Overclocking 6600K

[trents]

But you passed four hours, right? That should be good enough.

 

[EarthDog]

Look at your event viewer and see.

 

[-Ice]

But you passed four hours, right? That should be good enough.

Yeah, just wanted to try it and see.

Look at your event viewer and see.

What am I looking for exactly?


New issue now; I tried enabling XMP for my RAM and I noticed my 5th monitor, the one connecte to the iGPU, keeps flickering. Never did that on non-XMP settings and stopped once I returned to non-XMP. Is this indicative of faulty RAM? I'll run memtest tomorrow maybe, but how/why would RAM XMP affect iGPU performance?

 

[trents]

Temporarily return the CPU to non overclocked state and see if the Monitor still flickers using RAM on XMP. It might just be that you need to bump the voltage up a tad on the IMC when CPU is overclocked.

 

[-Ice]

Just finished crunching some numbers and I would really like some input here as I cannot make sense of it.




On the first two tests, I had all 5 monitors connected and displayed. The first test had the flight sim displayed across all three display monitors on Surround configuration with a resolution of 5896x1080. The second test had the flight sim displayed only on the left-most monitor at 1920x01080 resolution but the display monitors were still on Surround configuration. I thought maybe because all 5 monitors were still giving a picture, maybe the GPU was drawing both the game and the desktop and that still equated to quite a load on the GPU as it still had to render 9736x1080 resolution even if the game was only on 1920x1080.

For the third test, I disabled the monitors via Windows Display Settings so I only had one 1920x1080 display monitor and two 1920x1080 touchscreens. Even if the GPU were drawing all three screens, it was only rendering 5760x1080 resolution so I should have better numbers. Nope. The difference between the 2nd and 3rd 4.5GHz test was only 4.8% and the difference between the 3.9GHz test was only 1.2%

GPU utlization for the first test (game on Surround) was 90.4% on both 3.9GHz and 4.5GHz and I guess that's why the average framerates are nearly identical. Second test (game on one screen) was 77% for 3.9GHz and 81% for 4.5GHz so I guess that's why the OC framerates are higher than stock framerates. The third test (5760x1080 res) only had 79% for 3.5GHz and 81% for 4.5GHz and so the framerate difference isn't too far apart.

With all that data, is the game really CPU or GPU bound? Or is there some other limit that I seem to be hitting? Vsync is not enabled in the game and is set to use the game settings in the nVidia control panel. I'm using 60hz screens.

 

[trents]

What I have always understood from everything I read, the way to test for GPU v. CPU bound is to lower the eye candy (level of detail) in the game settings. If the frame rates improve then the GPU was the bottleneck. If not, the CPU is the bottleneck.

I would suggest not using that old flight sim for this testing since its old engine may also be a limiting factor and complicate the results. Do you have a relatively new game you could repeat this test with?

 

[-Ice]

I understand what you're saying, but testing on a different sim, the "logic" may not necessarily translate to this old flight sim. The primary reason of doing tests here is because this is what I spend most of my free time on. Getting something like Project Cars 2 from 30fps to 90fps is useless if I can't translate that into Falcon BMS, if that makes sense. That's also the reason for testing in both Surround and 1-screen-only setups; to see exactly how much performance I am losing with Surround and being able to see whether it is worth the framerate drop.

I can lower the graphics settings to medium or low in BMS.