How CPU bound are games today?

[DildePer]

I've currently on my E2180 @ 3375 Mhz. SiSoft Sandra benchmarks tell me that my chip is outperforming an E6750 and just slighty an E8400 in some benchmarks.

Would I see any FPS gain at all, if I upgraded to say... a Q6600 or E8400 and OC'ed them?

Or are games today simply more GPU bound?

[inCore]

Unless you're gaming at low resolutions, such as 800x600 and 1024x768, you're GPU bound. You will only see a significant fps gain in those resolutions. Sure, there are some games that will profit more than others, but you're usually a lot better off upgrading your graphics card. Although, your 8800GTX is fine for now.

For some graphs: http://www.tomshardware.com/charts/desktop-cpu-charts-q3-2008/benchmarks,31.html

[nefariouscaine]

I'm inclined to say a quick yes - most games have been shown to benefit from the increase level of cached offered - your 1mb vs 6mb in the case of the e8400 - generally its better to have more to a degree - i'm going to paraphrase a review of your processor from xbit labs "We have already mentioned earlier in this article that reduced to 1MB L2 cache is a significant slowdown for the Pentium E2000 series in gaming applications"

its not purely a thing of clock for clock as you must understand - IMO I say get a e8400 or e8500 and push them up some more - you should get a healthy little boost for much less $$ than your video card upgrade would cost

[stevecs]

Depends on the game. For example I play the X series (x3 reunion, x3:terran conflict) among others. Graphics no problem w/ a 280GTX even at 2560x1600. CPU is a killer as it's a single threaded app like most games. 3ghz QX6850 slows down to 10-15fps. at 4ghz it becomes 'playable' at about 20fps or so (though not for dog fighting). This is true (and exact same figures) even at 1280x800. So it really depends on the type of game.

[Level 88]

Yeah, I'd hang onto that E and let it pay for itself a bit longer (i.e save for when nehalem becomes more palatable to the back pocket), those aren't bad clocks and you're not going to see much gain unless you're a SupCom freak, for instance. There are always ways of reducing the impact of the CPU by dialling down physics and other eye candy if it's a problem.

How's your gaming experience? If you're happy with it then there's not much point in spending cash for a few FPS.

[RazzleUltra]

I find RTSs seem to bog down for me when there's lots of stuff going on. Given that I've got a 4870 512MB and I'm playing at 1920x1200 with no AA, I'm presuming it's CPU bound (3GHz E6700)

[YukonTrooper]

Depends on the game. Don't listen to the "GPU only matters" talk. Try playing most RTS games, or Source based games on a weak processor.

[G.Foyle]

CPU cache size influence in low resolutions:
http://img.pclab.pl/zdjecia/artykuly/mbrzostek/e5200_e7300/ut3_cache.png
http://img.pclab.pl/zdjecia/artykuly/mbrzostek/e5200_e7300/crysis_cache.png
All CPUs are clocked the same, memory clocks and timings are the same.
In bigger resolutions the difference diminishes; in 1680x1050 there's practically no difference between E5x00, E7x00 and E8x00 (clock for clock).

[Ozzfest05]

The real answer is yes I gained over 20 fps from overclocking my cpu I use FRAPS and in game FPS counters the cpu is the limiting factor with todays high end GPU's its true so many times if your looking at getting a really highend GPU like 4870x2 or something similar and you have say a 2.66 cpu you would be better off getting something like a 4870 on sale and grab a e8400 e8500 or e8600 and oc it a little like the new 8400's with EO will do 3.6 3.8 on low voltage with little to no effort or experience in OC'ing

[YukonTrooper]

Good point about the cache size being a factor, Xoqolatl. When I upgraded to C2D, I went with the E6420 just for that reason.

[inCore]

Sure cache will make a difference, I'm just inclined to saying that going from 141 to 182 fps in Unreal Tournament is probably one of your smaller concerns. That's also at 2.66 GHz, your E2180 can obviously do better than that.

[Xello]

As has been said probably depends on the game and general system setup. This might not affect you but in some high-end setups such as tri-sli and i believe crossfire-x, extra cpu power can help even up to 1920x1200. Check out this Hexus article for a comparison of 3.2ghz nehalem vs 3.2ghz penryn on a tri-sli system:

http://www.hexus.net/content/item.php?item=16285&page=1

@ 1920x1200, there are some tangible differences between the two, which i was pleasantly surprised at :)

[demonkevy666]

CPU cache size influence in low resolutions:
http://img.pclab.pl/zdjecia/artykuly/mbrzostek/e5200_e7300/ut3_cache.png
http://img.pclab.pl/zdjecia/artykuly/mbrzostek/e5200_e7300/crysis_cache.png
All CPUs are clocked the same, memory clocks and timings are the same.
In bigger resolutions the difference diminishes; in 1680x1050 there's practically no difference between E5x00, E7x00 and E8x00 (clock for clock).

no conroe ?

I never seen a fps vs with conroe 2mb vs 4mb

this is nice to know too.

[Asgard_thor]

:P you 2180 aint better then MY e6750 *sticks tongue out*

[G.Foyle]

Cache size has much smaller impact on performance than FSB frequency. Stock E5200 is FAR slower than similarly clocked E7300; setting FSB to 333 makes it only a bit slower.