Benchmark for Black Shark?

[AlphaInfinity]

Ok I am a little confused. The cockpit resolution setting is suppose to set the resolution and refresh for the SHVAL camera right? It doesn't have anything to do with cockpit texture resolution?

Edited December 24, 2008 by AlphaInfinity

[TX-EcoDragon]

Ok I am a little confused. The cockpit resolution setting is suppose to set the resolution and refresh for the SHVAL camera right? It doesn't have anything to do with cockpit texture resolution?

 

From the manual:

 

"Res. of Cockpit Displays. For aircraft that have displays such as mirrors, multifunction displays, moving maps, etc., you can adjust the resolution of these displays using the dropdown list. Note higher resolution settings can negatively impact the smoothness of game play"

 

 

I didn't mean for my post to be a final post - it's a work in progress for me, I want to test a few other settings, and I would also like to see how performance changes as resolution does, a higher video card OC, as well as 4x AA etc. . . I just decided to post it unfinished rather than whenever I get it "done". . .if I ever do! :-D

Edited December 24, 2008 by TX-EcoDragon

[TX-EcoDragon]

Windows 7 preliminary testing

 

Windows 7 Preliminary testing in BlackShark has some very interesting results!

 

Remember the run I posted above for 3.06 GHz? Well, no need to go look up there for it. . .here it is:

 

Windows XP E8400 @ 3.06 GHz (speedstep enabled) DX9.0c 2x1Gigs of PC8000 RAM at 850Mhz, 8800GTS (G92) at stock clocks 178.24 drivers

Frames: 16232 - Time: 480193ms - Avg: 33.803 - Min: 5 - Max: 61

 

The following is the same run, but in Windows 7 (64 bit), with DirectX11, all other settings the same ( more than 2 gigs of RAM doesn't appear to make a difference in BS):

 

Windows7 core 0 E8400 @ 3.06 GHz (speedstep enable) D11 2x2gigs PC8500 at 850Mhz, 8800GTS (G92) at stock clocks 178.24 drivers

Frames: 13802 - Time: 480225ms - Avg: 28.740 - Min: 7 - Max: 60

 

So that's not so great. . .at this point it's not looking very good for Windows 7. . .but it's supposed to run like a better version of Vista, which supposedly is strong in Black Shark. . .so what gives? Ahhh, the CPU affinity trick you say?? Well, lets see if that makes up the lost performance!

 

For this run, settings are the same, only I enable both CPU cores in the taskmanager:

 

Windows7 core 0+1 core E8400 @ 3.06 GHz (speedstep enabled) DX11 2x2gigs PC8500 at 850Mhz, 8800GTS (G92) at stock clocks 178.24 drivers

Frames: 21729 - Time: 480340ms - Avg: 45.236 - Min: 23 - Max: 62

 

So umm. . . can you say AWESOME?!??!!!?! Not only did I get the lost performance back, but I set something of a record for this benchmark on my machine! If you take a look at my previously posted Windows XP run at the same settings the best I managed was - Avg: 42.185 - Min: 10 - Max: 62 and that was overclocked to 3.9 GHz!!!!!!

 

 

Even my runs at 3.960GHz with 2x2Gigs of PC8500 RAM at 1100MHz, only averaged: 44.439fps in XP. . .so simply running Windows 7, and using both CPU cores gave me .8 avg fps better performance at a stock 3.06 Ghz than at nearly 4.0 GHz in 32 bit Windows XP!!

 

I can't wait to see what it does at 3.96 GHz!!! Well, yeah I can, I need to sleep now that it's 6:30 am!!!! :music_whistling:

 

[Edit to add the results from the 3.96 GHz runs]

 

Windows XP 32 E8400 @ 3.960GHz 2x2Gigs of PC8500 RAM at 1100MHz, 8800GTS (G92) at stock clocks vSynch and triple buff on on 178.24 drivers

Frames: 21281 - Time: 480107ms - Avg: 44.439 - Min: 10 - Max: 63

 

Windows7 (64) core 0+1 E8400 @ 3.960GHz 2x2Gigs of PC8500 RAM at 1100MHz, 8800GTS (G92) at stock clocks 178.24 drivers vSynch ON

Frames: 26647 - Time: 480074ms - Avg: 55.506 - Min: 27 - Max: 62

 

Windows7 (64) core 0+1 E8400 @ 3.960GHz 2x2Gigs of PC8500 RAM at 1100MHz, 8800GTS (G92) at stock clocks 178.24 drivers vSynch OFF

Frames: 28235 - Time: 479994ms - Avg: 58.823 - Min: 32 - Max: 93

 

Impressive results no? Under the same conditions I had an increase of 11.1 frames per second better on average when overclocked to 3.96 1100MHz DDR2, and 14.4 frames per second better when vSynch and triple buffering are forced off in the driver control panel (the normal way in which benchmarks are run).

