminder that M59 Stable is launch is coming soon (less than 2 weeks)! Your bug is labelled as Stable ReleaseBlock, pls make sure to land the fix and get it merged into the release branch ASAP so it gets enough baking time in Beta (before Stable promotion). Thank you!

I am removing ReleaseBlock-Stable tag, since this issue has been around in M58 and is not truly a release blocker. Please feel free to add it back if you think otherwise. 

I don't understand. I thought this issue was both reproduced and that a
simple fix was available. Why not just push out?

I second that, please reconsider a fix for stable if it is a one-liner.

I have also been eagerly watching this issue, currently this causes me to restart chrome every two days.

Apologies for the delay - was hoping to come up with a more robust fix, but I think a quick patch in the meantime is warranted. I'll try to get something out in the next week.

Some image heave websites are really bad, example: http://www.kathmandu.com.au/womens/clothing/pants-and-shorts.html

Running: Version 59.0.3071.115 (Official Build) (64-bit)

Visual Studio Code is badly affected too.

Why is this still broken?  I thought the fix was supposed to be pretty easy
and Chrome on a 4k display is pretty broken for gmail until you know what
to do.

vmpstr@, can you add some code to double the bytes_limit_when_visible here when on a high-resolution display (display width > 3k?). I don't have the bandwidth to handle this right now.

Relevant code:
https://cs.chromium.org/chromium/src/content/renderer/gpu/render_widget_compositor.cc?rcl=7c9efb2c74e9e20e4c4c23eafdeff07da09b4745&l=621

Not sure how to detect screen resolution in the renderer though?

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/0375ae80fe1d94e9d0fc175ce4312d1b3a6bc168

commit 0375ae80fe1d94e9d0fc175ce4312d1b3a6bc168
Author: Vladimir Levin <vmpstr@chromium.org>
Date: Tue Jul 18 03:06:40 2017

content: Double tile memory for large monitors (4k)

With the current memory limits, the compositor experiences frequent out
of memory issues when running on 4k monitors. This patch doubles the
memory allowed to use when the screen resolution is high enough
(>= 3500px by default).

R=ericrk@chromium.org, piman@chromium.org

Bug: 695427
Change-Id: I73884ae93d7ccd32fd76b8e40f3a596a02e4eaae
Reviewed-on: https://chromium-review.googlesource.com/575445
Reviewed-by: Eric Karl <ericrk@chromium.org>
Reviewed-by: Antoine Labour <piman@chromium.org>
Commit-Queue: Vladimir Levin <vmpstr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#487364}
[modify] https://crrev.com/0375ae80fe1d94e9d0fc175ce4312d1b3a6bc168/content/renderer/gpu/render_widget_compositor.cc
[modify] https://crrev.com/0375ae80fe1d94e9d0fc175ce4312d1b3a6bc168/content/renderer/gpu/render_widget_compositor.h
[modify] https://crrev.com/0375ae80fe1d94e9d0fc175ce4312d1b3a6bc168/content/renderer/gpu/render_widget_compositor_unittest.cc

The patch in #39 should alleviate the out of memory problems on large monitors. ericrk@ was there a thought for some sort of a more robust solution in the future, or do you think this is enough to close this as fixed?

When will this get pushed out?

This should be in the next Canary release, and it should make it into 61. According to https://www.chromium.org/developers/calendar, it should make it into stable around Sept 6, but will be available in other channels before that.

I also have this issue on a 1080p display Intel graphics. 

The issue is still reproducible on the following webpage: http://www.kathmandu.com.au/womens/clothing/pants-and-shorts.html
Just maximize the browser view on the external display and move the cursor around. For me the menu on the left side is still flashing.

Chrome: Version 61.0.3163.79 (Official Build) (64-bit)
Scaling: 4k scaled to 3008x1692 (middle text size setting)
HW MacBook Pro, Retina, 15-inch, Mid 2015; Intel Iris Pro 1536 MB; Dell 4k monitor