I installed a Gigabyte 670 today and it seems pretty great, at least most of the time.
After playing around with clocks a bit I went into BF3 to test it out, and everything ran as it should, but trouble began when playing WoW. I got a black screen that lasted maybe 4-5 seconds, as if the display driver had stopped working, and then the core reset to stock, un-boosted clock (980 MHz) and memory clocks dropped all the way down to 162 MHz! After a few minutes memory clock went back up, but I had to restart Precision X to re-apply core clock (which I guess is normal).
I'd only applied a very modest 35 MHz OC to the core and none to the memory as it's my first nVidia card so wanted to start out slow; according to GPU-Z the GPU never went above 73% of the TDP and temps were hitting low 50s. I checked the GPU-Z log file and noticed the voltage (which, when under load, always runs as 1.175) had dropped down to 0.9xx mv at the same time the clocks went haywire. I'm wondering whether my PSU is being run to its limit (it only has 408W on the 12v rail) or if it's WoW that's being weird (it didn't happen in BF3).
I've only got a single HDD, a 2500k at 4.2 GHz and 3 fans running so I'd be surprised if the PSU is actually running at full capacity.
I read this guide: http://www.overclock.net/t/1265110/the-gtx-670-overclocking-master-guide
and noticed my "Kepler Boost" seems to vary from 52 to 13, depending on the clock offset. The boost clock is 1058 at stock and the card runs at 1110 under load (+52 Mhz) but if I OC the core, the Kepler Boost simply doesn't 'boost' as much, despite temps still being in the 50s.
I'm really at a loss here and wondering whether I should get a more capable PSU, return the card or admit defeat and run it at stock (which is really a very boring way of using an expensive graphics card).
After playing around with clocks a bit I went into BF3 to test it out, and everything ran as it should, but trouble began when playing WoW. I got a black screen that lasted maybe 4-5 seconds, as if the display driver had stopped working, and then the core reset to stock, un-boosted clock (980 MHz) and memory clocks dropped all the way down to 162 MHz! After a few minutes memory clock went back up, but I had to restart Precision X to re-apply core clock (which I guess is normal).
I'd only applied a very modest 35 MHz OC to the core and none to the memory as it's my first nVidia card so wanted to start out slow; according to GPU-Z the GPU never went above 73% of the TDP and temps were hitting low 50s. I checked the GPU-Z log file and noticed the voltage (which, when under load, always runs as 1.175) had dropped down to 0.9xx mv at the same time the clocks went haywire. I'm wondering whether my PSU is being run to its limit (it only has 408W on the 12v rail) or if it's WoW that's being weird (it didn't happen in BF3).
I've only got a single HDD, a 2500k at 4.2 GHz and 3 fans running so I'd be surprised if the PSU is actually running at full capacity.
I read this guide: http://www.overclock.net/t/1265110/the-gtx-670-overclocking-master-guide
and noticed my "Kepler Boost" seems to vary from 52 to 13, depending on the clock offset. The boost clock is 1058 at stock and the card runs at 1110 under load (+52 Mhz) but if I OC the core, the Kepler Boost simply doesn't 'boost' as much, despite temps still being in the 50s.
I'm really at a loss here and wondering whether I should get a more capable PSU, return the card or admit defeat and run it at stock (which is really a very boring way of using an expensive graphics card).
Facebook
Twitter