raptor2012rom

Distinguished
May 2, 2013
151
3
18,695
Well I got a bit of an old PC with problems. (previous thread). Managed to fix most of the problems. Especially crashing. After a while while checking in the settings I saw that my GPU (asus r7 260x) was running at PCIE 2.0 x8 , so I reseated it. Worked on PCIE x16 but crashed after a while and after restarting the pc the card switched to x8 and somehow was stable but the memory reading/writing speeds went from 6GB/sec to 3GB/sec. So I reseated/cleaned the GPU from dust and the connector and also switched to another 6-pin power connector for the GPU. Now works on PCIE 2.0 x16 like it should but I get random crashes , like was playing enlisted and crashed at 80c and also 1 minute in OCCT at CPU stress test i got a BSOD. Can't manage to fix the crashes.
Here is a link to the last BSOD dump file: dump file
Thx

PC specs:
Mobo: Asrock 960gm-vgs3 fx
CPU: AMD FX 6300 3.5ghz (stock not OC)
GPU: ASUS R7 260x OC 2GB GDDR5 128-bit
RAM: 2x Corsair Vengeance LPX DDR3 1600mhz CL9 run at (9-9-9-24-41)
HDD: Seagate 1TB 7200rpm
OS: Windows 10 Pro 64-bit
 
Solution
i got that advice for occt elsewhere. setting ram to defaults is to check if PC is stable there.
try it and see if you still get errors
it could be you need to give CPU or ram more voltage, one increment at a time.

Colif

Win 11 Master
Moderator
lets see if I can get it right this time

conversion of dumps

report

File: 111722-37234-01.dmp (Nov 18 2022 - 06:38:46)
BugCheck: [SYSTEM_SERVICE_EXCEPTION (3B)]
Probably caused by: memory_corruption (Process: csrss.exe)
Uptime: 0 Day(s), 0 Hour(s), 26 Min(s), and 02 Sec(s)

have you tried running DDU since fixing the cables?

crash mentions gdi which is graphics driver interface, so it might be the card to blame, not the drivers.
 

raptor2012rom

Distinguished
May 2, 2013
151
3
18,695
lets see if I can get it right this time

conversion of dumps

report

File: 111722-37234-01.dmp (Nov 18 2022 - 06:38:46)
BugCheck: [SYSTEM_SERVICE_EXCEPTION (3B)]
Probably caused by: memory_corruption (Process: csrss.exe)
Uptime: 0 Day(s), 0 Hour(s), 26 Min(s), and 02 Sec(s)

have you tried running DDU since fixing the cables?

crash mentions gdi which is graphics driver interface, so it might be the card to blame, not the drivers.
Man. You helped me alot and thanks for that. And no , I didn't reinstall drivers with DDU since reseating and changing cables but I will. The card might be broken but I bet the motherboard isn't right either , auto settings on the mobo overvolts the CPU and takes it to instability land. While searching for memory corruption on the web found a guy that said his crashes began as soon as his PC used the virtual memory (link) and somehow was related to HDD. I could try flash another vBIOS on the card and maybe fix it? But it's already on the last one. Need to rule the broken part out because I can get cheap replacement parts for this but at the same time the PC is old and won't try throwing money at it for no reason.
 

raptor2012rom

Distinguished
May 2, 2013
151
3
18,695
try running occt without xmp on, default bios settings, see if you still get core errors.
Will try. Also from what I've read might be something when the GPU is accessing the virtual memory in RAM. Tested the VRAM with a bunch of stress tools and passed every one of them.
My total paging file size for all drives is 2816 MB. Is that ok ?
 

raptor2012rom

Distinguished
May 2, 2013
151
3
18,695
page file sounds about normal

virtual memory is page file + ram
Oh ok. Needed to get that out of the way. I'll tweak with RAM timings a bit later and see if that fixes core errors and crashes.
But could XMP crash GPU driver/memory ? Also OCCT's 3d test for the GPU gets my video card to like 90c in 5mins. I don't seem to have a temp issue because while in games hardly passes 80c maybe a 82 if the house gets hotter.
 

Colif

Win 11 Master
Moderator
i got that advice for occt elsewhere. setting ram to defaults is to check if PC is stable there.
try it and see if you still get errors
it could be you need to give CPU or ram more voltage, one increment at a time.
 
Solution

raptor2012rom

Distinguished
May 2, 2013
151
3
18,695
i got that advice for occt elsewhere. setting ram to defaults is to check if PC is stable there.
try it and see if you still get errors
it could be you need to give CPU or ram more voltage, one increment at a time.
Just came home and i'll start testing. Well I have my cpu set to 1.20v right now at 3.5ghz , 1.25v also crashed. At auto voltages was overheating alot and thought till now it was a stable voltage (maybe it is maybe not) , I'll start with different RAM settings then with CPU and come with an update. I'll note my current BIOS settings too because I switched off pretty much almost all power saving features.
 

raptor2012rom

Distinguished
May 2, 2013
151
3
18,695
have storms here this weekend so be in when I can.
No problem bro. I'll keep testing.
Till now I got to test the following settings:

CPU at 1.20v / RAM at 1333mhz on auto timings @ 1.5v = CRASH
CPU at 1.30v / RAM at 1600mhz on auto timings @1.5v = CRASH + CPU overheating and throttling to 1.4ghz ingame
CPU at 1.25v / RAM at 1600mhz on 9-9-9-24-2-41CLK (memory's default on manual) @1.5v = CRASH
CPU at 1.20v / RAM at 1600mhz on 9-9-9-24-2-41CLK (memory's default on manual) @1.6v = DIDN'T CRASH YET! but got core5 erors in OCCT 1 minute in.

I'll test more. But I was thinking if maybe the 1.6v on RAM is stable and it wont crash anymore , does the core erors matter so much ? I mean when the GPU was running at PCIE x8 mode I got core4 and core5 errors in OCCT but the PC was very usable , at least in the few games I play and while doing some photoshop work.
 

raptor2012rom

Distinguished
May 2, 2013
151
3
18,695
Didn't crash anymore. The core errors are still there and might be a voltage thing. So overvolting the ram to 1.6v (stock 1.5v) made them stable.
I think from some reason when the GPU was accesing the virtual memory in RAM , the PC crashed. Was also a bit cooler in the room where I have the PC lately but as monitoring goes the PC did crash on normal temps too CPU&GPU wise.