Blow It All Away Reza Hussain's tech blog

28Apr/100

Overclocking & Left 4 Dead 2

Here is my setup:
Athlon II x2 240 cpu
Dfi 785g-m35 mobo
4gb 1600 8-8-8-24 1.65v crucial ddr3 ballistix tracers
corsair hx520 psu
ati 4670 gpu

What I did was record a timedemo in L4D2 and then ran the timedemo on my system at stocked and overclocked speeds.

Stock Setup:
200fsb
Ram @1333
2.8ghz cpu frequency

Overclocked Setup
300fsb
Ram @ 1600
3.75ghz cpu frequency

The first test was using the settings in l4d2 that I normally play with.

Results:

Stock 69.57 fps & 10.360 fps swing
Oc'ed 73.22 fps & 11.54 fps swing

At first I thought maybe there really isn't any gain from overclocking. Then I decided to check if my gpu was the bottleneck. For the second test I turned all of the graphics in l4d2 down as far as they would go.

Results:

Stock 93.91fps & 8.803 fps swing
Oc'ed 118.62fps & 11.324 fps swing

That is a 24.71 increase in fps on the oc'ed setup when I lowered the gpu settings, indicating that I am gpu bottlenecked. It is roughly a 34% overclock giving a 27% boost in performance.

Other thoughts:
The max memory speed this mobo supports is 1600. In reviews of the crucial ddr3 ballistix 1600, it is demonstrated that they can run at 2100 9-9-9-24.

See here and here

So I should have gotten a mobo that could support higher memory frequencies, at least above 2100. Now, running higher frequencies at relaxed timings on ram doesn't necessarily mean more speed. It's benefit comes from being able to run a higher fsb when your memory divider is already maxed.

I tried running the DFI board with higher than 1600 memory speeds, and it was just not stable.

-Rez

   

Archives

Tags

atx case bang for buck bang for buck cpu bang for buck gpu case design choosing pc components computer case crucial ddr3 custom case design defcon dfi 785g google nofollow informedselection.org iphone codesigning iphone game led grow lamp led lamp left 4 dead 2 overclocking phonophrenic spam

Blogroll

Meta

SEO Powered by Platinum SEO from Techblissonline