Any point to upgrading from 4Ghz Core 2 for games?

Big.Wayne · 23 Mar 2010 at 23:47

Devrij said:

Hi Devrij, thanks for posting results up also!

. . . . your HD5770 GPU seems to be a lot more loaded than Cesar67's HD5850? . . . hmmm can you crank up the eye-candy anymore, lash on some AA/AF or whatever but give the GPU something more to do . . . It's hard to be sure as your image fell to pieces when enlarged but I'd say that was about 95% GPU load? . . . so "Full HD" 1920x1080, all graphic options high and try a few different blends of AA/AF see what happens!

I'd say both yours and Cesar67's processors were working hard but I'm still not convinced they are capping the GPU's? . . . hopefully we can find out soon! :cool:

///Rage · 24 Mar 2010 at 00:10

I think it would not make sense now to buy a dual core CPU now when developers are finally coming round to multithreading.

Th3D0n · 24 Mar 2010 at 00:30

///Rage said:
I think it would not make sense now to buy a dual core CPU now when developers are finally coming round to multithreading.

Big.Wayne · 24 Mar 2010 at 00:47

Hey ///Rage, actually we are coming at this from the other angle . . . i.e fast dual core owners sussing out if their chips can still do the do!

Cesar67 · 24 Mar 2010 at 01:02

Tomorrow mate.

I think that second run, I bumped everything up to 'High' (Maximum in game) Maybe Bloom or HBAO was off. Will run more tomorrow.

The first screenshot was medium.

Don't you think 95% CPU usage is a sign that more resources are needed to maximise performance ?

Did i mention it was unstable ?

Did you know this game has a full core running at 100% just sitting idle at the log in screen ? There is a screenshot in one of these sub forums about that.

Vir · 24 Mar 2010 at 09:36

Go Quad if you play Bad Company 2 a lot. It runs so much better on my i7 920 (stock) then on my older E8400 @ 4Ghz

Cesar67 · 24 Mar 2010 at 19:51

E8400 @ 4.2GHz - HD5850 @ 1000/1200

Game settings:

[WindowSettings]
Width=1920
Height=1080
Fullscreen=true
RefreshRate=60
VSync=false
[Sound]
Quality=high
VoipEnable=true
SpeakerCount=0
[Graphics]
Effects=high
Soldiers=high
Vehicles=high
Overgrowth=high
Undergrowth=high
StaticObjects=high
Terrain=high
Shadows=high
Bloom=true
HSAO=true
MSAA=3
Water=high
MainQuality=custom
Texture=high
DxVersion=11
Aniso=4
Detail=high
RenderAheadLimit=0
Fov=68

Big.Wayne · 24 Mar 2010 at 20:47

Cesar67 said:

Looks good Cesar67

. . . I think maybe you could give the GPU a bit more work to do? . . . did you try x8/x16 Anisotropic-Filtering?

keyser · 24 Mar 2010 at 20:47

wake me up when theres decent benchmarks for dual vs quad. (i.e. 1920x1200)

Cesar67 · 24 Mar 2010 at 21:26

Big.Wayne said:
Looks good Cesar67 . . . I think maybe you could give the GPU a bit more work to do? . . . did you try x8/x16 Anisotropic-Filtering?

As far as i know, That is every setting in the game at their highest (Yes 8XAA & X16AF), I only have a 1920X1080 monitor so cant raise the resolution.

Sure the CPU is not flatlining but, Surely those times when it peaks over 95% are the times in game when i have jerkiness and lag ?

Keyser don't you have anything worthwhile to put into the discussion ?

Big.Wayne · 24 Mar 2010 at 21:37

Hey Cesar67,

I'm finding all this a bit confusing!

. . . can you confirm what AA/AF settings you are actually using as in the [config] you posted above it says MSAA=3, Aniso=4

Regarding the CPU holding you back? . . . surely there is absolutely no difference from a CPU working at 33% to a CPU working at 93% :confused:

If it's not maxed @ 100% then there is still untapped power there?

CAT-THE-FIFTH · 24 Mar 2010 at 21:45

This article did look at core scaling:

http://www.techspot.com/article/255-battlefield-bad-company2-performance/page7.html

Cesar67 · 24 Mar 2010 at 21:57

Big.Wayne said:
Hey Cesar67,

I'm finding all this a bit confusing! . . . can you confirm what AA/AF settings you are actually using as in the [config] you posted above it says MSAA=3, Aniso=4

Regarding the CPU holding you back? . . . surely there is absolutely no difference from a CPU working at 33% to a CPU working at 93%

If it's not maxed @ 100% then there is still untapped power there?

That is confusing mate, Ingame it says 8XAA & 16XAF, The .ini file uses the 3 & 4 values to represent 8AA & 16AF though.

Well, I am not any kind of expert at all though, Does the system not keep that ~7% free for other uses ? Would the two cores actually go completely to 100% ? Is it not managing it's resources by leaving a small percent free ?

Do you have any ideas why the game feels so unresponsive and laggy even when the GPU is only running at 50% ? What else could be bottlenecking my system ?

