Overclocking Performance, Benchmarks, Temperatures and Power Consumption
If you do need to squeeze out some more FPS, this page is for you.
Overclocking in general is not easy and is incredibly time consuming taking hours, days maybe even weeks to get it just right. Getting it just right means you have to make sure that there is no tearing or artifacting while the benchmark is running and of course no freezing or bluescreens.
I will show you my settings but just remember, not all cards overclock the same, even if you bought the same card. These overclocks are by far not the farthest I could push it, but I wanted to get this review out in a timely manner so just use these as a reference.
To start off, let me show you GPU-Z’s before and after screenshots and I highlighted the improvements.
I spent a lot of time on the GPU and left the memory behind, but you can of course go higher.
I overclocked the GPU by 130Mhz and the memory only by 5Mhz, but it does make a nice difference you can see below.
To overclock, I used EVGA’s Precision X OC, like GPU-Z I will paste below a before and after screenshot.
Using the “Curve” button
Which brings up the Curve menu
With that, you can set your curves on how you would like the fans to ramp up once the heat rises.
Clicking the Fan Curve button actually brings up this menu and the Fan curve menu, but I won’t go into this one right now, I will a little later in the review.
You are also provided 10 different profiles to save your work to. The way I used it was I overclocked a little, to where it was stable, then I saved it as profile 1. Using Profile 1, I overclocked a little more till it was stable, then I saved it as Profile 2. I used the same method each time till I reached profile 4.
Now on to the benchmarks.
So on the comparison, on the initial bench, we received a decent score of 20,654 and the overclock gave us a score of 21,652. While it looks amazingly high, there was only a 4.61% improvement and with that only a 2-degree bump in temperature. As for power, we only saw an increase of 0.29%.
Breaking down the results more, we can see that there was a 5.97% improvement in performance on the Graphics score from 28,477 up to 30,285. Keeping up with that improvement, the physics score increased by 4.051% from a decent score of 13,616 to 14,191. Lastly, the combined score was only bumped up 1.91% from 9,040 to 9,216. So all across the board we saw an improvement with only a slight overclock. Let’s see what in store for our future in Time Spy.
I need to start it off by saying, if you remember the initial results, they were only 93% better than all other results, but this tiny but in speeds gave us an extra percentage up to 94% better than all other results. Please remember, many if not all other results have the CPU and the memory, the physical DDR4 overclocked as well, everything here aside from the GPU and the negligible GDDR5X speeds are overclocked here. On to the comparison.
For the longest time it seemed people were trying to get the highest 3DMark score, to get to that top and now people are fighting to get the best Time Spy score. We’ve already climbed 1% on this tiny overclock, imagine what more can be done with a CPU and Memory overclock and some more attention to the video card, you can easily pass 10k, get some LN2 in there and you can even hit the famous K|NGP|N’s score… well maybe not but you can try. As of writing this review, K|NGP|N’s score was 32,089, but he is running these cards in Quad SLI.
The main Time Spy score here went from a decent 8,583 to a nice 9,146, that’s a 6.13% improvement. The biggest improvement here though was the Graphics score, that come up to a 10,277 from the prior score of 9,468, a 7.87% gain. Now the bad part was the Average wattage went up 12.53%, though oddly enough the temperature dropped an almost equal 12.20%. Not incredibly odd that the temperature dropped since we did bump up the fan curve.
Ok, now back to the game benchmarks.
For the games, we are going to break the result sections by resolution so that we don’t have a microscopic diagram. We will start off at 1920 x 1080.
1920 x 1080
With this tiny increase, we see a 4.96% increase in performance. Even though the increase was slight on the average frame rate, the see the Max wattage increase greatly and the average wattage increased by 7.31%, with the fan curve we dropped the temperature by 8.43%. Any decrease in temperature is welcome.
Let’s see if 2560 x 1440 does any better.
2560 x 1440
Here we can see there was a frame rate gain by 4.75% from 74.1 frames per second to 77.8 frames per second. You might notice that the Maximum frame rate on the stock is actually 34.70% above that of the overclocks 132.29 frames per second, but please remember while the Max is a nice to have, that is literally a blip of performance. By blip I mean, for a split second the score shot up to 202.59 then shot right back down to let’s say somewhere in the lower 100’s.
With that bump, the wattage again climbed up on the max side from 363 to 414, a 12.32% jolt of power but like the Max Frame Rate, it was a blip and only there for a second. The average power consumed rose 7.53%, which is not horrible but with the slight frame increase, not warranted, though this does mean there is still a lot of headroom for overclocking.
Now on to 4K.
