• 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.

Is there any (semi) automated way to test a video card?

Commissario
Joined
17 Oct 2002
Posts
33,490
Location
Panting like a fiend
Hi

I've got a 7850 here which I'm fairly sure is faulty, but only shows the problem up in games (anything from a couple of minutes to several hours*, usually 30 minutes+ if it's going to do it).

I've rebuilt my old machine (q6600, 4gb ram etc) up to rule out problems with my newer one (which has been running without a crash since I swapped out the 7850).

But I don't really have the time/patience** to spend playing a game on the older machine until it crashes as it's so random that I could be playing it for 2-3 hours+ and not have one.

Hence, are there any programmes or ways to run a videocard as if it was in a normal game? (IE not something that's going to stress it to the max, but use it like you were running around/in and out of areas).
I'm thinking something like a game demo, but with it going from area to area rather than just the one level? (so it's having to actually load/unload stuff from the card multiple times).

About the only thing that is immediately noticeable with the 7850, is that in Firefox I can almost immediately get some odd squares on the screen in patterns, but I've read that can be FF being flakey (I've noticed them on the test rig which was a completely clean install of windows 7 pro, with the latest FF etc).



*It's extremely unlikely to be heat related as the machine was running ok when my room was hitting 30c, but sometimes crashing within a few minutes when the room is <25c (HW monitor suggests the videocard only ever reaches about 60-64c max).


**I rebuilt and reinstalled the old machine a couple of weeks back, but have only had time for a couple of sessons (combination of being under the weather, decorating, and running around after family :p)
 
Very simple, find some overclocking software for the GFX card and test at 100mhz overclock on core and memory.

You should have 50~100 stability margin for reliability to cover higher heat soak in extended gaming sessions etc....little point in testing anything at factory settings as you will learn nothing about stability margins which will be reduced in the cards later life for various reasons ;)

The same goes for your CPU don't forget and regardless of the fault something is going to give away the source of the problem.
 
Hi

I've got a 7850 here which I'm fairly sure is faulty, but only shows the problem up in games (anything from a couple of minutes to several hours*, usually 30 minutes+ if it's going to do it).

I've rebuilt my old machine (q6600, 4gb ram etc) up to rule out problems with my newer one (which has been running without a crash since I swapped out the 7850).

But I don't really have the time/patience** to spend playing a game on the older machine until it crashes as it's so random that I could be playing it for 2-3 hours+ and not have one.

Hence, are there any programmes or ways to run a videocard as if it was in a normal game? (IE not something that's going to stress it to the max, but use it like you were running around/in and out of areas).
I'm thinking something like a game demo, but with it going from area to area rather than just the one level? (so it's having to actually load/unload stuff from the card multiple times).

About the only thing that is immediately noticeable with the 7850, is that in Firefox I can almost immediately get some odd squares on the screen in patterns, but I've read that can be FF being flakey (I've noticed them on the test rig which was a completely clean install of windows 7 pro, with the latest FF etc).



*It's extremely unlikely to be heat related as the machine was running ok when my room was hitting 30c, but sometimes crashing within a few minutes when the room is <25c (HW monitor suggests the videocard only ever reaches about 60-64c max).


**I rebuilt and reinstalled the old machine a couple of weeks back, but have only had time for a couple of sessons (combination of being under the weather, decorating, and running around after family :p)

Aside from the squares in firefox, which is a known bug regarding Mozilla and AMD drivers which Mozilla need to fix, what are the other problems you mention? Do you get crashes, black screens, driver stopped responding messages or coloured stripey line images where the pc locks up until you restart?
 
If it was loaded for long periods at 30C ambient maybe the solder points may have been ruined. Just a thought though, exact nature of problem would be needed to say more.
 
The problem is basically the computer goes black & locks up, then I have to reboot - sometimes it'll restart fine, sometimes it'll need the machine to be turned off fully, then back on to get it to start (it'll give me a "no graphics card" beep, or if I've left the onboard video on as well*, resort to using that with the 7850 not found if I just press reset).

It was happening at random before the room hit 26c , but didn't actually happen once when the room was at 30c during the hottest week of the summer (so far).

