"Processor speed doesn't matter for gaming"

[xchrissypoox]

So I'm procrastinating a project and someone said something like "processor speed doesn't matter, which I tend to agree with, so I did a couple of benchmarks in crysis at various processor speeds.

 

Test methodology: Okay so I ran HOC crysis benchmark at 3 different speeds on all very high settings (no filtering) at 1440x900 (my monitor's max resolution till I can get me that 24" ultrasharp) and I would look at the average fps of the first two runs (I couldn't get it to put benchmark results into the folder so this is all I could figure out since I wanted this done by 10PM).

 

Here are the results.

27.31fps 2-cores HT disabled at 1446Mhz (125 bclk, 12 multiplier) turbo mode off

33.74fps 2-cores HT disabled at 3.4Ghz (170 bclk) 3.57Ghz turbo mode on

34.34fps 4-cores HT disabled at 4009Mhz (200 bclk) turbo mode off

(note: awesome government graph maker)

 

I've been using my comp at 1446Mhz since I about a couple hours ago, and you wouldn't even know it was going this 'slow' then again everything else on it is fairly fast. I think that the fact that at 1446Mhz it is still within (almost) 7 seconds is pretty good, and I'd say as far as crysis is concerned get a dual core (do they even make ones slower that 2.5Ghz now?) and a good video card and play to your hearts content... though crysis kinda sucks.

 

So I think it is safe to say that this could play most any game at 1446Mhz that isn't totally CPU bound. I didn't really prove anything, and this is pretty narrow, but maybe I've intrigued someone who will do more with the subject matter. Feel free to add on to this, and sorry I don't have much to say, I'm not much of a articler.

 

 

Oblivion tests on page 3

Edited September 30, 2009 by xchrissypoox

[molotovdaskwerl]

so it's not a big difference, it like makes a difference, but it's not a big cept from 1.4ghz on a dual core to a 4ghz at a quady

[xchrissypoox]

so it's not a big difference, it like makes a difference, but it's not a big cept from 1.4ghz on a dual core to a 4ghz at a quady

I'd say its not much of a difference in crysis no matter what speed the processor is going (since I wouldn't cry over 5 frames to only buy a 2Ghz dual core instead of a 4Ghz quad and a couple of visual tweaks would have you over 30 fps) but I think it might show a larger difference if the graphics card is weaker, but I don't have another graphics card. I'll post some other games that are more cpu dependent like oblivion and half-life later. My brief experiment isn't enough to convince me that all gaming computers should run at 2Ghz though it just says a couple things about crysis and processor speed with a top of the line graphics card.

 

My findings weren't ground breaking but I've never seen it displayed to extremes so I thought I'd put it out there since I see lots of people talking about how they want a 4Ghz overclock and I'm just thinking "why?" which seems hypocritical (look at sig) but for normal use I just run at 3.4Ghz with 2 cores (the middle one on the chart) and when I want to impress people I turn everything on.

Edited September 26, 2009 by xchrissypoox

[rourkchris]

4Ghz is a glass ceiling that's been hard to break. It's the challenge. 3Ghz was the same way not too long ago. Interesting findings. I'd like to see more.

[xchrissypoox]

4Ghz is a glass ceiling that's been hard to break. It's the challenge. 3Ghz was the same way not too long ago. Interesting findings. I'd like to see more.

Yeah I guess that could be the reason. But it seems pretty easy to hit it on the i7 (maybe I just have a good processor 1.21vcore bclk200x20) 4.2Ghz has been the challenge for me since I have inadequate cooling to hit it with all 4 cores on right now

 

Feel free to contribute some info on your games, all i have is oblivion, wow, crysis, and the orange box.

 

Edit: I'm about to add the results of the video card on high at 1446Mhz to make sure that increases the fps as it should

Edited September 26, 2009 by xchrissypoox

[ccokeman]

As long as you are not GPU limited you will see an improvement. At lower resolutions you will see that improvement with less than mid range settings

[slick2500]

I was always under the impression that Crysis was gpu bound? Do that same test with a Source engine game and see what happens.

 

I see what your trying to prove but different game engines are limited by different things.

[xchrissypoox]

As long as you are not GPU limited you will see an improvement. At lower resolutions you will see that improvement with less than mid range settings

Yeah but the whole fear of slow CPUs is that they bottleneck the computer and will make you get lower fps, I know I've seen people so not to get below X.XX Ghz for gaming (example 2.5Ghz sounds low for gaming to anyone reading this right?) and so I wanted to see how much you would have to lose (visually) to get 35 fps average. I'm testing all very high, except shadow quality, and shader quality which are on high. It is just so that people won't think of the 1446Mhz cpu as failing to meet the greater than 30fps standard.

 

Edit: as I said all settings to very high, except shadow quality, and shader quality which are on high the a dual core i7 at 1446Mhz did 34.73fps which is pretty good

 

Beginning Oblivion testing.

Edited September 26, 2009 by xchrissypoox

[xchrissypoox]

I was always under the impression that Crysis was gpu bound? Do that same test with a Source engine game and see what happens.

 

I see what your trying to prove but different game engines are limited by different things.

 

Yeah I know it was just the most accessible at the time, I'll do half-life tomorrow.

[Flibo]

Yeah I know it was just the most accessible at the time, I'll do half-life tomorrow.

 

I've done some experiments, and I can say for quite sure processor speed really matters in game running HL and HL2 engines. Especially in the original Half-Life and its mods you barely need a graphics card so it's great for testing the effect of increasing processor speed. I'd like to see some kind of a graph indicating the effect

 

EDIT: Actually, testing in original Half-Life isn't a very good idea since there isn't a stress test for it as far as I know and there's a fps limit (1000 fps).

Edited September 26, 2009 by Flibo

[WhenKittensATK]

You might see better test results on a higher resolution.

 

Games to try:

HL2

Supreme Commander

Far Cry 2

 

There are already Supreme Commander results somewhere on this forum. I can't remember the exact results, but I believed it was in favor the a better processor, better frames.

Edited September 26, 2009 by Krazyxazn

[tkrow21]

Isn't Crysis being bottlenecked by the GPU?