protonscr

TF2: Huge, lasting drop in framerate

steamclosed MOVED: Team Fortress 2reviewed
ValveSoftware/steam-for-linux#140 · opened 2012-12-20 by chmod222 · updated 2013-02-15 · 24 comments · github
Cchmod222 2012-12-20 github

Occasionally, sometimes instantly on round start / game join, sometimes halfway into a round, my framerate drops from mostly (see issue #138) playable levels into unplayable 10 - 15 frames per second.

The game will stay that way until I force a reinitialization of the graphics by changing any advanced graphics setting (e.g. Motion Blur, Filtering) or completely restart the game (stuttering persists even in main menu resulting in delayed button clicks and hover states).

One thing I now know is that it's not an out of memory condition, TF2's memory consumption stayed the same pre- and post bug and was overall well below my total RAM (2.4 GiB usage, 8 GiB avail). It also is not connected to the 100%-CPU bug that was fixed yesterday, because it since happened again.

This bug makes the game pretty much unplayable for me, sadly.

Processor Information:
    Vendor:  GenuineIntel
    Speed: 3301 Mhz
    4 logical processors
    4 physical processors
    HyperThreading:  Unsupported
    FCMOV:  Supported
    SSE2:  Supported
    SSE3:  Supported
    SSSE3:  Supported
    SSE4a:  Unsupported
    SSE41:  Supported
    SSE42:  Supported

Network Information:
    Network Speed:  

Operating System Version:
    Ubuntu 12.10 (64 bit) [Unity]
    Kernel Name:  Linux
    Kernel Version:  3.5.0-21-generic
    X Server vendor:  The X.Org Foundation
    X Server release:  11300000

Video Card:
    Driver:  NVIDIA Corporation GeForce GTX 560 Ti/PCIe/SSE2

    Driver Version:  4.3.0 NVIDIA 310.14
    Desktop Color Depth: 24 bits per pixel
    Monitor Refresh Rate: 60 Hz
    VendorID:  0x10de
    DeviceID:  0x1200
    Number of Monitors:  2
    Number of Logical Video Cards:  1
    Primary Display Resolution:  1920 x 1080
    Desktop Resolution: 3600 x 1080
    Primary Display Size: 20,51" x 11,54"  (23,50" diag)
                                            52,1cm x 29,3cm  (59,7cm diag)
    Primary Bus: PCI Express 16x
    Primary VRAM: 1023 MB
    Supported MSAA Modes:  2x 4x 8x 16x 

Sound card:
    Audio device: 20K2

Memory:
    RAM:  7956 Mb

Miscellaneous:
    UI Language:  English
    LANG:  de_DE.UTF-8
    Microphone:  Not set
    Total Hard Disk Space Available:  679553 Mb
    Largest Free Hard Disk Block:  184560 Mb

Installed software:

Recent Failure Reports:
?ghost 2012-12-20 github

I just had this happen. Game is still running, in case someone has any ideas on what data to collect. I am on #steamlug on FreeNode

EDIT: Adding system information
EDIT2: Interesting is that now it's just hl2_linux process which is at over 100% CPU, not steam. If I enable multicore rendering, two cores are used (200% CPU), with same fps.

Processor Information:
Vendor: GenuineIntel
Speed: 2499 Mhz
4 logical processors
4 physical processors
HyperThreading: Unsupported
FCMOV: Supported
SSE2: Supported
SSE3: Supported
SSSE3: Supported
SSE4a: Unsupported
SSE41: Supported
SSE42: Unsupported

Network Information:
Network Speed:

Operating System Version:
Debian GNU/Linux testing (wheezy) (64 bit)
Kernel Name: Linux
Kernel Version: 3.2.0-4-amd64
X Server vendor: The X.Org Foundation
X Server release: 11204000

