Originally Posted by yofghi
I'm not sure about using evbot as I have yet to try it but when running benches, whete do you see this 573 wall/fail safe kick in? I noticed that evga precision bugged on me during benches and was showing me a similar clock while benching. I thought it was from an unstable oc but it was just precision. GPUz was showing correct clock speeds and so was the OSD from rivatuner. As for the 7ghz mem clock that's really normal for these cards. I'm stable at 1.3v with a 1372core/3505 or 7ghz memory. A lot of people have reported 7+ghz stable clocks. If you are basing your clock off of what evga is showing I would get a few more monitoring programs to verify. Also make sure you aren't pushing past your max power. I haven't sen it before but may have something to do with it also. 1.4v is a lot to hold under 115% power.
Awesome man. Thanks! +1 rep
Yes, solely using Precision to monitor the speeds. Humm, I'll setup GPU-Z and other tools going forward. Yes, it was Precision showing the "wall" being hit and showing the clock drop to 573 and remaining there. A few other tell-tail signs is that the EVGA OC Scanner tool, using a combination of GPU and PhysX, stops the PhysX part when this "wall" is hit. And also, the FPS in the test drops from 95 FPS down to 52 FPS when the "wall" is hit.
I thought all of these things failing all indicated a "fail-safe" kicking in. Or, maybe OC Scanner sees Precision X reporting "too low of a clock" and just stops the PhysX part. I'll try other tools to bench as well (Heaven, Valley, etc). I was just installing and trying things out quick before returning the card.
As for watching power target, yep I keep a very close eye on that and walk the line around 105 to 110% (with the LN2 bios of 115% max). I then use a combination of "lowering the voltage/increasing core clock" and that's where I get to 1330 before the "wall" is hit. Everything only lasts 30s to a minute, before the wall.
I should mention that I leave the memory at stock when I am trying to find the core's limit. Sometimes I'll bump voltage to memory and PEX via EVBot - doesn't seem to matter.
I should also mention that I never get a BSoD, no apps crash, no "Display Driver" crashing, etc - all things typical of a failed overclock. The only thing that happens that prevents me from overclocking is this "Wall" being hit - and that's it. Now, if I just set something like +250, sure that locks up of course.
The last thing I find odd is the Precision X Overvolting tool: For example: If I don't touch it and set EVBot to 1.3V, volt meter showing 1.34V indicating it is indeed getting 1.3V, then the highest clock only goes to like 1228 or something in the high 1100s (with like +160, just to push it). I would expect this to be enough; but, it isn't at +148, +150, +160, etc. Only when I move the Overvolting tool in PrecisionX to max does the clocks immediately jump to 1319 or 1330, as expected.
Starting to raise an eye brow at PrecisionX...
I keep the temps down around 65 to 70 degrees. Maybe 75C if pumping a lot of volts. I know to stay away from the 80C mark and have seen the thermals kick in, yep. That's why I am dumbfounded at this "573 Mhz" wall - cause I only see 64C on the temp, PT only at 98 to 103%, etc. I don't see what is triggering this "wall"; but, I'll try other monitoring tools to see if it is really there.
Btw, if it matters: 4930k @ 4.8 Ghz, bus at 100.3, memory at 2133, Asus RIVE w/Gen3 in BIOS but no NVidia hack (stilla t PCIe 2.0), 1000W PSU.Edited by eduncan911 - 10/21/13 at 8:38am