Benchmarking topic & performance comparisons

[Maurice]

As Warren suggested, this should be a new topic so I took the liberty of starting one to answer a post by Warren in another thread. Hope you don't mind Warren  . Anyway, here is a quote from Warren which I do not agree with:

"My take away is that regardless of system specs, tweaks, and mods, in terms of FPS, unwarped on a TH2Go is faster than warped, and both are faster than using multiple cards and heads.  The numbers don't lie."


Not so fast there Warren, numbers can & do lie sometimes  . My experience is that I see no real measurable difference between warped & unwarped (1 fps at best but too close to call) and my old TripleHead with a GTX480 gave me pretty much exactly the same frame rate & overall performance as my 690 with 4 outputs & surround (I use  three outputs).
Bottom line is that it is impossible to generalize about anything; that's what I think but it would be interesting to get more inputs from anyone who has tried the above combinations.

Maurice

[fsaviator]

No issues, Mau.

I used the FSmark07 benchmark flight for all my tests.  I ran all the tests after the system had been running for a while, and I have all 300 seconds of numbers, over Seattle.  No guessing involved.

While every system setup will give different numbers based on processor, software, settings... whatever, it doesn't matter.

The point was to see what the framerate hit was on this particular system (my old system).  Tomorrow, I'm going to pop the SSDs into my new system as I build it and run the same tests.  That will give me an idea of difference multi-core and bus make.  For these tests I'll have to still use my old video cards though as I don't have two of the new 670's.

I'll post all the results in this thread.

[fsaviator]

Moved this here.
Well, at the risk of hijacking this thread, I'll tell you all what I've found to help some make up their minds.  I'll start a new thread called Benchmarks or something once I'm done but since we're on this topic, a quick look at the numbers with three projectors at 1024x768 tell me this:

Tweaked, with a TH2Go, no warping, and limiting FPS to 29 = AVG is 23.21
Tweaked, with a TH2Go, no warping, and unlimited FPS = AVG is 34.02
Ultra High Setting, with a TH2Go, no warping, and unlimited FPS = AVG is 18.52

Tweaked, with a TH2Go, warping, and limiting FPS to 29 = AVG is 16.77
Tweaked, with a TH2Go, warping, and unlimited FPS = AVG is 20.41
Ultra High Setting, with a TH2Go, warping, and unlimited FPS = AVG is 19.66

Tweaked, with two identical dual head video cards, no warping, and limiting FPS to 29 = AVG is 11.96
Tweaked, with two identical dual head video cards, no warping, and unlimited FPS = AVG is 14.53
Ultra High Setting, with two identical dual head video cards, no warping, and unlimited FPS = AVG is 13.85

My take away is that regardless of system specs, tweaks, and mods, in terms of FPS, unwarped on a TH2Go is faster than warped, and both are faster than using multiple cards and heads.  The numbers don't lie.

the worst part is that at some point in tweaking I had limited my FPS and never thought about it and I find now that I would have gained and average of 11 FPS just by setting it back to unlimited.

[jskibo]

Warren, just curious why you need two of the 670s to test?

[fsaviator]

Not sure what I was thinking.  I was referencing connecting the three projectors and my lower eicas to the one computer.  I can do it with the one card, but I need the adapters.  I'll have to see what I have buried in the junk drawer.  If I have them, I'll run the tests connected to the one card

[jskibo]

Should be able to use the two DVI and one HDMI for the main display so only HDMI - DVI needed for test. I bought a DP to VGA for my lower.

[fsaviator]

It's  not about the connections on the card, but what I'm connecting it to.  My projectors are XGA's so I need VGA all around, as is my lower EICAS.  The card has two DVI, a Display Port, and an HDMI.  I think I have a DP to VGA, and I know I have an HDMI to VGA somewhere.  If not, I'll pick one up while I'm back in the US next week.

[fsaviator]

OK, back to this topic as I upgrade.

3.5GHz i7-3770K (No OC)
MSI Z77A-G45 Mainboard
16GB G-Skill DDR3-2400 Trident X Ram (No OC)
Galaxy GTX670 GC with 4GB (No OC)
ONE LCD only through the GTX670


-A fresh Windows 7 64-bit install on a 60GB OCZ SSD
-Fresh FSX SP1 Accleration Pack installed on a 128GB Kingston SSD
-Latest Nvidia driver for the 670

No tweaking of FSX
No tweaking of Win7
No tweaking of 670 Drivers.

Using the Fsmark07 Benchmarking Package with Fraps

TEST 1 (right after install with default Ultra High) = 40.31 FPS
TEST 2 (using the same settings my old system had) = 42.81 FPS
TEST 3 (All sliders and settings maxed) = 20.49 FPS

On the maxed test the FPS dropped to 12-16 over Seattle, about 18 over Boeing field, about 22 between Boeing field and SEATAC then about 18 over SEATAC

With the TH2Go and three views (as well as a fourth monitor which would be my lower EICAS), same tests:

TEST 1 (right after install with default Ultra High) = 12.71 FPS
TEST 2 (using the same settings my old system had) = 17.27 FPS
TEST 3 (All sliders and settings maxed) = 8.51 FPS

With the TH2Go and one stretched window across three LCDs (as well as a fourth monitor which would be my lower EICAS), same tests:

TEST 1 (right after install with default Ultra High) = 62.36 - Wow!  It appears that running the TH2Go here is a plus
TEST 2 (All sliders and settings maxed) = 27.18

I decided to throw in a bonus as I had everything all laid out already...

With three seperate views running off of three of the video card heads (DVi-D, DVI-A, HDMI).

TEST 1 (right after install with default Ultra High) = 15.28
TEST 2 (All sliders and settings maxed) = 5.90

Interestingly though the picture seems much clearer and more detailed through the three card heads as opposed to the TH2Go


Ouch!
Kind of demonstrates the performance hit you take in order to get the wrapped view.  This also doesn't take warping into account

[nicd]

Some interesting data there Warren. So with the three tests without the TH2GO - was that with a single stretched window, or 3 windows? Just want to make sure we're comparing apples with apples (i.e. 3 windows in both rounds of tests).

On a related point .. setting FPS to unlimited is an interesting topic. A recent issue I found is that, on some systems, FS9 and FSX will slam the audio feed if FPS is set to unlimited. I tore my hair out trying to fix an awful clicking, overdriven sound (using PMDG and NGX sound sets). Found an obscure forum post somewhere that suggested not using unlimited FPS, and that did the trick. Can even set the FPS to something like 100 instead of unlimited and it fixes the issue. But only relevant to some systems apparently - including high-end ones. I mention it here to save someone else's time (and hair!).

[fsaviator]

I was using just one LCD view, not stretched.

I will run a round of tests this evening with one stretched window across the three screens to add those numbers for comparison.  Then I plan to run the TH2Go tests using various tweaks.

Then I'm putting my new system back in the cockpit so I can get some flying done this weekend!

[fsaviator]

Some interesting data there Warren. So with the three tests without the TH2GO - was that with a single stretched window, or 3 windows? Just want to make sure we're comparing apples with apples (i.e. 3 windows in both rounds of tests).

I went ahead and ran the same tests with one stretched view across the three monitors through the TH2Go.  The results were very interesting.  The framerates were better than a single screen without a TH2Go.  It would appear that the TH2Go actually has a little rendering power built in

I also ran the same tests without the TH2Go but with three separate views all running off the card