waltherone
Gawd
- Joined
- Aug 12, 2003
- Messages
- 589
Wow, this ended up LONG, but if you don't like to read about computer parts, I don't know why you're on this forum
I wasn't going to post this for fear of people thinking I was a moron, so take it with a grain of salt, but it's becoming more consistent for me now.
I've got an eVGA 7900gt that was for the most part worked flawlessly for me from the start. I did the 1.4v mod(via the conductive ink pen) a few days after getting it, and was running it for a while at 640/865, cooled via an NV Silencer (and inside a P180 with a big sanyo-denki fan on the front, and all other fans on high...my case sounds like a jet engine when I turn all the fans up for gaming). I put a Vantec ramsink on the rectangular IC (in the following picture jacked from XS forums, I put the sink on the IC that's on the right of the two that are circled).
http://home.earthlink.net/~waltherone/images/HotICs.jpg
I didn't sink both of them because, in case no one noticed, they aren't the same height, so you can't use a single sink to effectively cover both of them. The two smaller ones are about 1 or 1.5mm shorter than the rectangular. So I left that one unsinked, and sinked the rectangular. It ran for a good while at those speeds playing a lot of oblivion and benchmarking since it was a new machine. As it was a new machine, maybe a month later I decided to sleeve all my wires to neaten the thing up a bit on the inside. Once I reconnected everything, the wire on this DFI board that connects the Molex and Floppy connectors just below the CPU was a bit stiffer due to the sleeving, and eventually (with help from heat while gaming, etc softening the thermal tape) pushed the sink off the IC and left it laying diagonally against the IC and any solder points next to that IC. This happened WHILE gaming I can only assume, as I'd been playing oblivion for an hour or so when suddenly I got a symptom many of you are probably familiar with, the game just froze, computer wouldn't respond. Hit Ctrl-Alt-Del, even that wouldn't respond. Then about 20 seconds later, everything caught up, the game minimized and the system tray was open, mouse would move, all was well. I click the Oblivion button on the taskbar to get back into game, as soon as it opens, all freezes again. So I thought, ok, wtf. I exit the game, reboot, go back and start testing, to find that the computer locks up any time I try doing ANYTHING 3D. 2D stuff is fine, but any time the video card has to process any 3d stuff, all goes to hell. I opened the computer, found the sideways ramsink, removed it (leaving the IC unsinked) and all was well again, so I thought yay, false alarm, and nothing BAD was shorted when it fell off and landed on all those solder points.
Well, not long after, it started happening again. Like before, if i did auto detect in coolbits, it would automatically detect the default speeds for the card. If I try anything higher, it would work until the card heated up a bit under load, and then the same symptoms. If the card was already hot via my tinkering with testing and such, and I tried to increase speeds over stock, the coolbits test would lock up the computer for a bit, or cause the screen to flash until it went back to normal and told me the test failed at those settings. So I got curious, and pulled the card, checked my volt mod just for giggles, and put another sink with fresh thermal tape back on that same IC, made sure that sleeved wire was tucked/tied away, and put the card back in. And everything functioned well again, until today when I got curious again.
I pulled the card this afternoon before work, and put two ramsinks next to each other, with one sitting on the two smaller square IC's (making sure not to touch the small resistor to the left of them, C502 in the picture) and the second ramsink on the rectangular IC like before, just moved over a bit kind of hanging off the right side (right side if you're looking at the above picture), so the two sinks were right up against each other. I put the card back in, ran 3dmark06 first, it was artifacting like CRAZY just on the first scene alone. Closed that, opened ATI Tool, the artifact and the regular, rotating 3d test block were artifacting BADLY as well. I'm not talking little dots, I'm talking big streaks of color all over the image. So I thought ok, what the hell, I can already guess the problem. Removed the card again, took off the rectangular IC's ramsink first, and noticed that the IC I had sitting on the two smaller square IC's (the two to the left in the above picture) was also resting against the pins coming out of the rectangular IC on the left (in the above picture) side. Removed that sink all together, put one back on just the rectangular IC like I had before, plugged the card back in, and now here I am, just ran a full pass of 3dmark06 at 640/850 speeds, and ran ATI tool for a lil while, no artifacts for until about a minute and a half in, no artifacting in Oblivion at all.