Video Card:
Driver: NVIDIA Corporation GeForce GTX 260/PCIe/SSE2

Driver Version:  3.3.0 NVIDIA 310.19
Desktop Color Depth: 24 bits per pixel
Monitor Refresh Rate: 59 Hz
VendorID:  0x10de
DeviceID:  0x5e2
Number of Monitors:  1
Number of Logical Video Cards:  1
Primary Display Resolution:  1680 x 1050
Desktop Resolution: 1680 x 1050
Primary Display Size: 18.62" x 11.65"  (21.93" diag)
                                        47.3cm x 29.6cm  (55.7cm diag)
Primary Bus: PCI Express 16x
Primary VRAM: 896 MB
Supported MSAA Modes:  2x 4x 8x 16x 

Sound card:
Audio device: SigmaTel STAC9721,23

Memory:
RAM: 5984 Mb

Miscellaneous:
UI Language: English
LANG: en_US.UTF-8
Microphone: Not set
Total Hard Disk Space Available: 102350 Mb
Largest Free Hard Disk Block: 61332 Mb

Installed software:

Recent Failure Reports:
Tue Dec 18 16:51:23 2012 GMT: file ''/tmp/dumps/crash_20121218175120_1.dmp'', upload yes: ''CrashID=bp-7099f4c7-75c3-4aad-965c-410442121218''
Wed Dec 19 22:16:29 2012 GMT: file ''/tmp/dumps/assert_20121219231625_1.dmp'', upload yes: ''CrashID=bp-69ea299d-6f3a-42bb-be23-fa6b22121219''

?ghost 2012-12-20 github

I captured about 30 seconds of strace of the hl2_linux process. 99% of it is sched_yield(), repeated hundreds, if not thousands times per second, with some other calls sprinkled in here and there. I can provide the file (200kB compressed) upon request. I also have a gdb backtrace, glxinfo output and pmap for the process.

DDissCent 2012-12-20 github

I can confirm this issue on openSUSE 12.2 x86, NVidia 310.19. It happens mostly after 3-4 map changes and is only solved by restarting the game.

AAnAkkk 2012-12-20 github

It's normal that hl2_linux is at 100% if you are using mat_queue_mode, it's supposed to use all of your CPU and should increase your FPS.

ZZeXx86 2012-12-21 github

Same bug here, MainThrd use 120% CPU and FPS is around 10. It does not affect me in offline game.

DDragonsbrethren 2012-12-22 github

I'm not sure if it's related or not, but I get a similar issue when changing shadows to high or HDR to full from a lower setting. My framerate will go from playable down to ~10 until I restart TF2. If I start TF2 with these already set to their highest, my framerate is normal.

Processor Information:
Vendor: AuthenticAMD
Speed: 2913 Mhz
2 logical processors
2 physical processors
HyperThreading: Unsupported
FCMOV: Supported
SSE2: Supported
SSE3: Supported
SSSE3: Supported
SSE4a: Unsupported
SSE41: Unsupported
SSE42: Unsupported

Network Information:
Network Speed:

Operating System Version:
Linux Mint 13 Maya (64 bit)
Kernel Name: Linux
Kernel Version: 3.2.0-23-generic
X Server vendor: The X.Org Foundation
X Server release: 11103000

Video Card:
Driver: NVIDIA Corporation GeForce GT 240/PCIe/SSE2

Driver Version:  3.3.0 NVIDIA 310.14
Desktop Color Depth: 24 bits per pixel
Monitor Refresh Rate: 60 Hz
VendorID:  0x10de
DeviceID:  0xca3
Number of Monitors:  1
Number of Logical Video Cards:  1
Primary Display Resolution:  1920 x 1440
Desktop Resolution: 1920 x 1440
Primary Display Size: 14.17" x 10.63"  (17.72" diag)
                                        36.0cm x 27.0cm  (45.0cm diag)
Primary Bus: PCI Express 16x
Primary VRAM: 512 MB
Supported MSAA Modes:  2x 4x 8x 16x 

Sound card:
Audio device: Realtek ALC662 rev1

Memory:
RAM: 2506 Mb

Miscellaneous:
UI Language: English
LANG: en_US.UTF-8
Microphone: Not set
Total Hard Disk Space Available: 28223 Mb
Largest Free Hard Disk Block: 1940 Mb

Installed software:

Recent Failure Reports:
Fri Dec 21 22:17:05 2012 GMT: file ''/tmp/dumps/assert_20121221171701_1.dmp'', upload yes: ''CrashID=bp-4a77eb2d-00a5-460b-af67-a21fa2121221''
Fri Dec 21 22:21:06 2012 GMT: file ''/tmp/dumps/crash_20121221172101_1.dmp'', upload yes: ''CrashID=bp-8281e352-40a3-4577-aae4-8fe702121221''
Fri Dec 21 22:24:27 2012 GMT: file ''/tmp/dumps/crash_20121221172414_2.dmp'', upload yes: ''CrashID=bp-296b014d-6353-4140-8ecc-30b162121221''
Fri Dec 21 22:24:29 2012 GMT: file ''/tmp/dumps/crash_20121221172427_1.dmp'', upload yes: ''CrashID=bp-cca028d5-f499-46cc-9b91-0c5e62121221''
Fri Dec 21 22:24:35 2012 GMT: file ''/tmp/dumps/assert_20121221172433_1.dmp'', upload yes: ''CrashID=bp-e06041e7-6428-4e2b-a301-6c0792121221''
Fri Dec 21 22:25:49 2012 GMT: file ''/tmp/dumps/assert_20121221172544_1.dmp'', upload yes: ''CrashID=bp-a1546248-ddc7-4f94-9901-5a6e52121221''
Fri Dec 21 23:04:20 2012 GMT: file ''/tmp/dumps/crash_20121221180414_1.dmp'', upload yes: ''CrashID=bp-8bd419e8-12c4-4e77-a45f-3d18b2121221''
Fri Dec 21 23:12:04 2012 GMT: file ''/tmp/dumps/crash_20121221181159_1.dmp'', upload yes: ''CrashID=bp-ddfc1058-75a3-4b48-b0d3-fa02d2121221''

ZZeXx86 2012-12-22 github

I can fix this bug with changing game settings such as antialiasing which restarts video renderer.
So there must be something wrong with initialization I guess. Maybe it can be caused by higher latencies or loading times, don't know. When this bug is affecting me, my ping is around 170ms and FPS around 10, normally my ping is around 5-30ms depending on server.

Eeisbehr 2012-12-24 github

I used to get this when alt-tabbing the (fullscreen) game and on some map-loads. Disabling the KDE desktop effects (OpenGL) before starting TF2 seems to make it less likely. With desktop effects enabled it persisted after game restart, X restart and full reboot (Or I was very unlucky), without desktop effects a game restart will fix it.

Cchmod222 2013-01-03 github

It appears that if I disable my secondary monitor, this doesn't occur. At least it hasn't happened for quite some time now, but instantly when I forgot to disable my screen.

Eeisbehr 2013-01-03 github

I don't have a second monitor. Seems rather random to me when it happens. Not even sure anymore desktop effects have anything to do with it.

ZZeXx86 2013-01-03 github

I dont have desktop effects enabled nor any compositor and I am affected. I don't have any second monitor too.
What's more it happens on totally different distributions with very different environments.

Eeisbehr 2013-01-09 github

After todays/yesterdays TF2 update the bug is still there.

EElectricTriangle 2013-01-12 github

I'm getting ridiculously high CPU use and low frames as well, seemed to start about two client updates ago. For what it's worth I can't seem to get TF2 to split into multiple threads either, I just have the hl2_linux taking up 300-400% CPU time. The high CPU use only seems to happen for me after connecting to a server too.

Processor Information:
Vendor: AuthenticAMD
Speed: 3300 Mhz
6 logical processors
6 physical processors
HyperThreading: Unsupported
FCMOV: Supported
SSE2: Supported
SSE3: Supported
SSSE3: Supported
SSE4a: Supported
SSE41: Unsupported
SSE42: Unsupported

Network Information:
Network Speed:

Operating System Version:
Debian GNU/Linux 7.0 (wheezy) (64 bit)
Kernel Name: Linux
Kernel Version: 3.2.0-4-amd64
X Server vendor: The X.Org Foundation
X Server release: 11204000

Video Card:
Driver: ATI Technologies Inc. ATI Radeon HD 4600 Series

Driver Version:  3.3.11653 Compatibility Profile Context
Desktop Color Depth: 24 bits per pixel
Monitor Refresh Rate: 60 Hz
VendorID:  0x1002
DeviceID:  0x9490
Number of Monitors:  1
Number of Logical Video Cards:  1
Primary Display Resolution:  1440 x 900
Desktop Resolution: 1440 x 900
Primary Display Size: 16.14" x 10.12"  (19.02" diag)
                                        41.0cm x 25.7cm  (48.3cm diag)
Primary VRAM Not Detected

Sound card:
Audio device: ATI R6xx HDMI

Memory:
RAM: 16080 Mb

Miscellaneous:
UI Language: English
LANG: en_US.UTF-8
Microphone: Not set
Total Hard Disk Space Available: 108070 Mb
Largest Free Hard Disk Block: 8876 Mb

Installed software:

Recent Failure Reports:

JJonno888 2013-01-21 github

Hi. I think I've also had similar symptoms to this problem. Basically the hl2_linux process was chewing up all the processor time and the game was unplayable. I changed a bunch of BIOS settings and this seems to have had a positive impact. I need to spend a bit more time narrowing down the exact change that has made the difference but below is a basic summary of what I've disabled.

  • Disabled any automatic overclocking
  • Disabled Quiet Mode (I think this limits fan speed ?)
  • Disabled C1E support
  • Disabled all virtualization related feaures. (I'm not running VM's so I don't care)

The game is a lot more playable now. It still stutters here and there but definitely made a big difference. I'm only using a NVIDIA GT 640 but I know this card works fine with TF2 under windows. Below is a summary of my system (Sorry I'm not at the machine right now so I can't be more precise.)

  • AMD X2 Athlon Dual Core (Not a great CPU, more of a HTPC cpu type of thing.)
  • Nvidia GT640
  • 12 GB RAM
  • Openbox desktop environment
  • Ubuntu server 12.04
  • Latest nvidia-experimental package (3.10.14 I think?)
Hh0m3 2013-01-30 github

I can confirm this bug,

I'n my case i get 60~ FPS on Highest settings but out of nowhere or when i start the map the fps drop to below 10 then it's unplayable, i can even navigate trought the menu because the mouse is slow.

I'm repared that i have less lag when the AA is already at 2x or Off (like the FPS from 60 drop only to 12/15 FPS). And if change AA or HDR settings and head back to normal. But sometimes this just crash the game.

Just a detail, wine don't have this bug. I can play TF2 pretty well on wine. I will just stay there until the next TF2 upgrade or a solution :)

Processor Information:
Vendor: AuthenticAMD
Speed: 3200 Mhz
6 logical processors
6 physical processors
HyperThreading: Unsupported
FCMOV: Supported
SSE2: Supported
SSE3: Supported
SSSE3: Supported
SSE4a: Supported
SSE41: Unsupported
SSE42: Unsupported

Network Information:
Network Speed:

Operating System Version:
Debian GNU/Linux 7.0 (wheezy) (32 bit)
Kernel Name: Linux
Kernel Version: 3.6.8-home
X Server vendor: The X.Org Foundation
X Server release: 11204000

Video Card:
Driver: NVIDIA Corporation GeForce GTX 460/PCIe/SSE2/3DNOW!

Driver Version:  4.3.0 NVIDIA 310.19
Desktop Color Depth: 24 bits per pixel
Monitor Refresh Rate: 60 Hz
VendorID:  0x10de
DeviceID:  0xe22
Number of Monitors:  1
Number of Logical Video Cards:  1
Primary Display Resolution:  1920 x 1080
Desktop Resolution: 1920 x 1080
Primary Display Size: 20,04" x 11,26"  (22,95" diag)
                                        50,9cm x 28,6cm  (58,3cm diag)
Primary Bus: PCI Express 16x
Primary VRAM: 767 MB
Supported MSAA Modes:  2x 4x 8x 16x 

Sound card:
Audio device: Realtek ALC892

Memory:
RAM: 8113 Mb

Miscellaneous:
UI Language: English
LANG: pt_BR.UTF-8
Microphone: Not set
Total Hard Disk Space Available: 469452 Mb
Largest Free Hard Disk Block: 11900 Mb

Installed software:

