Head scratching moment - help needed
Discussion
I've just completely rebuilt my home/gaming rig after a year of problems.
The problems turned out to be compatibility issues between the mobo and the video card - all of which are now history.
New processor (core duo 2.13GHz)
New mobo
New memory (1Gb DDR2 1066MHz)
The video card remains (a modest 7800GTX)
Everything works fine - it benchmarks highly and runs like a dream.... but...
As a test I ran up NFS-C and put it in the highest detail mode that my monitor will handle and gave it a play. Great for a few minutes but then the whole thing just stalled - a good 2 seconds where nothing happened. Initially the screen rendering went a bit funny (lots of blank black vertices) and the game recovered. I can live with that as it was probably a background task kicking in and chewing up the memory bandwidth (and I did have all the settings turned up to full to test the limit) but the whole thing freezing has me flumoxed.
I checked the services and task manager and there is simply nothing visible running in the background that could have this kind of impact.
The only thing I can think of is that the swap file was being thrashed but of course there was nothing being logged at the time so I can't verify that.
Anyone got some suggestions?
The problems turned out to be compatibility issues between the mobo and the video card - all of which are now history.
New processor (core duo 2.13GHz)
New mobo
New memory (1Gb DDR2 1066MHz)
The video card remains (a modest 7800GTX)
Everything works fine - it benchmarks highly and runs like a dream.... but...
As a test I ran up NFS-C and put it in the highest detail mode that my monitor will handle and gave it a play. Great for a few minutes but then the whole thing just stalled - a good 2 seconds where nothing happened. Initially the screen rendering went a bit funny (lots of blank black vertices) and the game recovered. I can live with that as it was probably a background task kicking in and chewing up the memory bandwidth (and I did have all the settings turned up to full to test the limit) but the whole thing freezing has me flumoxed.
I checked the services and task manager and there is simply nothing visible running in the background that could have this kind of impact.
The only thing I can think of is that the swap file was being thrashed but of course there was nothing being logged at the time so I can't verify that.
Anyone got some suggestions?
Card has no dust on worth mentioning - the fan works cleanly *but* it is the slimline cooler and fan arrangement rather than the double bay setup that appears to be more common
Plenty of space around it and a fan aimed pretty much directly at it too (actually something I worry about as it means the card could potentially pick up dust as a result)
I will check through the nvidia diagnostics to see if it has temperatures - i don't remember seeing it before though.
Plenty of space around it and a fan aimed pretty much directly at it too (actually something I worry about as it means the card could potentially pick up dust as a result)
I will check through the nvidia diagnostics to see if it has temperatures - i don't remember seeing it before though.
Black lines, stalling.
Overheat. Now hear me out before going "but everything's new Sarah! It can't be gunk!" Does any of the equipment installed (CPU, Gfx, case fans) have thermal control (i.e. the hotter the system gets the faster the fans go)? I know most modern GFx cards are like this and quite often it's the drivers that regulates this. With the CPU being a 100% load running the game the card overheated and stalled (thermal protection, it clocked itself down), the drivers picked up on this and accelerated the fans up to speed.
It is plausible. Personally if you're building a gaming rig I would give the system as many fans (or indeed go with water) as possible; use something like a fan bus controller to stop/slow them when not needed so you don't have a hurricane going on whilst typing.
Overheat. Now hear me out before going "but everything's new Sarah! It can't be gunk!" Does any of the equipment installed (CPU, Gfx, case fans) have thermal control (i.e. the hotter the system gets the faster the fans go)? I know most modern GFx cards are like this and quite often it's the drivers that regulates this. With the CPU being a 100% load running the game the card overheated and stalled (thermal protection, it clocked itself down), the drivers picked up on this and accelerated the fans up to speed.
It is plausible. Personally if you're building a gaming rig I would give the system as many fans (or indeed go with water) as possible; use something like a fan bus controller to stop/slow them when not needed so you don't have a hurricane going on whilst typing.