Now the conclusion to all this rambling: That little cluster of IC's is VERY close together, and there are a couple of other components (resistor, etc) nearby just outside the cluster. You guys make sure your shit isn't shorted I've gotten the same symptoms as a LOT of people I've read about on eVGA's forums, and on this forum. These people are RMAing their cards, getting replacements, and they're still getting the same symptoms. NOTE, without any ramsinks at all, I get those symptoms as well as soon as the card heats up, so it obviously can happen to stock cards. But for all the folks modding their cards, I've also gotten those same symptoms twice now, under two different occurrances, when the ramsink shorted some pins on a nearby part. Those IC's are very tiny, and very close together, and ramsinks are typically a standard size more or less. If you used any thermal paste to apply your ramsinks, you could very well have shorted some pins on one of those IC's. They are VERY tiny, and the pins are close enough together to barely even get a needle point in between them. It's not exactly hard to short them to each other. Also note, the card functions fine much of the time, just artifacts like MAD, and, depending on what is shorted to what, might or might NOT have the freezing/flashing screen symptom.
Also, I read either on here or on XS forums that on the GTX cards, these IC's are on the front/top of the card(the side with the GPU). So they're getting a little breeze from that huge fan on the GTX, which partly explains why these problems are mostly in the GT cards. A few GTX owners have chimed in with issues, but not even 1/4 of the number of GT card owners that have problems.
Tomorrow starts the new project, filing down half of the underside of one of my ramsinks so I can mount a single sink covering all three IC's without shorting them to each other or anything else (one of these Vantec sinks covers them perfectly, but as stated before, they're not the same height).
Just some food for thought, take it with a grain of salt or take it as gospel, I know there will be a little of both on this site But it might be worth pulling your card and looking CLOSELY at.
I wasn't going to post this for fear of people thinking I was a moron, so take it with a grain of salt, but it's becoming more consistent for me now.
I've got an eVGA 7900gt that was for the most part worked flawlessly for me from the start. I did the 1.4v mod(via the conductive ink pen) a few days after getting it, and was running it for a while at 640/865, cooled via an NV Silencer (and inside a P180 with a big sanyo-denki fan on the front, and all other fans on high...my case sounds like a jet engine when I turn all the fans up for gaming). I put a Vantec ramsink on the rectangular IC (in the following picture jacked from XS forums, I put the sink on the IC that's on the right of the two that are circled).
http://home.earthlink.net/~waltherone/images/HotICs.jpg
I didn't sink both of them because, in case no one noticed, they aren't the same height, so you can't use a single sink to effectively cover both of them. The two smaller ones are about 1 or 1.5mm shorter than the rectangular. So I left that one unsinked, and sinked the rectangular. It ran for a good while at those speeds playing a lot of oblivion and benchmarking since it was a new machine. As it was a new machine, maybe a month later I decided to sleeve all my wires to neaten the thing up a bit on the inside. Once I reconnected everything, the wire on this DFI board that connects the Molex and Floppy connectors just below the CPU was a bit stiffer due to the sleeving, and eventually (with help from heat while gaming, etc softening the thermal tape) pushed the sink off the IC and left it laying diagonally against the IC and any solder points next to that IC. This happened WHILE gaming I can only assume, as I'd been playing oblivion for an hour or so when suddenly I got a symptom many of you are probably familiar with, the game just froze, computer wouldn't respond. Hit Ctrl-Alt-Del, even that wouldn't respond. Then about 20 seconds later, everything caught up, the game minimized and the system tray was open, mouse would move, all was well. I click the Oblivion button on the taskbar to get back into game, as soon as it opens, all freezes again. So I thought, ok, wtf. I exit the game, reboot, go back and start testing, to find that the computer locks up any time I try doing ANYTHING 3D. 2D stuff is fine, but any time the video card has to process any 3d stuff, all goes to hell. I opened the computer, found the sideways ramsink, removed it (leaving the IC unsinked) and all was well again, so I thought yay, false alarm, and nothing BAD was shorted when it fell off and landed on all those solder points.