3840 x 2160
Looks like I might need to work a little more on that overclock but at least we can see there was an improvement on the average frame per second count. On the overclock, we see a 40.74 frame per second mark, which is a 5.42% improvement over the stock clocked 38.53 frames per second. Here though we can see the average wattage an unwarranted 12.56% though the temperate did drop 7.23%. It looks like I may have raised the voltage too high, though soon I will show you how to adjust this and more, so stay tuned.
1920 x 1080
OK, now here we can see the overclock a little more in effect. We can see on average, there was a boost of 3.95% up from its original 119.3 frames per second to its new 124.2 frames per second. The average wattage consumed only increased 3.46% and the temperature dropped 7.70%. Please remember that increasing the fan curve will increase the fan noise, though ever so slightly in this case. Let’s move on to 2560 x 1440.
2560 x 1440
Once again, across the board we see an improvement. The average FPS went from 109.4 to 113.6, a 3.70% increase. Average wattage actually dropped slightly to 2.06%, down from 340 to 333 Watts, that always nice and following its precedence the temperature dropped down to 75°C from 81°C, a 7.41% decrease in temperature. What does 4K have in store for us?
3840 x 2160
It seems that the performance trend continues but the wattage consumption monster is back. On average, there was a 7.58% improvement in performance, from 70.7 frames per second to 76.5 frames per second. With that slight improvement, there was a 11.5% increase in power consumption, up 46Watts to 400 Watts. Thankfully the 7.32% cooling improvement makes up for a bit of that wattage spike, though you can drop it some by lowering that voltage some.
Alright well, at least we did see some more improvement with Thief, how receptive to this change do you think Laura will be?
1920 x 1080
The overclock continues to be effective in Tomb Raider as well, improving the average frame rate by 3.71% from 329.9 frames per second to 342, though at such a high frame rate was it needed? OF COURSE IT IS. That increase though resulted in a 11.37% increase in watts being used from 343 to 387Watts. The thermal improvement trend though continues down from 83°C to 74°C, a welcomed 10.84% improvement. Let’s check out 2560 x 1440.
2560 x 1440
Tomb Raider again benefits across the board with this minute overclock, while not a huge gain, a gain is a gain. On average, we can see a 4.37% increase in frames per second but with it comes the power monger sucking up an additional 9.56% more power at 387 versus 350. Thankfully the fan curve once again come s in to save temps coming in a 76°C, 8.43% better than its stock temps at 83°C. Let’s see if 4K does any better.
3840 x 2160
At 4K, there is almost identical results to that of the 2K set. On average, there is a slight increase in our frames per second count, a 4.83% improvement at 101.3 over 96.4 frames per second. A huge bump in power consumption is again occurring here, on the overclocked we are pulling an additional 12.28% watts at 391 coming in over the stock speeds at 343 Watts.
While there was an improvement in Tomb Raiders score, it was a tiny one on the already huge speeds this card provides. Let’s see if Tom Clancy’s The Division a more demanding and newer title can utilize this overclock a little better.
1920 x 1080
The Division is showing some promise at first, we can see a 6.85% improvement in average performance, 135.8 frames per second over 126.5 frames per second. The improvement brings on a bit larger of a power draw, on average it pulled in 406 Watts when on stock it only drew 355 Watts, that’s a 16.67% increase. Once again, the cooling is on top coming in at 76°C, 9.52% under the stocks cooling of 84°C. The overclock seems to have decreased the CPU utilization 14.75%, so the GPU did the most work.
This was the lower resolution, let’s see what 2560 x 1440 will bring us.
2560 x 1440
The overclock helped a bit here, on stock the card hit 90.3 frames per second but overclocked we saw a 7.86% improvement with scores a t 98 frames per second. Watts again on average we a bit high at 11% over the stock run at 356 Watts and again, the cooling improvement was 9.52% at 76°C over the stocks 84°C. The overclock here though seemed to rely a little more on the CPU with an increase of 12.97%. Last but not least, 3840 x 2160 is up next.
3840 x 2160
The overclock here makes the game a little more playable at 4K, reaching closer for that magic 60FPS. On average, the overclock came in at a decent 8.91% increase at 53.9 FPS over the stocks 49.1 FPS. The overclock increased the power consumption 13.99% at 393 Watts over the stock performance 338 Watts. Here the average CPU utilization was almost identical at a measly 5.74%, 35% on the overclock and 33% at stock.
It is evident the overclock here was not very good, but surely you can get it much higher, I may work on it some more later. There was plenty of headroom, but of course in any overclock you want stability over speed and they all take some time. Now, like the stock speeds, I will take some time to talk about the thermals.
Here you will notice the thermals never went above 77°C, a few points away from its 84°C maximum GPU temperature by design, many times it was lower than 77°C. The noise did go up a tiny bit, but with the side panel on it is barely noticeable, if anything it can and will be overlooked.
So here I showed you the basics on EVGA’s Precision X OC, but I will go into more detail on the next page and maybe you can overclock it tons more.