- Jun 30, 2004
- 16,559
- 1,984
- 126
Not intending this to grow into a long discussion, but a few questions. I'd offered advice to a forum brother here:
http://forums.anandtech.com/showthread.php?t=2437398
And I'll be candid: I'm still fairly new to graphics overclocking, but inclined to think I've mastered it. Of course, many will say that my processors are "easy overclockers," and the new utilities like Precision and Afterburner make graphics overclocking easy and effortless.
In the process of providing a link to a GTX 980 OC'ing guide in the above thread, I noticed that it promoted the idea of using 3DMark for stressing and testing. Nor am I new to that; seems like I'd purchased an old version before. So I got it again.
I've run through the Firestrike Extreme benchmark on this puppy (i7-2700K @ 4.7; 16GB 4x4 @1600 and 9-9-9-24).
I overclocked my 2x SLI GTX 970s to 1,468Mhz core and 7,600Mhz memory.
Now, I see that (whoop-it-ee-doo) I score as "better than 87% other comparable configurations" or "other results as a whole" -- whatever the precise wording -- I'll have to check again. And fine -- I'm not looking for bragging rights. I just want decent to good results.
But I also see that my core clocks on the 970s never went over the stock values of 1,316 and 7,000. That is to say, these were the values registering in AfterBurner's GPU clock monitoring. Of course, for air-cooling, it looked as good as Ice-Station-Zebra: the temperatures never exceeded 68C for the top and hotter card.
Is this normal -- for 3DMark to limit the clocks to the stock value, even if the overclocks were effective before and after the test?
And why does Futuremark do this? Or -- would this be for my deferring the download of the latest upgrade to 3DMark?
Put this way -- is it their intention to provide an absolute hardware benchmark? What are they doing there? Is there another menu option I haven't discovered?
http://forums.anandtech.com/showthread.php?t=2437398
And I'll be candid: I'm still fairly new to graphics overclocking, but inclined to think I've mastered it. Of course, many will say that my processors are "easy overclockers," and the new utilities like Precision and Afterburner make graphics overclocking easy and effortless.
In the process of providing a link to a GTX 980 OC'ing guide in the above thread, I noticed that it promoted the idea of using 3DMark for stressing and testing. Nor am I new to that; seems like I'd purchased an old version before. So I got it again.
I've run through the Firestrike Extreme benchmark on this puppy (i7-2700K @ 4.7; 16GB 4x4 @1600 and 9-9-9-24).
I overclocked my 2x SLI GTX 970s to 1,468Mhz core and 7,600Mhz memory.
Now, I see that (whoop-it-ee-doo) I score as "better than 87% other comparable configurations" or "other results as a whole" -- whatever the precise wording -- I'll have to check again. And fine -- I'm not looking for bragging rights. I just want decent to good results.
But I also see that my core clocks on the 970s never went over the stock values of 1,316 and 7,000. That is to say, these were the values registering in AfterBurner's GPU clock monitoring. Of course, for air-cooling, it looked as good as Ice-Station-Zebra: the temperatures never exceeded 68C for the top and hotter card.
Is this normal -- for 3DMark to limit the clocks to the stock value, even if the overclocks were effective before and after the test?
And why does Futuremark do this? Or -- would this be for my deferring the download of the latest upgrade to 3DMark?
Put this way -- is it their intention to provide an absolute hardware benchmark? What are they doing there? Is there another menu option I haven't discovered?
Last edited:
Facebook
Twitter