Closed Bug 1311985 Opened 8 years ago Closed 8 years ago

massive performance decrease with certain Kongregate games when using Flash async drawing mode

Categories

(Core Graveyard :: Plug-ins, defect)

49 Branch
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED DUPLICATE of bug 1340276

People

(Reporter: c.simon, Unassigned)

References

Details

(Keywords: regression, Whiteboard: repo?)

Attachments

(2 files)

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:49.0) Gecko/20100101 Firefox/49.0 Build ID: 20161019084923 Steps to reproduce: Use any Kongregate flash game with Firefox 49.0.2 and Flash 23.0.0.195. Actual results: AsyncDrawing seems to be used, even if wmode is set to "window" Expected results: According to the Adobe page for 23.0.0.198 (current beta) (https://forums.adobe.com/thread/2224252) wmode window should not use AsyncDrawing.
Our intention is to stop supporting windowed-mode plugins entirely, so all Flash instances will use async drawing or fall back to windowless. We intend to remove windowed-mode plugin support in Firefox 53. I'd like to morph this bug to be specifically about performance problems when a specific game is using async drawing mode. Could you provide the following details: * A specific game URL that we can use for testing * Attach the contents of your about:support data to this bug, in particular I'm interested in the graphics configuration data I need to figure out whether this is a regression in all configurations, or something specific to certain configs which may not be hitting accelerated codepaths.
Component: Untriaged → Plug-ins
Flags: needinfo?(c.simon)
Product: Firefox → Core
Summary: AsyncDrawing is enabled, even though the wmode is set to "window", causing massive performance decrease → massive performance decrease with certain Korgregate games when using async drawing mode
Blocks: 1229961
Keywords: regression
Attached image kongregateGameMemory49.png (deleted) —
I tested this on http://www.kongregate.com/games/Ninjakiwi/bloons-monkey-city I noticed a spike in Memory on FX49. FX48 was fine. Graphics Features Compositing Direct3D 11 Asynchronous Pan/Zoom wheel input enabled WebGL Renderer Google Inc. -- ANGLE (NVIDIA GeForce GTX 745 Direct3D11 vs_5_0 ps_5_0) Hardware H264 Decoding Yes; Using D3D11 API Direct2D true DirectWrite true (6.3.9600.18468) GPU #1 Active Yes Description NVIDIA GeForce GTX 745 Vendor ID 0x10de Device ID 0x1382 Driver Version 9.18.13.3266 Driver Date 2-16-2014 Drivers nvd3dumx,nvwgf2umx,nvwgf2umx nvd3dum,nvwgf2um,nvwgf2um Subsys ID 106510de RAM 4095 Diagnostics AzureCanvasAccelerated 0 AzureCanvasBackend direct2d 1.1 AzureContentBackend direct2d 1.1 AzureFallbackCanvasBackend cairo
Attached image kongregateGameMemory48.PNG (deleted) —
Flags: needinfo?(twalker)
Whiteboard: repo?
Here are the peak numbers I observed for the FlashPlayerPlugin on Win 7 in a VM on Mac. On latest Nightly (21061028) with asyncdrawing on: -- virtual memory reserved: ~1059MB -- Physical memory committed: ~510MB withasyncdrawing off: -- virtual memory reserved: ~856MB -- Physical memory committed: ~450MB On Firefox 49.0.2 -- virtual memory reserved: ~768MB -- Physical memory committed: ~475MB On Firefox 48.0.2 -- virtual memory reserved: ~660MB -- Physical memory committed: ~493MB There doesn't appear to be a regression in the use of physical memory. But these numbers are observations at various times when I felt the game was putting the system under heavier load. In general, on this VM under test, the total memory usage would be about 30% with just the Win 7 OS running. Memory usage would then rise to about 70% with Firefox running and would ramp up further to between 85-90% when the flash game was running in the browser. Memory usage would briefly spike even higher when the game was doing something like loading up a new challenge. These general measurements would be similar no matter which browser was in use. During those peak memory use times the Firefox UI would become laggy. ie, press new tab button a few times and noting would appear to happen, then a few seconds later the tabs would spawn. Note: the virtual memory reserved would slowly ramp up over time. The nightly number is high it was under test by far the longest as I left it running over lunch.
Flags: needinfo?(twalker)
Latest version of Shockwave Flash seems to be worse, not particularly in memory usage, but in hanging the browser. It is so bad with this version, that it's not even possible to login to play the game. File: NPSWF32_23_0_0_205.dll Path: C:\Windows\system32\Macromed\Flash\NPSWF32_23_0_0_205.dll Version: 23.0.0.205 State: Enabled Shockwave Flash 23.0 r0
Tracy can you please retest here with the latest flash and see if things have improved. I've been testing various kongregate games in other bugs and I'm not seeing a lot of issues with perf or memory.
Flags: needinfo?(c.simon) → needinfo?(twalker)
Nothing has improved since my last comment #5. It is still not possible to log in to the http://www.kongregate.com/games/Ninjakiwi/bloons-monkey-city game. The login button is unresponsive. Tested with ADP 25.0.0.127 in 32-bit Nightly 55.0a1, 20170315030215 on Win 10 Surface Pro. Also it took over a minute for the audio/music of said login page to unload after closing the tab with the kongregate page in it. meaning, the music was still playing well after it's tab was closed. Don't have these issues if asyncdrawing is disabled.
Flags: needinfo?(twalker)
Summary: massive performance decrease with certain Korgregate games when using async drawing mode → massive performance decrease with certain Kongregate games when using Flash async drawing mode
Status: UNCONFIRMED → RESOLVED
Closed: 8 years ago
Resolution: --- → DUPLICATE
Product: Core → Core Graveyard
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: