10 replies to this topic

[G4LV4TR0N]

I do have this odd issue that my game can crash without any notice. It just crashes to desktop and is very annoying because it can happen few times in a row effectively blocking me from finishing game. Here is what I know:

- Only certain maps are affected, with Terra Therma crashing most of time. Crimson Strait, Forest Colony, Emerald Taiga, Grim Plexus, Viridian Bog and others also crash but not that often.

- Some maps like HPG Manifold or most Siege(Invasion) maps are not affected and never crash.

- Crash can happen in every game mode but Invasion is most prone while Scouting isn't affected that much.

- There is no indication that it will crash, sometimes it happens when there's a big battle with me in middle of it but sometimes it happens when nothing is happening.

- I suspect video card driver, but it doesn't mean it can't be fixed manually.

- It's unrelated to user.cfg.

- It's unrelated to GPU overclocking.

- Logs don't have any specific information.

Could it be that there are more textures being loaded in Invasion mode and on those maps like Terra Therma and those do cause problems?

[NARC BAIT]

well, your using your old GPU, with a new motherboard ... how heavy duty is the power supply ... maybe on the last system, it was within tolerance, and now, its occasionally drifting beyond it ....

some of them ATI drivers were a bit less than desirable ... so its hard to rule that out ...

[Tarl Cabot]

Memory - The crashes may happen when something comes into view to be loaded.

These threads you have posted , though noted, may still be relevant, especially if anyone else tries to assist.

https://mwomercs.com...rrain-textures/
https://mwomercs.com...ost__p__5816804

G4LV4TR0N, on 06 July 2017 - 12:19 PM, said:

Awesome work and one of best posts on forums. I am also switching my CPU from E8400@4GHz to Pentium G4600, but my video card will be same old Radeon HD5670. I guess I will have to edit my config now and modify most of cvars by 180 degree because I was trying to move as much as possible from CPU to GPU but with my current config it will be better to move from GPU to CPU.

Edited by Tarl Cabot, 25 July 2017 - 03:51 PM.

[G4LV4TR0N]

NARC BAIT, on 25 July 2017 - 05:58 AM, said:

well, your using your old GPU, with a new motherboard ... how heavy duty is the power supply ... maybe on the last system, it was within tolerance, and now, its occasionally drifting beyond it ....

some of them ATI drivers were a bit less than desirable ... so its hard to rule that out ...

PSU will easily handle my video card, it's just 65TDP and my new CPU/MOBO draw less power than previous ones. I had same crashes before so definitely not CPU/MOBO related. I have underclocked my video card to ~75% of normal clocks and crashes were still there hence either something is faulty(doubt it as I can run all demanding tests for days without problem) or there's a driver problem, I have tried Catalyst and Crimson and it's all same. As drivers are outdated because card is not supported it could be also possibility that MWO game is just using some newer drawing techniques/effects that ain't supported by outdated driver properly - I have noticed that some maps like HPG Manifold or Tourmaline Desert never crashed for me and it could be that they were made for older CryEngine revisions or maybe using different default config sets.

I have tried config that hard locked my video RAM usage to something like 500-600MB but there is no difference when comparing to max 1000MB and during game with my 30 FPS limit GPU hoovers around 60% and spikes to 80% ceiling in very rare situations. Definitely not temperatures neither, my card is now running cooler than stock reference models. Because of this I am ruling out faulty hardware.

Might wonder if I would get different video card, would it help. But as I don't want to spend and I want low TDP my options are limited. GTX 750 Ti, GT 1030, GTX 1050, HD 7790, RX 550, RX 460 are probably all in my range, all with 2GB GDDR5. I do wonder does memory bandwidth make any difference for MWO because GT 1030 is 64-bit card while others are 128-bit.

[Tarl Cabot]

I do not see you saying anything about the system's memory, not the GPU memory.

[G4LV4TR0N]

System memory is fine.

[NARC BAIT]

have you tried stress testing the memory ? is it running an XMP profile ?

scouting mode should cause a lot less stress than invasion because of the total player count, and the way that impacts the game through the battlegrid ...

[G4LV4TR0N]

If you are talking about system memory, then I had same error on two different machines, one with DDR2 and second with DDR4. I could also try DDR3 one, except I know it's not RAM.

Edited by G4LV4TR0N, 27 July 2017 - 09:04 AM.

[NARC BAIT]

yeah ... I'm just wondering because its where I'm spending my time at the moment, with ... well weird results ... I've read that on an intel platform, when you activate an XMP profile, it ups the base clock to 125 MHz, and the timings expect to take that into account .... but its not really translating onto the AMD ddr4 platform at the moment .... and I was just wondering how your CPU might have been interpreting things ....

have you ran the repair tool lately ? one corrupt texture might be crashing out the streaming system and bringing the whole thing down ... you might only see it occasionally if its related to a mech/pattern that's not loaded very often ... a losmech ...

usually I get lazy and use a dxdiag to see the last bunch of programs that made WER info before I decide that its worth going to look for something that doesn't usually contain a lot ....

[G4LV4TR0N]

Tried Repair Tool many times, been using it for everything even game patching before but now Portal is mandatory. As for DDR4, my motherboard supports DDR4 2400MHz but I use DDR4 2666MHz stick. So if anything, it's underclocked. I have raised timings a bit but not enough to make RAM running at 2400MHz faster than default 2666Mhz. Also ran Windows Memory Test and MemTest86+ for few hours. I did one debug session where game crashed with "log_WriteToFileVerbosity 4", but output ain't clear. That's what I get from "dxdiag":

Quote

---------------
Diagnostics
---------------

Windows Error Reporting:
+++ WER0 +++:
Fault bucket 120720655473, type 4
Event Name: APPCRASH
Response: Not available
Cab Id: 0

Problem signature:
P1: DllHost.exe
P2: 10.0.15063.0
P3: 4f42de92
P4: StackHash_2264
P5: 0.0.0.0
P6: 00000000
P7: c0000005
P8: PCH_C8
P9:
P10:


+++ WER1 +++:
Fault bucket 120720655473, type 4
Event Name: APPCRASH
Response: Not available
Cab Id: 0

Problem signature:
P1: DllHost.exe
P2: 10.0.15063.0
P3: 4f42de92
P4: StackHash_2264
P5: 0.0.0.0
P6: 00000000
P7: c0000005
P8: PCH_3A
P9:
P10:


+++ WER2 +++:
Fault bucket 120720655473, type 4
Event Name: APPCRASH
Response: Not available
Cab Id: 0

Problem signature:
P1: DllHost.exe
P2: 10.0.15063.0
P3: 4f42de92
P4: StackHash_2264
P5: 0.0.0.0
P6: 00000000
P7: c0000005
P8: PCH_2A
P9:
P10:


+++ WER3 +++:
Fault bucket 120720655473, type 4
Event Name: APPCRASH
Response: Not available
Cab Id: 0

Problem signature:
P1: DllHost.exe
P2: 10.0.15063.0
P3: 4f42de92
P4: StackHash_2264
P5: 0.0.0.0
P6: 00000000
P7: c0000005
P8: PCH_B2
P9:
P10:


+++ WER4 +++:
Fault bucket 120720655473, type 4
Event Name: APPCRASH
Response: Not available
Cab Id: 0

Problem signature:
P1: DllHost.exe
P2: 10.0.15063.0
P3: 4f42de92
P4: StackHash_2264
P5: 0.0.0.0
P6: 00000000
P7: c0000005
P8: PCH_A6
P9:
P10:


+++ WER5 +++:
Fault bucket 120720655473, type 4
Event Name: APPCRASH
Response: Not available
Cab Id: 0

Problem signature:
P1: DllHost.exe
P2: 10.0.15063.0
P3: 4f42de92
P4: StackHash_2264
P5: 0.0.0.0
P6: 00000000
P7: c0000005
P8: PCH_C7
P9:
P10:


+++ WER6 +++:
Fault bucket 120720655473, type 4
Event Name: APPCRASH
Response: Not available
Cab Id: 0

Problem signature:
P1: DllHost.exe
P2: 10.0.15063.0
P3: 4f42de92
P4: StackHash_2264
P5: 0.0.0.0
P6: 00000000
P7: c0000005
P8: PCH_0F
P9:
P10:


+++ WER7 +++:
Fault bucket 120720655473, type 4
Event Name: APPCRASH
Response: Not available
Cab Id: 0

Problem signature:
P1: DllHost.exe
P2: 10.0.15063.0
P3: 4f42de92
P4: StackHash_2264
P5: 0.0.0.0
P6: 00000000
P7: c0000005
P8: PCH_4C
P9:
P10:


+++ WER8 +++:
Fault bucket 120720655473, type 4
Event Name: APPCRASH
Response: Not available
Cab Id: 0

Problem signature:
P1: DllHost.exe
P2: 10.0.15063.0
P3: 4f42de92
P4: StackHash_2264
P5: 0.0.0.0
P6: 00000000
P7: c0000005
P8: PCH_23
P9:
P10:


+++ WER9 +++:
Fault bucket 120720655473, type 4
Event Name: APPCRASH
Response: Not available
Cab Id: 0

Problem signature:
P1: DllHost.exe
P2: 10.0.15063.0
P3: 4f42de92
P4: StackHash_2264
P5: 0.0.0.0
P6: 00000000
P7: c0000005
P8: PCH_DD
P9:
P10:

"c0000005" is indeed related to memory but it's too generic, it could also be any programmable error or driver issue. I am now testing game in DX9 mode to see if it helps. As for config I have been crashing with user.cfg and without user.cfg. Also, as I said two most common crashing maps are Invasion mode Terra Therma and to much lesser extent Crimson Strait. On those maps it's very common that after I crash in-game and try to rejoin I am getting crashed instantly, few times in a row. At some point I am able to rejoin and continue playing. My take is still texture/shader issue but probably more related to my video card driver than MWO game itself.

[NARC BAIT]

seemed to find some random stuff implying that 'StackHash_2264' errors stopped occurring only after disabling the NX instruction bit ... potentially just disabling DEP .... normally I wouldn't recommend disabling it ... as its meant to be 'protecting' memory from improper access ....

if you can 'REALLY' say for sure that your clean from virus etc, then you might want to try it as a test run, until it crashes ....