Well, not long after, it started happening again. Like before, if i did auto detect in coolbits, it would automatically detect the default speeds for the card. If I try anything higher, it would work until the card heated up a bit under load, and then the same symptoms. If the card was already hot via my tinkering with testing and such, and I tried to increase speeds over stock, the coolbits test would lock up the computer for a bit, or cause the screen to flash until it went back to normal and told me the test failed at those settings. So I got curious, and pulled the card, checked my volt mod just for giggles, and put another sink with fresh thermal tape back on that same IC, made sure that sleeved wire was tucked/tied away, and put the card back in. And everything functioned well again, until today when I got curious again.
I pulled the card this afternoon before work, and put two ramsinks next to each other, with one sitting on the two smaller square IC's (making sure not to touch the small resistor to the left of them, C502 in the picture) and the second ramsink on the rectangular IC like before, just moved over a bit kind of hanging off the right side (right side if you're looking at the above picture), so the two sinks were right up against each other. I put the card back in, ran 3dmark06 first, it was artifacting like CRAZY just on the first scene alone. Closed that, opened ATI Tool, the artifact and the regular, rotating 3d test block were artifacting BADLY as well. I'm not talking little dots, I'm talking big streaks of color all over the image. So I thought ok, what the hell, I can already guess the problem. Removed the card again, took off the rectangular IC's ramsink first, and noticed that the IC I had sitting on the two smaller square IC's (the two to the left in the above picture) was also resting against the pins coming out of the rectangular IC on the left (in the above picture) side. Removed that sink all together, put one back on just the rectangular IC like I had before, plugged the card back in, and now here I am, just ran a full pass of 3dmark06 at 640/850 speeds, and ran ATI tool for a lil while, no artifacts for until about a minute and a half in, no artifacting in Oblivion at all.
Now the conclusion to all this rambling: That little cluster of IC's is VERY close together, and there are a couple of other components (resistor, etc) nearby just outside the cluster. You guys make sure your shit isn't shorted I've gotten the same symptoms as a LOT of people I've read about on eVGA's forums, and on this forum. These people are RMAing their cards, getting replacements, and they're still getting the same symptoms. NOTE, without any ramsinks at all, I get those symptoms as well as soon as the card heats up, so it obviously can happen to stock cards. But for all the folks modding their cards, I've also gotten those same symptoms twice now, under two different occurrances, when the ramsink shorted some pins on a nearby part. Those IC's are very tiny, and very close together, and ramsinks are typically a standard size more or less. If you used any thermal paste to apply your ramsinks, you could very well have shorted some pins on one of those IC's. They are VERY tiny, and the pins are close enough together to barely even get a needle point in between them. It's not exactly hard to short them to each other. Also note, the card functions fine much of the time, just artifacts like MAD, and, depending on what is shorted to what, might or might NOT have the freezing/flashing screen symptom.
Also, I read either on here or on XS forums that on the GTX cards, these IC's are on the front/top of the card(the side with the GPU). So they're getting a little breeze from that huge fan on the GTX, which partly explains why these problems are mostly in the GT cards. A few GTX owners have chimed in with issues, but not even 1/4 of the number of GT card owners that have problems.
Tomorrow starts the new project, filing down half of the underside of one of my ramsinks so I can mount a single sink covering all three IC's without shorting them to each other or anything else (one of these Vantec sinks covers them perfectly, but as stated before, they're not the same height).
Just some food for thought, take it with a grain of salt or take it as gospel, I know there will be a little of both on this site But it might be worth pulling your card and looking CLOSELY at.