Question if you upgraded to a quad core for Crysis

[bigriggg]

so.........

 

[Swamp]

How much of an increase did you see in minimum framerate in laggy areas from a dual core after upgrading to a quad core?

Reason is I think I'm severely CPU Limited. I get 25 FPS or lower in 1680x1050 and if I drop it down to 800 x 600, its the same exact framerate although the CPU meter on my g15 says 92% or so.


If you OC'd you cpu more. To like 3.2-3.4ghz you will see a slight increase in fps. You wont see such a dramatic drop in fps when gaming. But if you are going from ur E6550 to a quad core. You will see a small increase of fps, depending on if your gpu is bottle necking ur cpu or cpu bottle necking ur gpu.

Even with my E6550 oc'd to 3.4ghz with my ATI 4890. my fps still jumps all over the place. In l4d2 max settings, 1680x1050 res my fps drops from 60fps to 200fps, depending on what is happening in the game. A quad core should even the fps out.

 

[error8]

so.........

You are getting higher CPU usage then me but it's still at 45%, which means that probably the fully loaded 2 cores would be spread out on the other two, so, doesn't really look like quad core optimization to me.
Look at the pictures I've posted in Dragon Age, with 80% usage. That is a game that likes cpu cores. If you would have gotten over 50%, that meant Crysis would have needed more then 2 cores. But it looks like it doesn't.

 

[Swamp]

My thing is, why does Cervat Yerli the CEO and President of Crytek said himself crysis runs so much better on a quad core over dual core? It prolly uses 2 cores for the game and puts any other load from the game onto the other cores. maybe like AI and physics?



So from looking at Bigrigg's screen shots. Must say something.

 

[Indus]

spend the time and drop your CPU clock back to stock and see what kind of difference you get

Not a bad idea. Will do when I'm home later tonight and post results.

Ok I did this last night and got some very interesting results about CPU bottleneck vs GPU bottleneck.

If I use 4x AA then there is absolutely no GPU bottleneck and the framerate stays the same at 800 x 600 to 1680 x 1050. However if I overclock the CPU from 2.33 stock to 2.93 I gain 23% more framerate.

If I use 8xQ AA then something very weird happens, I actually get a DECREASE in performance.. I got 5% more framerate with stock speed than my overclocked speed. (I did this 5 times with same results to make sure). And as for the GPU bottleneck yes I noticed a 11% decrease in performance if I go from 800 x 600 to 1680 x 1050.

 

[bigriggg]

You are getting higher CPU usage then me but it's still at 45%, which means that probably the fully loaded 2 cores would be spread out on the other two, so, doesn't really look like quad core optimization to me.
Look at the pictures I've posted in Dragon Age, with 80% usage. That is a game that likes cpu cores. If you would have gotten over 50%, that meant Crysis would have needed more then 2 cores. But it looks like it doesn't.

if you look at my pic, it shows about 25%, 25%, 50%, 75%, so if what your saying is if i had only 2 cores,one would be at about 100%, and the other would be at about 75%, then you would see a plateau, which means you would need 4 cores to run this for optimal performance, which i do run at max settings and in 64 bit mode.

 

[MrK6]

Later today I'll see if I have time to run some benchmarks on my 750 with two cores disabled vs. all four cores enabled and see if there's a significant performance delta.

 

[Indus]

So with the increase I got after my test of overclocking, will a quad give me better framerates @ 3.4 ghz too or no? Am I looking at $170 for 20-30% increase?

 

[Cogman]

if you look at my pic, it shows about 25%, 25%, 50%, 75%, so if what your saying is if i had only 2 cores,one would be at about 100%, and the other would be at about 75%, then you would see a plateau, which means you would need 4 cores to run this for optimal performance, which i do run at max settings and in 64 bit mode.

Wrongo.

The reasons you see 25, 25, 50, 75 is because windows is graciously switching which processor core each of your threads is running on. As a result, each core sees just enough of a spike to register 25% in the windows manager.

Until you receive over 75% on each core, the game probably isn't using 4 threads.

 

[bigriggg]

Wrongo.

The reasons you see 25, 25, 50, 75 is because windows is graciously switching which processor core each of your threads is running on. As a result, each core sees just enough of a spike to register 25% in the windows manager.

Until you receive over 75% on each core, the game probably isn't using 4 threads.

ok boss hog, sorry you're right, sometimes i just see things, made it's just my imagination, i did forget to take my pill today

 

[MrK6]

So with the increase I got after my test of overclocking, will a quad give me better framerates @ 3.4 ghz too or no? Am I looking at $170 for 20-30% increase?

A quad @ 3.4GHz compared to your E6550 @ 2.93GHz? Yes, most definitely. Compared to your E6550 @ 3.4GHz? Not significantly.

 

[Indus]

A quad @ 3.4GHz compared to your E6550 @ 2.93GHz? Yes, most definitely. Compared to your E6550 @ 3.4GHz? Not significantly.

Yes I wanted to go from E6550 @ 2.93 (600 mhz overclock) to Q9550 2.83 stock @ 3.4 or more. My memory can do 419mhz easily so depending on the CPU 3.5 or 3.6 might be possible.