I went too far and damaged my video card
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
I went too far and damaged my video card
Greetings,
I have always underclocked my video card with the notion that underclocking required very little scrutiny when compared to overclocking. That turned out to be false as I have learned the hard way.
I have a Geforce4 Ti 4600, cooled by a Zalman VF700. Due to the greater power draw of the 4600, my modded PSU's temps increased quite a bit, and where there is heat, there is noise. The 4600's stock speed is 300MHz core, 650MHz mem. I typically underclock to 200MHz core, 450MHz mem for basic desktop computing and when playing older games. This has lowered my modded PSU temps by about 1-2°C, which is not much of an improvement. I left it at that clock speed for some time, until recently when I decided to go significantly lower. I was aware that GPUs and RAM chips have certain limits as to what clock speeds they operate on, but I did not know that clocking too low can cause permanent damage. So I decided to try the extreme 75MHz core, 160MHz mem. The initial tesing went great. The video was stable, PSU temps were reduced by about 10°C and its associated fan noise was reduced to my satisfaction. After one week of use, the video started to show various artifacts from vertical lines to square dots, indicating that a memory chip is damaged. I tried pushing all the fans to full speed, resetting the clock speeds, reinstalling the 4600, but it is too late. The damage is permanent. It looks like I went too low for the memory clock and damaged a chip in the process.
Although this did not kill my video card (I am typing this with the damaged 4600), I will not be able to play games or watch any form of video. The next time you decide to do any underclocking, please be careful, and avoid the extreme until the proper research is done.
I have always underclocked my video card with the notion that underclocking required very little scrutiny when compared to overclocking. That turned out to be false as I have learned the hard way.
I have a Geforce4 Ti 4600, cooled by a Zalman VF700. Due to the greater power draw of the 4600, my modded PSU's temps increased quite a bit, and where there is heat, there is noise. The 4600's stock speed is 300MHz core, 650MHz mem. I typically underclock to 200MHz core, 450MHz mem for basic desktop computing and when playing older games. This has lowered my modded PSU temps by about 1-2°C, which is not much of an improvement. I left it at that clock speed for some time, until recently when I decided to go significantly lower. I was aware that GPUs and RAM chips have certain limits as to what clock speeds they operate on, but I did not know that clocking too low can cause permanent damage. So I decided to try the extreme 75MHz core, 160MHz mem. The initial tesing went great. The video was stable, PSU temps were reduced by about 10°C and its associated fan noise was reduced to my satisfaction. After one week of use, the video started to show various artifacts from vertical lines to square dots, indicating that a memory chip is damaged. I tried pushing all the fans to full speed, resetting the clock speeds, reinstalling the 4600, but it is too late. The damage is permanent. It looks like I went too low for the memory clock and damaged a chip in the process.
Although this did not kill my video card (I am typing this with the damaged 4600), I will not be able to play games or watch any form of video. The next time you decide to do any underclocking, please be careful, and avoid the extreme until the proper research is done.
Damage due to age is possible, but the card was lightly used until about a year ago when it was moved into my rig and received daily use since. Look at what I replaced the 4600 with: my really old TNT2. It has gone through almost nine years of heavy use and still going strong! So I have ruled that out from the beginning.Immortals wrote:Only in the sense that you're using more voltage than necessary... still shouldn't ruin the card. Has the OP considered it could just be a coincidence since it is quite an old card anyway?
I'd say it's much more likely that the card simply died of "natural causes" than by undervolting/underclocking. Perhaps an old ESD damage that just manifested itself, or something else.
That your old TNT2 is still going strong doesn't support any conclusion really. Older hardware is often more robust than newer stuff (easier to accidentally fry something made with a smaller process I suppose).
I have undervolted and underclocked tons of graphic cards and none of my cards have died in a way that made me suspect undervolting/underclocking to be the cause of their demise.
Have you checked the caps on the 4600 btw? A friend of mine once had a 4200 with bad caps.
That your old TNT2 is still going strong doesn't support any conclusion really. Older hardware is often more robust than newer stuff (easier to accidentally fry something made with a smaller process I suppose).
I have undervolted and underclocked tons of graphic cards and none of my cards have died in a way that made me suspect undervolting/underclocking to be the cause of their demise.
Have you checked the caps on the 4600 btw? A friend of mine once had a 4200 with bad caps.