FSXMark11

[Kosta 122]

I decluttered the post by removing the previous wrong test and following comments. If you make a new one, I'll remove this one too. Just post all the details, like the GPU clock etc.

[DescendDescend 25]

Here you go.

 

 

I ran the complete cycle twice and on the second brought up monitoring on the GPU. Like my tests last night the GPU never stressed over 60-65% and the GPU clock was no where near maxed.

 

GTX680 results in this test look identical to similar clocked i5's.

 

Sorry again about missing that line. I must fault a good bourbon for the snafu.

 

Hi Word Not Allowed

 

Just wondered if you were going to add these results to the sheet as they seem to be the only ones for a 680.

[Kosta 122]

I will. But waiting for the final test, as he said something about the clock...

[JohnFromCO 15]

I will. But waiting for the final test, as he said something about the clock...

 

I backed down Precision X and verified boost clock settings now don't disappear.

 

Never saw FSX run the card at 100% load (stayed at 40-60% throughout) or run max clocks for more than a few seconds here and there.

 

Results are essentially identical even with boost kicking in to 1324 (boost plus 200 offset) occasionally around the course.

[Kosta 122]

OK then I'll just enter them into the results. Thanks.

 

Infact, me overclocking my 580 doesn't bring squat for FSXMark11, so I guess the same is with 680.

[quadraspleen 7]

Hi guys,

 

OK, I totally rebuilt my rig from scratch. I think I had the fever or something; or I'm getting too old for this..anyway; totally new build; zeroed BIOS.

 

The ASUS Rampage Extreme has a stock speed for the 3930K of 3.8Ghz.

For the first test, I upped it by x1 to 3.9, then by 3 each time up to 4.5 - all of these wre using my memory's stock XMP profile #1 (timings are on the sheet) this takes the RAM to 2133. Other than that, I just raised the core clock. All other settings on auto except LLC and Power Phase control (Medium and Optimised).

The CPU Vcore hasn't really gone above 1.4v and usually hovers around 1.35. Temps in the medium to high 50's when runnig the tests.

I am going to do some proper manual tweaking of this board when I have the time. It looks to be a fiddlers paradise!

I noticed interesting stuff when not using the XMP profile at higher clock speeds; the FPS went down a couple. RAM speed is very important as well, it seems (I expect you guys know all of this already..this is all new to me..for years I never tweaked anything in FSX, and only had a proper O/Cable board since the P8P67 I had before this)

 

All tests performed 100% according to the procedures. Really interesting stuff. I'm going to do a split-build on my rig so I can have a permanently "new" FSX just for testing O/C's on my board. I reckon I can get much, much higher scores than this..Seeing as how I don't really do anything else with this PC LOL

 

Anyways, here is the new 3930K at 3.9;

 

4.2

and 4.5Ghz:

 

best

 

jake

[Kosta 122]

About split FSX installations: I too have two, but my system knows of only one. I think that's a better solution. I basically have 3 folders in each FSX specific folder/drive. You know of ProgramData FSX folder, aber 2 folders in Users etc. They are all called FSX. So, I created FSX, FSXvanilla and FSXcurrent. FSX is always whats active. If I want to exchange the installations I only move files accordingly. I think this is a better solution, since I can have two separate FSX folders where FSX.cfg resides, two separate shader folders etc.

 

About results: They seem pretty low. I compared against the FSXMark11 scores, and some comparable scores are quite higher. Given, those can be wrong also due to wrong WideViewAspect setting giving higher score. Compare just one:

Flying Fred 48.2 31.0 62.0 i7-2600k 4.46 On GTX 580 512 3072 1600 9 8 SSD No

 

I don't know who measured falsly, or is it the chip. Would be interested to know for sure.

[quadraspleen 7]

Hey Word Not Allowed,

 

That's a good idea re the different folder thing. I might give that a try. re the results..Hmm.. I have run a heap more than those I published (I'm very tired today LOL) and I'm not sure I can go with the 48 on a 2600 @4.46, but he may have tweaked his board to hell, who knows?

