CPU scaling with GeForce 295 GTX

[MUYA]

The article;
http://www.legionhardware.com/document.php?id=807

I certainly found this true with my q6600 at least in terms of 3D mark2k6

With the q6600@3Ghz I obtained a score of
http://service.futuremark.com/compare?3dm06=9788422

With q6600@3.4Ghz I obtained a score of 18473
http://service.futuremark.com/compare?3dm06=9875055


Might plonk in some dollars in a q9650 and o/c that to 4GHz or more?

[methimpikehoses]

I have one for sale! :o

[ATi]

I have one for sale! :o

why?!:o

[methimpikehoses]

why?!:o

My mobo can't hold the OC. If I'm going to buy a CPU and new mobo, I might as well go all the way with the new socket. :naughty:

[Bman212121]

I read the article earlier today. It all comes back to resolution. At 1680 you'll see some gains with a faster cpu, at 1920 those gains will be less pronounced. At 2560 they will almost disappear in most cases.

I really don't feel that a Q9650 is a cost effective upgrade over a Q6600. It costs about $350 to gain another 5 - 15% increase in fps. The only way you're really going to see larger gains is going to an i7 920 and replacing your board.

A Q9650 from 3ghz to 4GHZ in WiC is 50fps to 58 fps. A gain of 16%. Far Cry 2 is 75 to 84 fps. A gain of 12%. CoH 113 to 122. 8% gain. Fallout 3 65 to 70, 8% gain.

WiC shows the best scaling percentage wise but it still won't play a huge role IMO. The rest of them are all above 60fps anyway so those gains aren't going to be nearly noticeable.

[MUYA]

Hmm...i7 looks more appealing now...seems the 295 will benefit most by jumping it i7's!

[methimpikehoses]

I read the article earlier today. It all comes back to resolution. At 1680 you'll see some gains with a faster cpu, at 1920 those gains will be less pronounced. At 2560 they will almost disappear in most cases.

I really don't feel that a Q9650 is a cost effective upgrade over a Q6600. It costs about $350 to gain another 5 - 10% increase in fps. The only way you're really going to see larger gains is going to an i7 920 and replacing your board.

STFU Bman!

[Bman212121]

STFU Bman!

let me finish!!! Please, can I finish??!!?!

If you didn't already have a Q6600 I think that 9650 would be a good upgrade. ;)

[methimpikehoses]

let me finish!!! Please, can I finish??!!?!

If you didn't already have a Q6600 I think that 9650 would be a good upgrade. ;)

Actually if you can get a steady OC and have SLI, the difference is very noticeable.

But if your mobo can't hold voltage, it's not worth it.

[Bman212121]

Actually if you can get a steady OC and have SLI, the difference is very noticeable.

But if your mobo can't hold voltage, it's not worth it.

SLI what? SLI GTX 295's?

EDIT: I'm waiting.........

[methimpikehoses]

SLI what? SLI GTX 295's?

EDIT: I'm waiting.........

1 GTX 295

/waiting

[Bman212121]

Hmm...i7 looks more appealing now...seems the 295 will benefit most by jumping it i7's!

Actually here is another good question. Is there a particular title where it feels like the game is running slow?

One thing that would be note worthy is there are no minimum fps numbers shown. The cpu could potentially have a larger impact if you were being cpu limited and that was causing the slowdowns. The only game I can think of where I've seen cpu slowdowns is Far Cry 2. When the fire is on the screen during the benchmark I can see the SLI scaling shrink on the bars to very little while the fps drops. There are other times where I've seen the bars shrink, but usually the fps doesn't drop meaning that I'm already sitting on a frame cap or something else isn't scaling right.

Actually I can think of another time where that happens, if you play an MMO and there is a bloat load of characters on the screen.

[Bman212121]

1 GTX 295

/waiting

Is there a particular title where it is really noticeable? I'm trying to think of how CPU would affect fps more than the GPU would.


I need to paint a picture to try to explain what I want to say... Off to paint .net.

[methimpikehoses]

Is there a particular title where it is really noticeable? I'm trying to think of how CPU would affect fps more than the GPU would.

I need to paint a picture to try to explain what I want to say... Off to paint .net.

Well since I ran both chips on my rig within minutes of each other, I can say it was noticeable in Crysis and the benchmark Vantage.

[Bman212121]

Wow I'm so pissed right now... It's not the first time this has happened either. There is a certain key combo that causes you to go back, maybe ctrl + b or something. When trying to capitalize a letter I accidentally hit that combo causing all of my post to be lost. I usually start typing up long posts into notepad because of this, but I didn't do it and naturally it happens again. EDIT: I know exactly what it is now. It's Crtl + R causing the page to reload!!! :fu:

Yea Crysis is a weird game when it comes to benchmarks. Vantage obviously will show differences because it has tests specifically for the CPU.

My main point I was trying to make was when you're already GPU bound, increasing your clock speed shouldn't make a difference. If you're cpu bound then increasing the CPU will help with fps, only to the point where you're GPU bound again. So if there is place where you have low fps because you are GPU bound, increasing the cpu isn't going to do anything for that despite the average fps going up because of other parts where you might be cpu bound.