Perhaps most impressive is the minimum fps. . .they are almost as good as the average fps at stock clocks!!!! My track really killed the fps in a flew places on XP, but with 7, and both CPU cores, that's gone!!!!!

Also, I am using drivers that are not technically compatible with WIndows 7 in the interest of running the same driver as I'd used previously. Next I'll try the driver MS suggests, and perhaps the 185 series beta that shows big gains in many titles.

Edited January 10, 2009 by TX-EcoDragon
3.96 data

[Feuerfalke]

Some nice and detailed testing. Thanks for sharing! :thumbup:

[pumpkin]

Can someone please give me a link to the track. I can't find it in this thread.

Thanks!

[Airway]

Well what track are you talkin about?

[pumpkin]

the benchmark track they are talking about in this thread

[Axel40]

Well it's not here in this thread now.

 

Anyone care to link it or is it a state secret?

[hitman]

Windows 7 Preliminary testing in BlackShark has some very interesting results!

 

Remember the run I posted above for 3.06 GHz? Well, no need to go look up there for it. . .here it is:

 

Windows XP E8400 @ 3.06 GHz (speedstep enabled) DX9.0c 2x1Gigs of PC8000 RAM at 850Mhz, 8800GTS (G92) at stock clocks 178.24 drivers

Frames: 16232 - Time: 480193ms - Avg: 33.803 - Min: 5 - Max: 61

 

The following is the same run, but in Windows 7 (64 bit), with DirectX11, all other settings the same ( more than 2 gigs of RAM doesn't appear to make a difference in BS):

 

Windows7 core 0 E8400 @ 3.06 GHz (speedstep enable) D11 2x2gigs PC8500 at 850Mhz, 8800GTS (G92) at stock clocks 178.24 drivers

Frames: 13802 - Time: 480225ms - Avg: 28.740 - Min: 7 - Max: 60

 

So that's not so great. . .at this point it's not looking very good for Windows 7. . .but it's supposed to run like a better version of Vista, which supposedly is strong in Black Shark. . .so what gives? Ahhh, the CPU affinity trick you say?? Well, lets see if that makes up the lost performance!

 

For this run, settings are the same, only I enable both CPU cores in the taskmanager:

 

Windows7 core 0+1 core E8400 @ 3.06 GHz (speedstep enabled) DX11 2x2gigs PC8500 at 850Mhz, 8800GTS (G92) at stock clocks 178.24 drivers

Frames: 21729 - Time: 480340ms - Avg: 45.236 - Min: 23 - Max: 62

 

So umm. . . can you say AWESOME?!??!!!?! Not only did I get the lost performance back, but I set something of a record for this benchmark on my machine! If you take a look at my previously posted Windows XP run at the same settings the best I managed was - Avg: 42.185 - Min: 10 - Max: 62 and that was overclocked to 3.9 GHz!!!!!!

 

 

Even my runs at 3.960GHz with 2x2Gigs of PC8500 RAM at 1100MHz, only averaged: 44.439fps in XP. . .so simply running Windows 7, and using both CPU cores gave me .8 avg fps better performance at a stock 3.06 Ghz than at nearly 4.0 GHz in 32 bit Windows XP!!

 

I can't wait to see what it does at 3.96 GHz!!! Well, yeah I can, I need to sleep now that it's 6:30 am!!!! :music_whistling:

 

If its going to be like that, I cant wait to try it on my new rig! Ill be trying Windows 7 if and when I can download it with a serial key. Hope its not too late for me, wont see my new psu for another week or so.

 

Rep inbound.

[zokier]

TX-EcoDragon are you using vsync? seems so as the max fps floats around 60, which matches 60 Hz of most LCD-displays. You could get higher avg fps without vsync, but on the other hand the results are nearer real-life performance with vsync

[TX-EcoDragon]

OK, I've updated my last post to include the results at 3.96 GHz with Windows 7!

 

 

I can upload the track I used if you would like to compare to my data. Keep in mind this is just how *I* benchmarked - anyone else's data is from something different.

 

To compare to my data, use FRAPS (etc) to generate benchmark data between the instant the switch is flipped to turn on the helmet mounted sight, and then stop capturing 8:00 minutes later. All settings were posted back in Post 24 for driver level settings and in game settings unless specified changed for an individual run.

 

Here is the trackfile I used: http://www.megaupload.com/?d=VRED889L

Edited January 11, 2009 by TX-EcoDragon

[TX-EcoDragon]

TX-EcoDragon are you using vsync? seems so as the max fps floats around 60, which matches 60 Hz of most LCD-displays. You could get higher avg fps without vsync, but on the other hand the results are nearer real-life performance with vsync

 

Yes, I know I didn't organize everything really well, but I try to specify when vsynch is on or off for each section that is different from the generic settings posted at the start of post 24. Of course if the max fps are 60-62 then chances are it was on.

