1279433 - Page disappears while scrolling, it reappears when still

Status: VERIFIED FIXED

(Firefox for Android Graveyard :: Toolbar, defect, P1)

Description

[Ilgaz Öcal]

Attachment: Current device hardware& software information

User Agent: Mozilla/5.0 (Android 5.0; Mobile; rv:48.0) Gecko/48.0 Firefox/48.0
Build ID: 20160607191837

Steps to reproduce:

1) Navigate to a page which needs scrolling such as slashdot.org
2) Scroll


Actual results:

Page disappears while it is scrolling, appearing white and re-appears when scrolling action ends.


Expected results:

Page should smoothly scroll displaying elements

Comment 1

[Ilgaz Öcal]

I am using Firefox beta on Android as main browser for about a year and can say this wasn't happening before 48 beta

Comment 2

[Ilgaz Öcal]

Bug in action (30+ MB mp4 screen capture on the device)
https://goo.gl/photos/FEBH5cFSADd1Aeoa8

Comment 3

[Ilgaz Öcal]

Attachment: Firefox about: support page (as copy to clipboard didn't work)

Comment 4

[Ilgaz Öcal]

There is a improvement on current nightly major version (50) , it still struggles while scrolling but not at "show white screen" level. Comparing to other current Android browsers, it is still unacceptable performance. While on it, I have a "rooted" Asus Zenfone 2 (Intel based, 4GB RAM) here and available for any testing. System profile is in first attachment.

Comment 5

[Kartikaya Gupta (email:[e-mail])]

Thanks for the info! We were looking at this a bit this morning. It happens on some devices but not others. On my Nexus 4 I was able to fix one source of the problem by going to about:config and changing the displayport multipliers so that the displayport doesn't keep getting resized. Could you please this as well, to see if it helps? Go to about:config and change the following prefs:

apz.x_stationary_size_multiplier: should be 1.5, change to 1.25
apz.y_stationary_size_multiplier: should be 1.8, change to 1.5

This makes the "stationary" multiplier values match the "skate" multiplier values, and (on my device at least) causes less repainting which makes things faster.

One other thing you could try is set layers.low-precision-buffer to false (it's true by default) and see if that helps. Feedback on how these settings change your experience would be very useful. Thanks!

Comment 6

[Ilgaz Öcal]

I did test the about:config tweaks, there is a very slight performance increase but white display etc continues. While on it, here is my CPU&GPU info

<<< CPU >>>

SoC Model: Intel Atom Z3580
SoC Codename: Anniedale (Moorefield)
Core Architecture: Intel Silvermont
Manufacturing Process: 22 nm
Instruction Set: 64-bit x86
CPU Revision: Intel 506A0h
CPU Cores: 4
CPU Clock Range: 500 - 2333 MHz
Core 1 Clock: 1833 MHz
Core 2 Clock: 1833 MHz
Core 3 Clock: 1333 MHz
Core 4 Clock: 1333 MHz
CPU Utilization: 12 %
L1I Cache: 4x 32 KB
L1D Cache: 4x 24 KB
L2 Cache: 2 MB
Scaling Governor: interactive
Supported ABIs: x86, armeabi-v7a, armeabi
Supported 32-bit ABIs: x86, armeabi-v7a, armeabi

<<< Display >>>

Screen Resolution: 1080 × 1920
Technology: IPS LCD
Screen Size: 68 mm × 122 mm
Screen Diagonal: 5.5 inches
Pixel Density: 401 dpi (xxhdpi)
xdpi / ydpi: 403 / 403 dpi
GPU Vendor: Imagination Technologies
GPU Renderer: PowerVR Rogue G6430
Current GPU Clock: 457 MHz
Scaling Governor: simple_ondemand
Refresh Rate: 60 Hz
Default Orientation: Portrait
OpenGL ES Version: 3.1
GPU Version: OpenGL ES 3.1 build 1.4@3283119
OpenGL ES Extensions: GL_EXT_debug_marker
GL_EXT_blend_minmax
GL_EXT_color_buffer_float
GL_EXT_discard_framebuffer
GL_EXT_draw_buffers
GL_EXT_multi_draw_arrays
GL_EXT_multisampled_render_to_texture
GL_EXT_occlusion_query_boolean
GL_EXT_read_format_bgra
GL_EXT_robustness
GL_EXT_separate_shader_objects
GL_EXT_shader_framebuffer_fetch
GL_EXT_shader_texture_lod
GL_EXT_texture_filter_anisotropic
GL_EXT_texture_format_BGRA8888
GL_EXT_texture_rg
GL_EXT_texture_sRGB_decode
GL_IMG_multisampled_render_to_texture
GL_IMG_program_binary
GL_IMG_read_format
GL_IMG_shader_binary
GL_IMG_texture_compression_pvrtc
GL_IMG_texture_compression_pvrtc2
GL_IMG_texture_format_BGRA8888
GL_IMG_texture_npot
GL_KHR_blend_equation_advanced
GL_KHR_blend_equation_advanced_coherent
GL_KHR_debug
GL_OES_compressed_ETC1_RGB8_texture
GL_OES_depth24
GL_OES_depth_texture
GL_OES_EGL_image
GL_OES_EGL_image_external
GL_OES_EGL_sync
GL_OES_element_index_uint
GL_OES_fragment_precision_high
GL_OES_get_program_binary
GL_OES_mapbuffer
GL_OES_packed_depth_stencil
GL_OES_required_internalformat
GL_OES_rgb8_rgba8
GL_OES_shader_image_atomic
GL_OES_standard_derivatives
GL_OES_surfaceless_context
GL_OES_texture_float
GL_OES_texture_half_float
GL_OES_texture_npot
GL_OES_texture_stencil8
GL_OES_texture_storage_multisample_2d_array
GL_OES_vertex_array_object

Comment 7

[Kartikaya Gupta (email:[e-mail])]

Thanks. I tracked down the root cause of at least one problem in bug 1262807, it is fixed in the latest nightly build - can you try that and let me know if the behaviour is any better?

Comment 8

[Ilgaz Öcal]

There is significant improvement on this device with current nightly, here is the very same site (slashdot) scrolling performance video capture. However you will also notice some white screen issues when scrolling backwards.

https://goo.gl/photos/81Y2UXNghFapsSwo6

Comment 9

[Randall Barker [:rbarker]]

The native android fling animation currently allows the velocity to go much higher than the previous version. There is a patch to limit this for Bug 1280656. This may help some with checker boarding however the higher fling speeds mean it is still easier to out run the painting.

Comment 10

[Milan Sreckovic [:milan] (needinfo for best results)]

We will take the fix from bug 1262807 up to 48 if approved, but not tracking the difference for 48.

Comment 11

[Kartikaya Gupta (email:[e-mail])]

Is the latest nightly any better now that bug 1280656 has landed?

Comment 12

[Ilgaz Öcal]

It is like the Firefox 47, scrolling is not smooth as Chrome but there are no white screen type horrible performance issues.

Comment 13

[Kartikaya Gupta (email:[e-mail])]

Thanks for confirming! I'll close this bug and make sure we uplift bug 1280656.

Comment 14

[Alex Caziuc]

I verified this on Galaxy Note 5(6.0.1) and can't see this issue anymore and I would like to put it on verified fixed.

Ilgaz can you please verify this on your device to see if it is fixed?

Thanks

Comment 15

[Ilgaz Öcal]

It really seems to have been fixed, there is no white screen issue on this Asus Zenfone 2 (Atom running Android 6.01) anymore. Current beta, build ID 20160905140648.

Comment 16

[Alex Caziuc]

Thanks Ilgaz. I'll put it verified fixed since it is fixed.