Lots of fans already - hd power supply with twin fans, twin case fans at the back, one at the front, one at the side.
SATA drives are in caddy bays with their own fan extractors (and noise suppressors)
Can't really afford a water cooling system right now but the way things are going to could be what is needed.
The CPU temperature is beautifully low - it sits there at 35degC under load which is the sort of temperature I like to see.
SATA drives are in caddy bays with their own fan extractors (and noise suppressors)
Can't really afford a water cooling system right now but the way things are going to could be what is needed.
The CPU temperature is beautifully low - it sits there at 35degC under load which is the sort of temperature I like to see.
jimbro1000 said:
The CPU temperature is beautifully low - it sits there at 35degC under load which is the sort of temperature I like to see.
If you go in to the advanced options of the Nvidia control panel bit and play with the options it should bring up the temperature of the gpu. Tell it to log everything and to annoy you when the device overhears. Then thrash it. I don't think we're looking at a CPU overheat issue, either GPU or video memory cooking.
Possibly making sure you've got the latest drivers for the card would be a good idea as well (again, driver controlled cooling system might be more responsive on newer drivers).
I lay the blame firmly at the door of Windows. I cant see a fault with anything else and the game recovered rather than going completely. If it were an overheating problem then surely the heat will have carried on rising and the card would have turned it metaphorical toes up. I would think its most likely a programming glitch on either the game itself or on windows. Best thing to do is to see if you can re-create it. If you cant then why worry but if you can and want to rule out some hardware try dropping the spec on the game and see if that helps.
Plotloss said:
Power?
Did you upgrade the powersupply at the same time?
Did you upgrade the powersupply at the same time?
580W unit - none of this cheapo "free with the case" power supply. It isn't the best power supply money could buy but it is way above the dross.
The only thing I have changed in this respect (while the machine is being built and tested) is that it is not being run through a UPS so the voltage supply from the mains may not be as stable as it could be.
The whole reason this situation came about was because the mobo previously in use did not support pci-e 16x, it only went to 4x and this was causing timing problems even though it "shouldn't". The manual states it is a 16x supported slot but when I did some digging with the manufacturer it turns out to be a lie.
The problem manifested itself as graphical corruptions, vertices with random co-ordinates or obviously corrupted 2D images or textures. When I tried the card in other machines all of these problems disappeared despite very rigorous testing.
The graphical problems now have no semblance at all to the previous corruptions - this really is something different altogether.
road2ruin said:
I lay the blame firmly at the door of Windows. I cant see a fault with anything else and the game recovered rather than going completely. If it were an overheating problem then surely the heat will have carried on rising and the card would have turned it metaphorical toes up. I would think its most likely a programming glitch on either the game itself or on windows. Best thing to do is to see if you can re-create it. If you cant then why worry but if you can and want to rule out some hardware try dropping the spec on the game and see if that helps.
I agree with the why worry - but I've been down that road before and it turned out to be a faulty memory chip. Replaced under warranty but not before a whole load of data had been corrupted (none of it critical - that was all backed up - but all very annoying).
If it was a heat/load issue the heat will vary according to the load. If a series of scenes require a higher load then the heat problem will increase while a less demanding sequence will allow the heat to dissipate.
The symptoms you describe I have only seen with a bad PSU, I swapped everything out before trying the PSU, finally did the PSU and it was cured.
Have you got another one you could test with?
Its not neccesarily underpowered but could be on its way out, this is exactly what I saw previously...
Have you got another one you could test with?
Its not neccesarily underpowered but could be on its way out, this is exactly what I saw previously...
Plotloss said:
The symptoms you describe I have only seen with a bad PSU, I swapped everything out before trying the PSU, finally did the PSU and it was cured.
Have you got another one you could test with?
Its not neccesarily underpowered but could be on its way out, this is exactly what I saw previously...
Have you got another one you could test with?
Its not neccesarily underpowered but could be on its way out, this is exactly what I saw previously...