A good example of this is in WiC where the A-bomb drops, that's all GPU-bound and increasing the CPU from 2Ghz to 4Ghz won't change the fps one bit. There is another part in that test where the planes make a fly by. I might only hit 60fps there where an i7 could hit 100+ fps. The averages will go up with the faster cpu, but overall gameplay really won't change for the better because I'll still easily notice the lag at the A bomb, but when I'm already getting a smooth 60fps as the planes fly over 100 fps isn't going to make them look a lot smoother.

I'm not doubting that it you did see a difference, I'm just trying to interpret it so that it makes sense where it would really help.

[methimpikehoses]

I agree that it helps most in CPU bound applications, absolutely.

I also hate when I delete a whole bunch of text... usually happens to me when I'm writing an email then pause to think about what to say next, and end up deleting a whole train of thought.

:headexplode:

[ikjadoon]

From a friend:

"The big boost in Crysis scores that you obtained from your E8500 were mostly because the entire Crysis game executable "fits" in the L2 cache of the CPU. 6MB's is more than enough to house the engine locally in the CPU die while your older CPU only had 4MB's (available). Remember that the Quad Cores do not share cache (if you were thinking you had 8MB's available to the game) nor does the game provide code for this. The Crysis game engine is about 5% too big to fit into the 4MB cache of the 65nm CPU's. As you already know, having the game executable a thousandth of an inch away from the processing cores (vs. 4 inches away in the RAM banks) reduces communication latency immeasurably."

I guess the same is for the Q6600 VS Q9650 debate.

~Ibrahim~

[Bman212121]

Somehow I don't believe that... It may have some advantages related to cache, but it doesn't make sense that the cpu would load the entire game executable into L2.

[ikjadoon]

I mean, the guy (http://www.overdrivepc.com/) is reputable; he actually built the rig in my sig (and plenty more). Last clocks I saw, he was selling the E8600 @ 4.2GHz on air, backed with a warranty. The quote was from the CTO, Mario Cifaldi. I usually take his word on computer stuff, lol.

Let me see if I can find a bench...

[MUYA]

Actually here is another good question. Is there a particular title where it feels like the game is running slow?

One thing that would be note worthy is there are no minimum fps numbers shown. The cpu could potentially have a larger impact if you were being cpu limited and that was causing the slowdowns. The only game I can think of where I've seen cpu slowdowns is Far Cry 2. When the fire is on the screen during the benchmark I can see the SLI scaling shrink on the bars to very little while the fps drops. There are other times where I've seen the bars shrink, but usually the fps doesn't drop meaning that I'm already sitting on a frame cap or something else isn't scaling right.

Actually I can think of another time where that happens, if you play an MMO and there is a bloat load of characters on the screen.
If you read the report, i7 does benefit the 295...how much is relative to what you can justify for yourself as a purchase. Certainly in some situations, an overclock in quadcores seems to boost average fps. As you mentioned, min fps has been overlooked but, still average fps is still indicative as a form of measurement...not fool proof but still.

I hear your point about all the other intricacies of benching/gaming but, coming from a q6600@3Ghz and finding 400mhz more could benefit you (as I mentioned onionmarks). More specifically, for SLI reliant situations in which the 295 is an example, I would think (if you have cash/credit to burn for a 295...) you could probably consider such things. I hope you understand what I am trying to say.

I intend to bench crysis warhead and far cry2.

I play AoC a lot and will try to bench it with Fraps...but benching consistently in AoC is a problem. You probably have rely on feel, perception more than anything else and that is always subjective

[MUYA]

I agree that it helps most in CPU bound applications, absolutely.

I also hate when I delete a whole bunch of text... usually happens to me when I'm writing an email then pause to think about what to say next, and end up deleting a whole train of thought.

:headexplode:

My point too, I think a q6600@3Ghz may have been hampering the full potential of my 295 as you know. I was along the thinking that perhaps that q9650 might benefit me.

My q6600 has terrible thermal profile. It needs vcore of 1.45v to get to 3.4Ghz...watercooled and still OCCT brings it to 70 degrees plus!!!

When benching crysis it was logged hover around 50-60 degrees though


It idles at 44-47 degrees!!!

[methimpikehoses]

Well from my own experience testing last weekend, the Q9650 is substantially faster at 3.8-4.0ghz than a Q6600 at 3.6ghz, for whatever that's worth. Basically, a decent-high OC for each chip for use on air, and you will get noticeably better performance on the Q9650, at least with a dual-GPU chip.

That was my thought in getting a Q9650, to see if it could get more out of the GTX 295, which it did. But since my mobo isn't up to the challenge (I think I would need a P45 or a X48), I'll wait for a good deal on an i7 setup.

Fast CPU is FTW on a dual GPU setup fo sho.

[MUYA]

Well from my own experience testing last weekend, the Q9650 is substantially faster at 3.8-4.0ghz than a Q6600 at 3.6ghz, for whatever that's worth. Basically, a decent-high OC for each chip for use on air, and you will get noticeably better performance on the Q9650, at least with a dual-GPU chip.

That was my thought in getting a Q9650, to see if it could get more out of the GTX 295, which it did. But since my mobo isn't up to the challenge (I think I would need a P45 or a X48), I'll wait for a good deal on an i7 setup.

Fast CPU is FTW on a dual GPU setup fo sho.

I am thinking of an i7 but....thats too much capital! :p I can do it but, would have sacrifice on other things too so, I might just wait till the Q9650 drops some more :p

[methimpikehoses]

Sell your body.

[MUYA]

Sell your body.

That's not a sacrifice I am willing to make....or others willing to make on me! bwahahahaha!