So many people have commented Dual>Quad 'feels' so much smoother. The nature of gaming will always see peaks and troughs as mentioned previously, I am *assuming* that when these peaks come around, A quad can cope better.

I do see your point about the untapped potential still left, When i put together my quad system (Next week hopefully), I will compare like for like benchmarks.

Cesar67 · 24 Mar 2010 at 22:02

CAT-THE-FIFTH said:
This article did look at core scaling:

http://www.techspot.com/article/255-battlefield-bad-company2-performance/page7.html

I have read that and, His conclusion about the dual core does not seem to be true comparing with my experience with the game.

Don't get the wrong impression, It is very playable on the dual core, I just believe it can be smoother and more responsive. After all we always strive for the best we can achieve, I don't think this dual core system is cutting it for this game.

Again, I am far from an expert

Cesar67 · 24 Mar 2010 at 22:10

He actually updated that review:

Steve @ Techspot said:
Okay for those that were concerned about the dual vs. quad battle here is a little more info before I get some sleep.

The game does appear to use all four cores when available. Here we used a standard Core i7 920 processor at 2.66GHz for this test, please note HyperThreading has been disabled and a single Radeon HD 5850 graphics card was used. As you can see none of the cores are working very hard.

Here is the same Core i7 920 processor with two cores disabled as well as HyperThreading. As you can see neither core is maxed out but the CPU utilization is much higher.

So again in conclusion a decent dual core processor such as a Core 2 Duo E8xxx or Phenom II X2 should be enough to get the most out of your graphics card in this game. While it is quad-core optimized the game is not demanding enough on the CPU to warrant it based on what I have seen so far. Still quad-core processors are ideal but not entirely necessary.

Finally the Radeon HD 5850 produced the same average frame rate with 2 cores enabled as it did with all 4. I admit more testing needs to be done with real actual dual-core processors and I promise I will do my best to collect more data for you guys. Also thanks for all the feedback so far its great!

Devrij · 25 Mar 2010 at 12:20

Cesar67 said:
How does the game 'feel' to play mate ?

Do you believe a quad in your system would run the game better/smoother/faster ?

Taking into account your GPU is also pushed to its limits.

I notice the odd drop in framerate when there's a lot of effects going on, and I will admit it occasionally feels laggy, but overall it's fine with frames above 60fps the vast majority of the time. It doesn't feel smooth though, which niggles at me a bit. I'll post up a bigger pic with some higher AA/AF settings on there, but I think it'll just result in lower frames and the same workload.

Devrij · 25 Mar 2010 at 12:51

Results: same rig in spec, all settings high, HBAO on 4xAA 4xAF. Frames in mid40s dipping to low 20s. GPU is maxed out as you can see, and cpu seems to be almost maxed out, so here I'm pretty much entirely gpu bound, but whether my cpu is also a limiting factor is inconclusive due to my lack of gpu grunt.

Bossk128 · 25 Mar 2010 at 13:48

Traditionally is it not thought that high resolutions (especially) and high AA/AF require a faster CPU, and that all other graphical stuff (textures/lighting effects not to mention physics, plus again a bit of AA/AF) is GPU based?

E.G. reviewers when benching GPUs for raw speed do it at 640x480 so the CPU doesn't bottleneck the GPU. Of course modern GPUs will have FPSes in the hundreds, and nobody games at 640, so benchies have moved on.....

Edit: After reading the comments below, and thinking about it the above is bonkers - it's CPU benchies at 640! Ooops.

Angst · 25 Mar 2010 at 13:57

phill0i said:
@lanz i did a 10 min fraps benchmark with my [email protected] and it returned and average fps of 42 a min of 17 and max of 65, i then upgraded to a phenom2 x4 @3.6ghz on a am2+ board with ddr2 ram and repeated the benchmark ad it returned a average of 76 a min of 33 and a max of 119.
I ran both benchmarks on identical quality settings, on port valdez(forgot how to spell it), on the attacking team. i also used the same class and i did not use vehicles.
Also the fact that it is an am2+ board using the same ram as on the e7300 run will remove that as a likely factor for a performance increase.
.

Just taken this from the other quad core thread. its worthy to note that with the dual the gpu used around 40-60% and the cpu was nearly flatllining but with the quad the cpu usage was about 70-80% and gpu usage is about 100%. so if we continue to see a trend of cpu intensive games like we are at the moment it should be easy to conclude that it would be worth the upgrade.

Devrij · 25 Mar 2010 at 15:17

Bossk128 said:
Traditionally is it not thought that high resolutions (especially) and high AA/AF require a faster CPU, and that all other graphical stuff (textures/lighting effects not to mention physics, plus again a bit of AA/AF) is GPU based?

E.G. reviewers when benching GPUs for raw speed do it at 640x480 so the CPU doesn't bottleneck the GPU. Of course modern GPUs will have FPSes in the hundreds, and nobody games at 640, so benchies have moved on.....

I was under the impression it was the other way round? That running low res allows you to see how bottlenecked you are by your cpu, seeing as your gpu now has far less rendering work to do? In fact, that might be worth trying.