Recent Failure Reports:
Wed Jan 30 01:02:28 2013 GMT: file ''/tmp/dumps/assert_20130129230216_1.dmp'', upload yes: ''CrashID=bp-13143f2e-4580-4472-ac61-795862130129''
Wed Jan 30 01:06:57 2013 GMT: file ''/tmp/dumps/assert_20130129230644_2.dmp'', upload yes: ''CrashID=bp-a2009f49-33c1-49dd-a18b-d90032130129''
Wed Jan 30 01:25:57 2013 GMT: file ''/tmp/dumps/assert_20130129232540_3.dmp'', upload yes: ''CrashID=bp-7da6e66a-3cbb-4b67-b7c4-2a18f2130129''
Wed Jan 30 01:28:41 2013 GMT: file ''/tmp/dumps/assert_20130129232828_1.dmp'', upload yes: ''CrashID=bp-80fca01c-7e75-42f4-bdd1-619bf2130129''
Wed Jan 30 04:29:22 2013 GMT: file ''/tmp/dumps/crash_20130130022917_1.dmp'', upload yes: ''CrashID=bp-6f8ef8b3-0a07-4dff-8eee-bb3ed2130129''
Wed Jan 30 04:31:41 2013 GMT: file ''/tmp/dumps/assert_20130130023133_1.dmp'', upload yes: ''CrashID=bp-1915b742-1699-403f-9f1c-648e12130129''
Wed Jan 30 06:54:31 2013 GMT: file ''/tmp/dumps/assert_20130130045429_1.dmp'', upload yes: ''CrashID=bp-fa9d979e-348b-44b6-a615-1c8c72130129''
Wed Jan 30 07:41:54 2013 GMT: file ''/tmp/dumps/crash_20130130054152_1.dmp'', upload yes: ''CrashID=bp-dfa77ebc-63a6-4840-bbee-6bf2d2130129''
Wed Jan 30 07:44:14 2013 GMT: file ''/tmp/dumps/crash_20130130054412_1.dmp'', upload yes: ''CrashID=bp-db08baa4-7edd-4ff3-9b25-a2deb2130129''
Wed Jan 30 07:46:05 2013 GMT: file ''/tmp/dumps/crash_20130130054602_1.dmp'', upload yes: ''CrashID=bp-d71564bd-c473-48a8-aa76-4df3e2130129''
Wed Jan 30 07:48:09 2013 GMT: file ''/tmp/dumps/crash_20130130054807_1.dmp'', upload yes: ''CrashID=bp-ab966ec4-9de2-4099-827a-23ca62130129''
Wed Jan 30 07:52:50 2013 GMT: file ''/tmp/dumps/crash_20130130055248_1.dmp'', upload yes: ''CrashID=bp-2c31b2b8-c606-4862-867e-a76332130129''
Wed Jan 30 08:57:01 2013 GMT: file ''/tmp/dumps/crash_20130130065657_1.dmp'', upload yes: ''CrashID=bp-aa68a19a-6c22-425f-9f25-44cf22130130''
Wed Jan 30 09:02:50 2013 GMT: file ''/tmp/dumps/crash_20130130070247_1.dmp'', upload yes: ''CrashID=bp-750cc512-b49e-43cd-ab25-3cf392130130''

Eeisbehr 2013-01-30 github

I'll post my nvidia-smi output from issue #696 here as well.

While TF2 bugs out and drops to ~10FPS:

==============NVSMI LOG==============

Timestamp : Tue Jan 29 14:22:53 2013
Driver Version : 313.18

Attached GPUs : 1
GPU 0000:01:00.0
Memory Usage
Total : 1023 MB
Used : 289 MB
Free : 734 MB

While TF2 runs normally it looks like this:

==============NVSMI LOG==============

Timestamp : Tue Jan 29 14:18:51 2013
Driver Version : 313.18

Attached GPUs : 1
GPU 0000:01:00.0
Memory Usage
Total : 1023 MB
Used : 934 MB
Free : 89 MB

So, something is happening to the video memory. Without cache on the card, performance just degrades heavily.

Hh0m3 2013-02-03 github

Update: Also tested with new 310.32 nvidia Proprietary Driver and still with the same problem.