I've been running it in my Haswell rig, and my q6600 (which is the one I've rebuilt/installed as a test), It was crashing in the q6600 back at the end of may, start of june, then seemed fine in the Haswell for a while (but that may have just been I didn't play many games in june due to a couple of weeks of being rather ill at the end of june/start of july).

I'm going to install War Thunder, Sim City and Mechwarrior Online on the q6600, as when I had a similar problem with my original 7850 they were showing artefacts/crashing it far faster than Borderlands 2 did.

The squares in FF are oddly reassuring, I'd read about it being an issue but that seemed to be for older versions, so wasn't sure if it was a problem with the 7850 (I don't get them with the 5770).



*After it started happening on the Haswell I turned the onboard video on.
 
Try bumping the core voltage two notches in afterburner or trixxx and see if that cures it.
 
The problem is basically the computer goes black & locks up, then I have to reboot - sometimes it'll restart fine, sometimes it'll need the machine to be turned off fully, then back on to get it to start (it'll give me a "no graphics card" beep, or if I've left the onboard video on as well*, resort to using that with the 7850 not found if I just press reset).

It was happening at random before the room hit 26c , but didn't actually happen once when the room was at 30c during the hottest week of the summer (so far).

I've been running it in my Haswell rig, and my q6600 (which is the one I've rebuilt/installed as a test), It was crashing in the q6600 back at the end of may, start of june, then seemed fine in the Haswell for a while (but that may have just been I didn't play many games in june due to a couple of weeks of being rather ill at the end of june/start of july).

I'm going to install War Thunder, Sim City and Mechwarrior Online on the q6600, as when I had a similar problem with my original 7850 they were showing artefacts/crashing it far faster than Borderlands 2 did.

The squares in FF are oddly reassuring, I'd read about it being an issue but that seemed to be for older versions, so wasn't sure if it was a problem with the 7850 (I don't get them with the 5770).



*After it started happening on the Haswell I turned the onboard video on.

Sounds more like a dry connection on a component or the PCIx slot/holder.
For most people in a case like this you would be best to just try to narrow it down to Board or Card by substitution.
 
Cheers Jakus, I'm currently running a 5770 in my Haswell system (without issue), and moved the 7850 to my old machine for testing (substituting the card, as you say to try and narrow it down).
My gut reaction is it's got to be the card, as it was doing it on the Q6600 before I built the Haswell system, it just hasn't showed up since I rebuilt the q6600 (probably as I haven't had time to run it much*), and pretty much all the internal hardware has changed between the two builds (everything bar a couple of non boot drives).

It's just so irritating as it's so erratic and time consuming to test:) not to mention this is the second card of this model I've had (the first one went wrong with similar symptons - ie crashes and not being detected at boot - it also had fairly major graphical corruption in Simcity, but I had more time to test it back then).

Ltmatt, I'll try pushing up the voltage on the card if it crashes in the old machine.


*Only about 5-10 hours of testing, in mainly shortish bursts (hence something I could leave running would be great).
 
Intermittent faults can be a real problem to track down :(

If the card was made properly and thoroughly tested by yourself when new, you can expect to get a good long life out of it providing you keep the temps sensible and are very careful handling it etc, electronics does not normally fail until nearing the end of it's design life, the soldering of the electronics to the boards is much more likely to be unreliable especially the led free solder which is everything nowadays, due to it being brittle. High temp swings and card flexing is likely to speed up fracturing of these microscopic joints.

When we used to have to find these problem components/connections you would try everything from flexing to local heating or cooling and still they were hard to find.

The reason I said the above was because I wonder how many warranty refurbs have just had a quick trip through the oven :p which is unlikely to be as reliable as a new card !
 
As you say lead free solder has a lot to answer for nowadays :(
IIRC it's also susceptible to micro crystals forming which can cause random shorts that are very hard to spot, but tend to have catastrophic results for the assembly, I remember reading something about NASA trying to work out what causes them as they're worried about it destroying something expensive in orbit if they ever start to use lead free solder (I think aerospace, military and medical are currently exempt from the ban).

I'm deliberately leaving the test machine on at the moment, as that's what I normally do with my machine (and what I used to do when it was my main PC), so the card is under the same sort of conditions but in a different machine.
I'm also monitoring the temperatures quite closely :)
 
Back
Top Bottom