BF3 & 4 - Hyperthreading on or off, 4 core vs. 6 core, overclocking... oh my!

[Kitlope]

[I posted this first post a few months ago on TLB, some interesting Intel hyperthreading findings, second post was a little experiment I tried just this morning....read on]

I'm sure this will apply much for Bf4 since it's an upgraded version of frostbite engine, hence a big part of the reason why I'm posting. I've been messing around a lot with my pc the last couple weeks since SLI'ing it, trying to squeeze as much performance as I possibly can, mostly because of trying to get as close to a constant 120 frames to go with the 120 HZ monitor (and for lightboost to work properly, these high frames are needed). I've also decided to overclock the processor a bit as I think it's becoming CPU limited. Here's a couple good write-ups I stumbled across on whether to have Intel's hyperthreading enabled or not and if a 6 core processor makes a difference versus a 4 core. Throw some overclocking as well into the mix and we'll see if it makes a difference.

http://chipreviews.com/main-feature/main-news/frostbite-2s-limit-6-core-performance-in-battlefield-3/

http://chipreviews.com/main-feature/main-news/battlefield-3-revisited/

 

[Kitlope]

Ok I just did my own little experiment regarding this Hyperthreading on or off. Keep in mind this only affects Intel CPU's.

I played on a 64 player server running Dawnbreaker map. Nothing was changed, just disabled HT in the bios. The findings are incredible. Keep in mind I have a Intel 6 core with a mild overclock from 3.33 Ghz to 3.72. Because it's still a good processor and mostly because its a 6 core, I have tons of headroom regarding CPU usage. Let me say though I practically doubled the CPU usage with HT disabled. It didn't really appear to affect my overall framerate much with HT disabled vs. it being enabled and I'm thinking it's the simple fact I have decent headroom, either way. From what I've read on the net games don't really use HT but this appears to be changing, as the links in my above post show too.

My thinking is anybody with a typical Intel i5 non HT 4 core will max their CPU out a lot quicker than if they had HT. Only the i7's have hyperthreading. I just did some checking with the second pc, it runs a i5 3470 (typical ivy bridge, 3.2 Ghz, 6 MB cache) and it runs decent enough but the cores are all maxed out, all the time. I would love to throw in a i7 with HT and see the difference, but at $360.00 for a 3770 am not gonna do it. Can't say much for AMD as they don't use hyperthreading but do have more cores (6 & 8 core are common, unlike Intel), albeit at a bit of slower speeds.

Make of it what you will. Look at the top right CPU - Total graph. I don't claim to be an expert but regardless of anything wouldn't keeping the overall CPU usage as low as possible be a good thing?


Hyperthreading On:





Hyperthreading Off: