• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

CPU bottlenecking GPU

Associate
Joined
1 Dec 2009
Posts
340
Hey all,
Current system is a 920 @ 4Ghz with 2*4890s crossfire (16x 16x PCIe motherboard)

Still not enough to run Crysis in full at 1920*1200 with 2x AA playable but sure that Cysis was only made to make us all feel inadequate + want to upgrade ;-)

Anyway...whatever game I try, Crysis, Dragon Age, NFSS, etc. looking at the CPU utilisation on my G15 keyboard screen it doesn't really go past 15%.

Is this a good enough rough indication of how much the CPU is working? Thinking about it now I should run task manager on my second monitor to check, but overall upgrading from a Q6600 @ 3.1Ghz to 920 @ 4Ghz doesn't seem to make any difference for games that I can tell. Seems at least for games it's better to just spend a huge wad on the graphics card(s). Maybe I should have spent the £1000 on 2 5970s rather than upgrade the whole lot (had 2 * 8800GTs SLI before)

Have always read people's response to performance queries saying you need a faster CPU or you will be 'bottlenecking' your GPU but from my own findings it seems it's best to stick with a cheap half decent CPU/motherboard and spend the upgrade money on the GPU.
I guess I can still upgrade to 2*5970s though lol !
 
If he's upgraded from a Q6600 to an i7, he won't have the same motherboard - and the new i7 motherboards are both Crossfire and SLI compatable.

Even so, I'd have thought (looking at CF'd 4890 reviews) that you should be able to max out Crysis and still get decent frames - even at 1920x1200.

crysis_1920_1200.gif


That's using the same system as you (i7 920, 2 4890s), but the CPU is only clocked to 3.8GHz. The settings there are maxed out, and is run with 4x AA.

Full review here

I don't have any idea why you aren't seeing the results, unless it isn't running in Crossfire - check in the Catalyst suite.

I will run benchmarks again but what I remember was getting around 44fps with settings = High with 4* AA, but down to 30ish with settings = Very High and 0* AA. It was indeed this same article that made me decide on going for 4890 crossfire rather than Nvidia.
 
Is it not immediately obvious that a cpu that is barely loaded is in fact faster than you need?

No motherboard that can run two 8800gt's with a q6600 can run two ati cards, your alternative upgrade route would have required a new motherboard as well.

Did you resolve the problem of the two 4890s not fitting on your motherboard then?
Yes, I returned the 2 OC value 4890s and Overclockers promptly refunded me :D
The 2 XFX ones fit fine but I then had to spend another £110 on a PCIe XFi soundcard to replace my PCI one :-|
 
Overall I had this feeling I needed to upgrade (retail therapy!). I play WoW mostly, except for Dragon Age: Origins in-between.
My old setup was fine for these 2 and the odd spot of NFSS, Left 4 Dead, TF2 etc.

I initially decided that to upgrade my system it should be graphics card.
I then decided I should really upgrade the CPU/motherboard to something faster as to NOT bottleneck the new graphics card.

I even changed my monitor from a 20" to a 24" just to ensure I was making good use of the upgrade :confused:
Now though I just have this feeling that spending half what I did on just a graphics card upgrade would have given me more. Looking at these sort of benchmarks is what I mean...
Crysis%208xAA.png
 
My results are not quite as expected comparing to the graph, and I do have to reduce some settings such as shadows to make parts of the game playable.

Here is what I got just now (windows 7 btw)...

26/12/2009 23:54:18 - Vista 64
Beginning Run #1 on Map-island, Demo-benchmark_gpu
DX10 1900x1200, AA=8x, Vsync=Disabled, 32 bit test, FullScreen
Demo Loops=3, Time Of Day= 9
Global Game Quality: VeryHigh
==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 66.41s, Average FPS: 30.11
Min FPS: 17.91 at frame 144, Max FPS: 38.00 at frame 855
Average Tri/Sec: -9900692, Tri/Frame: -328775
Recorded/Played Tris ratio: -2.79
!TimeDemo Run 1 Finished.
Play Time: 61.08s, Average FPS: 32.75
Min FPS: 17.91 at frame 144, Max FPS: 38.95 at frame 861
Average Tri/Sec: -10314350, Tri/Frame: -314976
Recorded/Played Tris ratio: -2.91
!TimeDemo Run 2 Finished.
Play Time: 61.20s, Average FPS: 32.68
Min FPS: 17.91 at frame 144, Max FPS: 38.95 at frame 861
Average Tri/Sec: -10225838, Tri/Frame: -312901
Recorded/Played Tris ratio: -2.93
TimeDemo Play Ended, (3 Runs Performed)
==============================================================

Completed All Tests

<><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

26/12/2009 23:54:18 - Vista 64

Run #1- DX10 1900x1200 AA=8x, 32 bit test, Quality: VeryHigh ~~ Overall Average FPS: 32.715
 
Last edited:
Back
Top Bottom