• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

5850 system just plain old freezing now!

kidscruff

kidscruff

kidscruff

Associate
Joined
23 Feb 2003
Posts
372
I had been having problems with 5850 'VPU recover' kicking in at idle speeds, in Win7-64bit. Same idle issues occurred in XP-32bit. Thought I had it fixed after updating the motherboard BIOS, and all the system and graphics drivers (Cat 9.10).

However, because I only had a Corsair HX450W PSU, I thought I would replace with a Corsair HX650W PSU. So fitted the new PSU and all the cables. Only difference being the 650W has 2 PCI-E cables per PSU connector, 450W uses one PCI-E cable per PSU connector.

Win 7 booted fine, then system completely froze. KB + Mouse locked, no response. Rebooted, and system froze again after a couple minutes. Checked all cables - connected fine. Reset BIOS, system ran fine... Played Crysis and Bioshock demos on max settings - ran great. Unigine Heaven benchmark, which usually runs fine, crashed to desktop mid way. Re-ran, then got a brown/grey screen, and system freeze.

Rebooted and ran Heaven again - ran fine.

Before at least it seemed like a driver crash. The system freezing now seems to be happening in place of the driver crash (or VPU recover), but obviously more serious in that I have to hard reboot!

Anything to try?
 
Wish I had another one to try it in, but I don't.

Memtest86+ v4.00 ran for 5 hours (8 full passes) with zero errors, so I'm ruling out memory.
 
Have you got a copy of XP or Vista you can try? There's been a lot of people having problems with these cards in Win7.
 
I really don't know many people with a desktop/built PC.

I have no compatible spare parts. If I want to try a different graphics card I'll probably have to buy a cheap one. But that's basically throwing money away.

It's just a pain that previous software crashes have been replaced with system hangs. Such a nightmare to troubleshoot.
 
Overclocked cpu at all?

Almost all vpu recover issues I had were resolved by upping the voltage on my cpu because while the system on old card(a 3870x2 back when this happened) was fine at overclock X(can't remember what speed/voltage, was ages ago) when I stuck a 4870x2 in I got lots and lots of crashes from the vpu recover, tried drivers, reseating card, new windows install, new everything eventually it turned out my completely stable overclock before was just right on the edge and the extra power draw, probably through the mobo pci-e slot, was just making the difference. Either upping cpu voltage one notch, or dropping the overclock made it work fine.

vpu recovery, in my experience is usually software crashing from an unstable system rather than gpu instability.

But yeah, troubleshooting any crashing is a pain because no matter how obviously a specific type of crash normally points to one problem, when its your own system, its always something abnormal and not the obvious thing :p

EDIT:_ Forgot to say, quite a few people are having issues with powerplay on the 5850, including myself, for some reason it just won't stay stable using powerplay at all, to prevent it on boot I use gpu clock tool to straight away set it to 800/1150(easily stable on mine with stock voltage) and gpu clock tool overrides powerplay so you don't get speed's dropping. Fixes all my problems and the majority of problems for other people with issues, so thats definately somewhere to start, grab gpu tool, set the default 3d speed and keep it there, play around with things and see if you still get crashing, at least if you do you know its not the powerplay issue.
 
Last edited:
No, everything at stock speeds. Basically, optimised BIOS defaults and saved.

Will try uninstalling the video drivers and see if it still hangs. That should sort out whether it's plain hardware, or due to drivers.

Next I'll try the GPU clock tool, which sounds like a feasible solution. Which version to use though?
 
I would set your memory manually to 2T-8-8-8-24-trfc88 the rest you can leave, make sure to set up both sicks as you have to manually do both on some boards, and set the voltage on the ram to 1.65v
 
hyperst said:
I would set your memory manually to 2T-8-8-8-24-trfc88 the rest you can leave, make sure to set up both sicks as you have to manually do both on some boards, and set the voltage on the ram to 1.65v
Click to expand...

Have previously set the memory (both sticks) to 1333MHz, 7-7-7-20 1T, 1.64V, rest on Auto. Running currently at 1066MHz, on auto (gives 1.5V, 1T, 7-7-7-19). Prime95 runs fine for hours at the slower memory speed, so assumed it was stable.

