Performance - dGPU Hitman | Tomb Raider | Far Cry Primal | Time Spy

Performance - Dedicated GPU

Let's grab the GeForce GTX Titan X (Pascal) and a couple of games. We test at our usage (Very High / Ultra) quality settings which is the way you guys game at home with your dedicated graphics card. But if you purchase a high-end system, surely you will not be playing at medium graphics quality, right? As such we test in a real-world usage replication.

Now we did opt for the GeForce GTX Titan X (Pascal) for a reason. We always talk about the dreaded CPU bottleneck, but the reality is that in 98% of the PCs out there, the graphics card is the bottleneck. If we'd used, say, a GTX 980, the scores starting at 1080P would be pretty much very close to each other. The benefit of having a GTX Titan X lying around, is that this card hardly poses a bottleneck, maybe at 2560x1440 the GPU bottleneck starts, but anything lower than that should really show performance differences with changes in CPU and/or memory.

If you look above you can see that very well, at 1440P we run into the GPU being the limitation. At 1080P we already see a bit of performance benefit from the faster memory. Then at 720P we can see the effect of memory once you do not have a GPU bottleneck. BTW all games are tested at proper image quality settings the way you are used to and set up similar as in our graphics card reviews. 

Tomb Raider 2016 is also dependent on system memory and processor performance, but even a Titan X is utilized harshly at 1080P. As such, the results will be much higher when using a lower resolution, once we drop to 720P, we can observe a, well let's call it 10% performance increase. But again, this chart shows that your graphics card is always a bigger bottleneck in the system opposed to CPU or memory.

There are of course games that are less harsh on that processor and memory subsystem, and really benefit from changes therein. As you can see, at 1080P we already see that 10% performance benefit, but that is also indicative for CPU limitation, and that shows at 720P as here we finally fell into a situation where the GPU is not the bottleneck, but the CPU. The effect with fast, low latency memory like this kit really shows itself with this specific game title (and a Nvidia GeForce GTX Titan X of course).

Just for brevity's sake, I added Time Spy and yes, the faster memory improves almost nothing really. Above is 2133 MHz, below is 3200 MHz.

Please let me remind you again that EVERYTHING on this page benchmarks wise is relative, as we used the Titan X Pascal. With a 1070 or 1080 or RX 480 or Fury X the GPU will be the limiting factor and the memory performance differential will be much smaller. But we are testing memory today, hence I wanted to show the effect it CAN have.