I have an lower rated Antec unit I can trial it with but I don't think it has the right connector on it for the video card which means hunting down an adapter (oh joy)
Putting it back on the UPS would likely resolve the PSU problem unless it really is on its way out and I believe it is still under warranty so not the end of the world.
Just seems to be the latest in a long line of hassles. I had to rebuild 4 computers over the weekend and you get sick of it after a while. Most of it is just cause by dust but that is why I tend to put filters in for my computers. The kids' computers don't have that and I was dismayed at just how much dust they can accumulate in just a few months - it isn't even as if they are used that much!
The worst one was a few months back when a loaned out computer came back dead - everything in it was fried. Popped capacitors, melted jumpers, the works. It was only after a somewhat blunt conversation that it turned out that the building had been in close proximity to a lightning strike. The other computer in the building was protected by a UPS but not this one...
That sounds like a fairly heavy, low level problem to cause that kind of freeze, so I too would be looking at hardware or maybe drivers, as opposed to running programs and services.
Network cards and Hard disk buses seem to be two things that can cause the system to 'pause' fairly dramatically.
Network cards and Hard disk buses seem to be two things that can cause the system to 'pause' fairly dramatically.
I did a few more checks yesterday:
GPU core temperature stable at 65degC under very heavy load
CPU temperature peaked at 38degC
Mobo temperature stable at 31degC
Mobo supply voltages varied but stayed within 5% tolerance
All exactly as expected
The alternative power supply I had didn't have the right connector for the GPU or for the mobo (20pin ATX instead of 24pin) so I didn't attempt to swap it out
I spent a good 90 minutes yesterday trying to reproduce the problem but it was suspiciously absent. Until I can get it to happen again I will leave the system as-is...
Thanks for all the suggestions yesterday
GPU core temperature stable at 65degC under very heavy load
CPU temperature peaked at 38degC
Mobo temperature stable at 31degC
Mobo supply voltages varied but stayed within 5% tolerance
All exactly as expected
The alternative power supply I had didn't have the right connector for the GPU or for the mobo (20pin ATX instead of 24pin) so I didn't attempt to swap it out
I spent a good 90 minutes yesterday trying to reproduce the problem but it was suspiciously absent. Until I can get it to happen again I will leave the system as-is...
Thanks for all the suggestions yesterday
Just as a quick follow-up on this problem. I think I've found the root cause and is was rather unexpected.
I tried booting up yesterday and initially it would only go as far as the end of the post test. Not a good start at all.
Resetting the rig got a bit further - loading windows but then left hanging, I gave up after 20 minutes. The drive light was buzzing away but no progress.
Safe mode was no different - it gets to MUP.SYS and no further (been here before in the past)
Now having an inkling of what was wrong I pulled the slave drive on the primary IDE channel and tried again. This time no problem at all - all booted fine.
The slave drive works on the secondary channel (with the optical drive disconnected) so I will hazard a guess and say that it is the drive controller is faulty and blocking communication when the channel is shared. I will be dumping the contents of the drive to an archive and then sending it back for replacement under warranty.
It isn't the end of the story as the optical drive is causing problems but that is a problem for another day...
I tried booting up yesterday and initially it would only go as far as the end of the post test. Not a good start at all.
Resetting the rig got a bit further - loading windows but then left hanging, I gave up after 20 minutes. The drive light was buzzing away but no progress.
Safe mode was no different - it gets to MUP.SYS and no further (been here before in the past)
Now having an inkling of what was wrong I pulled the slave drive on the primary IDE channel and tried again. This time no problem at all - all booted fine.
The slave drive works on the secondary channel (with the optical drive disconnected) so I will hazard a guess and say that it is the drive controller is faulty and blocking communication when the channel is shared. I will be dumping the contents of the drive to an archive and then sending it back for replacement under warranty.
It isn't the end of the story as the optical drive is causing problems but that is a problem for another day...
Gassing Station | Computers, Gadgets & Stuff | Top of Page | What's New | My Stuff