 

Looking through the 2600/2700 results in this page and on the google spreadsheet, there do seem to be some anomalies (imo) For instance, if you scroll right to the top of this page, rex251 has a 920 @ 4.3 and he's seeing 35.9 (560 card tho) and then John further down, with his 2500 @ 4.7 is seeing 43.7. Me with my 3930 @ 4.5 gets 42.5 so there isn't much in it.

 

So I dunno..I'm very methodical (mostly) and love this kind of thing, being from the IT industry, so I was absolutely fastidious in my following of the instructions to the letter..I wouldn't be suprised, as you say, if ppl had got a tiny part of it wrong (the wideview and resolution for two)..But I think the trend is roughly your Ghz=your framerate give or take a few..The 5.5Ghz monster at the top basically gets in the 50's, and so on down; but there will always be exceptions..

 

Bear in mind that I haven't really tweaked much of the board yet beyond the main core speed, as it's new and my wallet is still a bit sore; but I don't mind losing a chip from time to time in the cause of experimentation, so who knows...Great test though..bit boring after watching it 30 or 40 times, though LOL

 

I will say, however, that I am seeing much smoother FSX with a lower clock and the same cfg settings on my new rig than on my old (faster RAM, PCH etc etc) so it's a good thing.

 

best

 

jake

 

EDIT: I'm going to re-run the tests at my stable 4.5 with wideview FALSE to see if I can replicate the higher settings - Also, does monitor size affect the test? (obv all same tested res but I remember reading that FSX likes its monitors native resolution? Mine is 1920x1200)

[Kosta 122]

Also, does monitor size affect the test?

 

No. Only the resolution. If you set lower resolution and run full screen on the Nvidia card, it's usually going to show small picture by default, not scale. Though you can set both.

[Kronzky 34]

Here are some results for a Radeon 6950.

While it wasn't good enough for the Top 10, at least the results were extremely consistent! :)

 

It seems the limiting factor was still the CPU, as the GPU activity never went above 60%.

 

[quadraspleen 7]

Interesting results. So at 5.0ghz the 3820 is a fraction slower than my 3930 @ 45.

 

In the interests of completism, and because I said I would, here are my 4.5 results with WideView FALSE. As you said, it is possible some people got this wrong. Who knows. I'm done tweaking and testing now. Stable at 4.6 and wanting to fly now...

 

[Kronzky 34]

Actually, I just bricked my brand-new Radeon this weekend (the BIOS flash didn't quite work out—but don't tell anybody...), so I've ordered a 560 (w/2GB) as a replacement.

 

It'll be interesting to see the results with identical hardware, and just a switched card.

[Ben Cap 89]

 

 

 

Actually, I just bricked my brand-new Radeon this weekend (the BIOS flash didn't quite work out—but don't tell anybody...), so I've ordered a 560 (w/2GB) as a replacement.

 

Same happened with me. I hosed my 580 2 weeks ago but just returned it for a new one.

[Kronzky 34]

Ok, so I got my replacement card (560GTX), and while it is a few FPS faster than the Radeon 6950, it is not really that significant (so perhaps at that range ATI vs NVIDIA is pretty close).

 

Just like the Radeon, the GPU utilization never went above ~60%.

I tried OC'ing it a bit, but that really had no measurable effect (I guess because it still had some "reserves" anyway).

 

So, here are the results:

[Ben Cap 89]

Ok, so I got my replacement card (560GTX), and while it is a few FPS faster than the Radeon 6950, it is not really that significant (so perhaps at that range ATI vs NVIDIA is pretty close).

 

Just like the Radeon, the GPU utilization never went above ~60%.

I tried OC'ing it a bit, but that really had no measurable effect (I guess because it still had some "reserves" anyway).

 

So, here are the results:

How did you manage to push the 3820 to 5GHZ? I thought it had a locked multiplier.