OCZ have suggested setting QPI/VTT voltage to 1.35V, which seems pretty high to me!
 
give my settings a go, have read that running 1t can cause stability issues and seeing as you are getting issues i would change memory setting,as regarding the QPI/VTT voltage mine is oc and i set my vtt near the voltage of cpu, around 1.31v

just to add, i have had some freezing in my setup(exactly as you describe) but put it down to my oc not quite stable, im still experimenting at the moment with it, just dropped my base clock down 5 to see if i was pushing it a bit
 
Last edited:
Nice, before changing the memory settings, I ran the windows experience index test which normally crashes the card. But this time crashed to a light-blue screen. Frozen.

Rebooted, and as Windows was about to load, I got a garbled display screen: mostly black, with a green/pink static pattern stripe along the bottom of the screen. Will post a picture later.

Off and on, it booted fine. Something isn't right with that graphics card!
 
Seems I didn't read the OP properly! :p

Does sound like the card is dodgy. Maybe it's worth trying to re-flash the Sapphire bios before RMAing the card. Someone else on here has fixed some issues with a bios flash. You can get a copy of the Sapphire one from the techpowerup bios collection.
 
i've have had this issue before. but since i flashed my card with asus bios it seems more stable.

the ram can make issues if it not set right. so far im at 1333mhz 6-6-6-20 1t 1.7v. but as said 1t can make stable issue.
 
Surely the ram stability should not be affected by the graphics card drivers being installed (or not), especially considering the crashes happen at idle and not when transferring or processing large amounts of data.

If I uninstall the Catalyst drivers, the computer will never crash, and Prime95 runs with no issues. Still I suppose there could be a problem with the GPU accessing system memory, so will try the lower timings anyways.

Cheers :)
 
Flippin Nora,

Ran memory at "OCZ spec" (i.e. their forum support):

CIA2 [Disabled]
Performance Enhance [Standard]
DRAM Timings CL 7-7-7-20
tRFC [88]
Command Rate [2T]
QPI/VTT Voltage [1.35]
DRAM Voltage [1.65]
Click to expand...

Computer went nuts with about 5 crashes in an hour, which messed up the registry and Windows couldn't repair itself... these were full-on GPU crashes with frozen, plain colour screens. So I've just had to reinstall Windows 7 :(

Now, Before I had had a chance to install the graphics drivers (or any drivers for that matter!), the screen FROZE!! :mad: So I take back my last statement - something is critical with the PC (my guess is memory or very flaky GPU at this point).

Memory timings had been set to Auto (Optimised defaults) in the BIOS, which gave 1066MHz 7-7-7-19-1T, 1.5V. So I've changed these back to "OCZ spec" 2T-7-7-7-20-tRFC88, with the exception of QPI Voltage at 1.25V.

Will see now how this runs before installing any drivers.

I NEVER had a single system freeze before installing the new 650W PSU (just a load of VPU recovers), so am also gonna try reseating the memory. Taking out the ATX power connector did cause a lot of bending in the motherboard.
 
you didnt forget to plug in the smaller 8 pin motherboard connection did you, just thought id ask just incase
also did you give the memory settings a go i put up there
 
kidscruff said:
Flippin Nora,

Ran memory at "OCZ spec" (i.e. their forum support):



Computer went nuts with about 5 crashes in an hour, which messed up the registry and Windows couldn't repair itself... these were full-on GPU crashes with frozen, plain colour screens. So I've just had to reinstall Windows 7 :(

Now, Before I had had a chance to install the graphics drivers (or any drivers for that matter!), the screen FROZE!! :mad: So I take back my last statement - something is critical with the PC (my guess is memory or very flaky GPU at this point).

Memory timings had been set to Auto (Optimised defaults) in the BIOS, which gave 1066MHz 7-7-7-19-1T, 1.5V. So I've changed these back to "OCZ spec" 2T-7-7-7-20-tRFC88, with the exception of QPI Voltage at 1.25V.

Will see now how this runs before installing any drivers.

I NEVER had a single system freeze before installing the new 650W PSU (just a load of VPU recovers), so am also gonna try reseating the memory. Taking out the ATX power connector did cause a lot of bending in the motherboard.
Click to expand...
it's looking like your ram is failing because it having issues running them at spec.
 
You must log in or register to reply here.
Back
Top Bottom