Overclocking Results

Overclocking a graphics card (or anything for that matter) and then reporting on it is always a tenuous adventure.  Our results could be abnormal on the high side or the low side, and taking it as gospel before picking your GPU option is usually an “at your own risk” decision.  That being said, what did we find?

First and foremost, it is worth noting that by using the latest NVIDIA driver, 335.23, the artificial limit of +135 MHz offset that existed on the reference card is now gone and you can push these Maxwell GPUs quite a bit farther than you could just a couple of weeks ago.  But, because all of these cards started at much clock rates than the reference card did we didn’t actually need that extra headroom.  Go figure!

We used the EVGA PrecisionX software for overclocking on all three cards to keep things simple (they all turn the same knobs on the GPU side).  We took the voltage option, which was limited to a +31mV add-on, and pulled it up to the top across the board and adjusted the GPU clock offset by 10 MHz increments running Metro Last Light for at least 10 minutes to test for stability before moving on.

EVGA’s GeForce GT 750 Ti FTW was able to hit an offset of +100 MHz, creating a base clock speed of 1289 MHz and a Boost clock of 1368 MHz. 

Galaxy Overclock

The Galaxy GTX 750 Ti GC was only able to get to the same +100 MHz offset, for a base clock of 1211 MHz and a Boost clock of 1289 MHz. 

PNY Overclock

Finally, the PNY, the surprise of the group, was able to get to +90 MHz offset for a base clock of 1292 MHz and a Boost clock of 1371 MHz, essentially matching the results of the EVGA card.

How did the real-world clocks pan out though?  Were these cards able to maintain Boost clock consistently?

This is a smaller subset of results to “zoom in” a bit, but we see once again that EVGA’s cooler is able to keep things smooth as glass, consistently staying at 1419 MHz (!!) on the GPU.  The Galaxy card was close to be able to do the same, but you can see the beginnings of some of that clock variance bubble up.  The PNY card on the other hand is hitting as high as 1423 MHz, but also falls to around 1300 MHz in some cases as the temperatures become an issue. 

As it turns out, with all that jumping around of the frequency on the PNY GTX 750 Ti OC, the EVGA card is actually averaging the highest overclocked speed at 1419 MHz.  This does beg the question – what is holding us back from getting even HIGHER on the EVGA card?  Perhaps we need another software update to let us move that power target?

Even though all three cards can overclock well above their stock speeds, and that equates to quite a bit over the reference speeds, clearly the EVGA card gives us the best result, followed closely by the PNY.  Galaxy’s GC card isn’t far behind and as I mentioned above, the same three cards being tested by someone else could end in a different result.  I do think that a 30% clock speed advantage for these cards when compared to the rated Boost clock of the reference GTX 750 Ti is something worth praising – clearly Maxwell has some headroom on the current 28nm process!

« PreviousNext »