Stable Win7 64bit apart from when playing games.
- Thread starter RimBlock
- Start date
I've always found Furmark to be an excellent way of testing a GPU's stability. If it can't handle 20 minutes of that something is wrong
good point, i completely forgot about that!
it will push your gpu up to 100% and give it a right good stress testing.
It seems to me like the temps are the enemy here, have you got any spare fans you could put in your case, make the airflow better or anything?
its cold in the uk, normally dont worry too much about the heat
Ok,
Installed CPUID last night and got some readings for power.
12V = 12.16 avg, 11.84 Min
Not graphed so cannot see if the min occured around the time of the hangs.
Other info gathered from 'out there'.....
Tech Harp
When you test the card out, be sure to note any visual abnormalities like corrupted textures. The game may even hang. This is a sign that one or more of the memory chips are failing at that clock speed. When that happens, clock down the memory bus by one step.
3D Guru
[Page 4]
So we fire off a hefty shader application at the GPU and start monitoring temperature behavior as it would mid-gaming, we literally stress the GPU 100% here. We measured at a set 21 degrees C room-temperature.
In idle you can expect a temperature of 45 degrees C / 113 F. Pretty normal. Yet once we push the GPU to 100%, the temperatures take a pretty hefty toll and settle at almost 70 Degrees C / 158F. This is also fine, actually pretty good for a pre-overclocked product.
My card was topping out around 79 deg but averaged around 65-68 deg
The PSU I have I believe is this one (came with the case).
I took the machine apart completely last night to clean it out. There is only the exhaulst fan and the PSU fans for cooling the midi case. Both where pretty dusty with the exhaust fan choked to the point of sounding like a helicopter. I had to take out the motherboard to get the CPU fan out for cleaning and not everything is sparklie new. The PSU I believe is suffering from dust buildup as well but everytime I try and clean a PSU it tends to fail soon after
so I have left it alone apart from a vacumn.
Testing after all the cleaning resulted in no real differences in temps etc although recovery time from a HL2 EP2 hang after jumping to desktop is more or less straight away.
Going on the notes above I may start underclocking the mem chips first and see if that fixes it. I am loath to accept a card which does not meet production specs especially as this card is meant to be overclockable on top of the factory overclock.
I will post a shot with the CPUID info tonight when I get home.
Cheers
RB
Installed CPUID last night and got some readings for power.
12V = 12.16 avg, 11.84 Min
Not graphed so cannot see if the min occured around the time of the hangs.
Other info gathered from 'out there'.....
Tech Harp
When you test the card out, be sure to note any visual abnormalities like corrupted textures. The game may even hang. This is a sign that one or more of the memory chips are failing at that clock speed. When that happens, clock down the memory bus by one step.
3D Guru
[Page 4]
So we fire off a hefty shader application at the GPU and start monitoring temperature behavior as it would mid-gaming, we literally stress the GPU 100% here. We measured at a set 21 degrees C room-temperature.
In idle you can expect a temperature of 45 degrees C / 113 F. Pretty normal. Yet once we push the GPU to 100%, the temperatures take a pretty hefty toll and settle at almost 70 Degrees C / 158F. This is also fine, actually pretty good for a pre-overclocked product.
My card was topping out around 79 deg but averaged around 65-68 deg
The PSU I have I believe is this one (came with the case).
I took the machine apart completely last night to clean it out. There is only the exhaulst fan and the PSU fans for cooling the midi case. Both where pretty dusty with the exhaust fan choked to the point of sounding like a helicopter. I had to take out the motherboard to get the CPU fan out for cleaning and not everything is sparklie new. The PSU I believe is suffering from dust buildup as well but everytime I try and clean a PSU it tends to fail soon after
so I have left it alone apart from a vacumn.Testing after all the cleaning resulted in no real differences in temps etc
although recovery time from a HL2 EP2 hang after jumping to desktop is more or less straight away.Going on the notes above I may start underclocking the mem chips first and see if that fixes it. I am loath to accept a card which does not meet production specs especially as this card is meant to be overclockable on top of the factory overclock.
I will post a shot with the CPUID info tonight when I get home.
Cheers
RB
good point, i completely forgot about that!
it will push your gpu up to 100% and give it a right good stress testing.
It seems to me like the temps are the enemy here, have you got any spare fans you could put in your case, make the airflow better or anything?
its cold in the uk, normally dont worry too much about the heat
I have never had problems in the Uk with heat in the PC even when I had 8 15K SCSI hard drives running....... noise yes, heat no
.More than the heat it is the dust that causes problems I think, that and the humid weather. The back USB port of my DLink DNS-323 NAS box is rusting
. Old cable TV points (maybe 15+ years old) here are usually heavily corroded, green and unsurprisingly dont work too well.I used to smoke tobacco (rollups) and I believe even after stopping smoking a year ago, the dust buildup in the PC is more than when I was smoking in the UK.
Now I have seem recall reading about an idea years back of putting one leg of a pair of tights over the case to act as a filter (high denier I believe, bigger holes in the thread). I do not recall what the article suggested doing with the other leg of the tights
although you could always use stockings and then you would have an excuse if your wife / girlfriend found the pair you bought for them in the draw and had a disapproving look on their face. Just hope she does not ask 'If you bought it for the computer to act as an air filter then where does the suspender belt go' 
.TBH this is the first time here in Singapore I have had the problem. My ATI4780 and 8400GS (not surprisingly in this case) ran fine.
Oh the fun of changing video cards
.Finally down the gym again tonight so might not be ble to do the memtest I completely forgot about over the weekend but will try.
Cheers
RB
Last edited:
ah yes the stocking dust filter tip
you can also by those air can cleaner things, they are pretty good actually.
i think it might be best you try and sell the graphics card or rma it if you can, you shouldnt have to underclock your card - but then it already is running faster then a standard one.
good luck whatever you decide.
you can also by those air can cleaner things, they are pretty good actually.
i think it might be best you try and sell the graphics card or rma it if you can, you shouldnt have to underclock your card - but then it already is running faster then a standard one.
good luck whatever you decide.
ah yes the stocking dust filter tip
you can also by those air can cleaner things, they are pretty good actually.
i think it might be best you try and sell the graphics card or rma it if you can, you shouldnt have to underclock your card - but then it already is running faster then a standard one.
good luck whatever you decide.
Cheers Toon,
I checked the PSU for dust last night and it was clear. I also checked the 4GB memory using memtest 4 boot CD and it passed.
I underclocked the 260 using VTune from Palits site and underclocking the memory had no effect, however, when I underclocked the GPU core from 625Mhz to 580Mhz the problems vanished. The shaders Mhz were also reduced as I did a linked underclock (Core and Shaders frequency reduce together).
I managed to get the core back up to 609Mhz (shaders were around 1340 down by about 12Mhz) and it was stable with no graphics glitches etc.
I have spoken to the shop and they have no stock so cannot replace. Spoken to the warrenty supplier and they have no stock until next week. Investigation doing a topup and replacing for something with better spec that they do have in stock. I suspect I will not be able to do this but no harm asking.
Oh and I ran the Furmark at extreme burn on 1920x1600 (Dell 24" monitor) with 16X AA and had no issues with the card. The card stabilazed at 80deg +-.
Oh well, at least the underclocking highlighted the core cause.
Is it worth the hassle of waiting a week, having to travel to the warrenty centre, wait to have the card tested and then, hopefully on the same day, take the replacement home for the sake of 16Mhz GPU clock speed.
Possibly, the required drop in speed may be an indicator of something worse to come and from the review I posted in this thread, it seems these cards should be even more overclockable.
Thanks for all the advise everyone.
Much appreciated.
RB