LLongJohnPiano 2013-02-06 github
Processor Information:
    Vendor:  GenuineIntel
    Speed: 2501 Mhz
    2 logical processors
    2 physical processors
    HyperThreading:  Unsupported
    FCMOV:  Supported
    SSE2:  Supported
    SSE3:  Supported
    SSSE3:  Supported
    SSE4a:  Unsupported
    SSE41:  Supported
    SSE42:  Unsupported
    
Network Information:
    Network Speed:  
    
Operating System Version:
    Linux (64 bit)
    Kernel Name:  Linux
    Kernel Version:  3.7.2-201.fc18.x86_64
    X Server vendor:  Fedora Project
    X Server release:  11301000
    
Video Card:
    Driver:  NVIDIA Corporation GeForce 8600M GT/PCIe/SSE2
    Driver Version:  3.3.0 NVIDIA 304.64
    Desktop Color Depth: 24 bits per pixel
    Monitor Refresh Rate: 60 Hz
    VendorID:  0x10de
    DeviceID:  0x407
    Number of Monitors:  1
    Number of Logical Video Cards:  1
    Primary Display Resolution:  1440 x 900
    Desktop Resolution: 1440 x 900
    Primary Display Size: 14.45" x 9.06"  (17.05" diag)
                                            36.7cm x 23.0cm  (43.3cm diag)
    Primary Bus: PCI Express 16x
    Primary VRAM: 256 MB
    Supported MSAA Modes:  2x 4x 8x 16x 
    
Sound card:
    Audio device: SigmaTel STAC9205
    
Memory:
    RAM:  3953 Mb
    
Miscellaneous:
    UI Language:  English
    LANG:  ja_JP.utf8
    Microphone:  Not set
    Total Hard Disk Space Available:  14083 Mb
    Largest Free Hard Disk Block:  4825 Mb
    
Installed software:
    
Recent Failure Reports:
    

I got the same problem with Steam under Fedora 18. IMHO, the game suddenly become limited by the CPU. 'GPU lags' tends to lower when you look at a simple sky or on the ground, but when that problem appears looking at the sky do not affect the framerate at ALL !
It seems that desactivating the multicore rendering makes the problem disappear, to what I've experienced. Or at least, if the problem occurs, it wasn't sensible for me.

Also, just before this problem triggers, there's a huge visible stutter. It's like "Working okay----30ms stutter----FPS down"

Hh0m3 2013-02-07 github

I've tryed to shutdown Multicore Rendering and don't work for me. For me even looking at sky, i get 9 FPS :/ One thing, i've installed CS Source from Steam today and play a few rounds. No bugs at all, it's seen that is a TF2 bug, not a Source bug. I'm downloading Team Fortress 2 Beta to see if has some difference.

EElectricTriangle 2013-02-11 github

I've noticed one thing that causes some very similar symptoms to the ones noted here, and in fact is the sole cause of my prior issues with TF2. Servers sponsored by Pinion ads. Perhaps it's due to flash not loading properly or something, but only when I have HTML MOTDs enabled and I go on a Pinion server do I have this issue.

Also perhaps possible 32-bit OS users may not experience my issue because it would load properly for them, Steam at present only being a 32-bit app and all.

Basically what solved it for me is blocking the ads for those servers. As much as I like sponsoring them it's rather pointless to if I can't play whilst doing it.

Eeisbehr 2013-02-11 github

I'm not playing on one of those servers and have html motd disabled, still get this issue. Don't think it's related.

Eeisbehr 2013-02-15 github

I'll just assume it's not magic.

Now also opened a report here
https://devtalk.nvidia.com/default/topic/529521/lasting-reproducible-frame-rate-drop-to-7fps-on-gtx-560-ti-313-18-driver
As I was able to reproduce this with a different OpenGL Application.

ZZeXx86 2013-02-15 github

@eisbehr Thanks! I am affected too. Its very strange bug. Interesting CS:Source is not affected by this issue.

Jjorgenpt 2013-02-15 github

Issue transferred to ValveSoftware/Source-1-Games#23.

This issue is closed - continue conversation there.

Nothing extracted yet.