 

I tried to do runs at most settings with vSynch both on and Off.

 

In sims tearing can be really obvious when looking out through cockpit bars, so I usually keep it on, as do most simmers I know, and yet almost all benchmarking review site don't use it - I understand why they don't, but that does largelly invalidate their testing from an actual gameplay perspective. Flight sims tend to have wild fluctuations in frame rate simply by looking up at the sky above you where they might hit triple digits, or looking at the ground where they might hit 10. The mathematics of this means that the average frame rate may get increased beyond what the user would actually see most of the time (ie it's higher than average) by turning off vynch. Black Shark doesn't seem to do this, but in many sims,I'll get 300 fps looking at the sky.

 

Something like Crysis doesn't do that as much - the user spends most all of the time surrounded by similar scenery, and looking out at a level attitude, and in close proximity to all the 3D objects. . .none of this applies to a flight sim.

 

While the 10 fps will lead to an obvious performance hit, the triple digit fps are transparent to users, and since vSynch can lower the minimum fps on many titles, and not just the max, I think real world testing should have vSynch on. I don't care how many max fps I get, I care about the average, and the minimums. vSynch will decrease performance further than just limiting max fps, so if there are big changes in min fps or much lower average fps then I want to show that. It is based on those numbers that I will set my graphics and system performance, not the max, and ultimately, I did all this testing only for myself to figure out how I want to run Black Shark.

 

I figure that you know that - and were just sharing info with me, which I appreciate. . .but I guess it prompted me to want to explain my reasoning!

Edited January 11, 2009 by TX-EcoDragon

[Axel40]

Thanks for posting the track link mate.

[Airway]

A thank you from me too, TX-EcoDragon :thumbup:

[ocram]

 

Here is the trackfile I used: http://www.megaupload.com/?d=VRED889L

 

Sorry, but where is the track file in there?

There are a lot of files but no *.trk.

 

I would like to compare your track. To me it seems like your Windows XP is to slow. You oced your 8400 to 3,9 GHz. I have my 8600 clocked @ 4,2 GHz and I have average FPS with the original stress test track around 80 - 90 with 1680x1050 and the same settings as yours. I dont think that my 280GTX is the only reason that doubles the FPS. And my AF/AA settings are actually at 16/4 (you used 8/2).

 

As far as I know BS could only use 1 core (correct me if I am wrong). So the best that can happen is that other cores using the all other programms that are running in the background. I dont know how that could result in doubling the FPS with Windows 7.

[pumpkin]

Thanks a lot for posting the download.

[7rooper]

Impressive performance gain when using two cores in windows 7. Can't wait to see what happens with 3 or 4 cores (how about an Intel i7?).

Guys, so far nobody has posted graphics settings stating if mirrors are on or off in their benchmark results. That's important for an apples to apples comparison.

 

Downloading the track. I'll post my results later on

[Slayer]

Windows 7 Preliminary testing in BlackShark has some very interesting results!

 

Remember the run I posted above for 3.06 GHz? Well, no need to go look up there for it. . .here it is:

 

Windows XP E8400 @ 3.06 GHz (speedstep enabled) DX9.0c 2x1Gigs of PC8000 RAM at 850Mhz, 8800GTS (G92) at stock clocks 178.24 drivers

Frames: 16232 - Time: 480193ms - Avg: 33.803 - Min: 5 - Max: 61

 

The following is the same run, but in Windows 7 (64 bit), with DirectX11, all other settings the same ( more than 2 gigs of RAM doesn't appear to make a difference in BS):

 

Windows7 core 0 E8400 @ 3.06 GHz (speedstep enable) D11 2x2gigs PC8500 at 850Mhz, 8800GTS (G92) at stock clocks 178.24 drivers

Frames: 13802 - Time: 480225ms - Avg: 28.740 - Min: 7 - Max: 60

 

So that's not so great. . .at this point it's not looking very good for Windows 7. . .but it's supposed to run like a better version of Vista, which supposedly is strong in Black Shark. . .so what gives? Ahhh, the CPU affinity trick you say?? Well, lets see if that makes up the lost performance!

 

For this run, settings are the same, only I enable both CPU cores in the taskmanager:

 

Windows7 core 0+1 core E8400 @ 3.06 GHz (speedstep enabled) DX11 2x2gigs PC8500 at 850Mhz, 8800GTS (G92) at stock clocks 178.24 drivers

Frames: 21729 - Time: 480340ms - Avg: 45.236 - Min: 23 - Max: 62

 

So umm. . . can you say AWESOME?!??!!!?! Not only did I get the lost performance back, but I set something of a record for this benchmark on my machine! If you take a look at my previously posted Windows XP run at the same settings the best I managed was - Avg: 42.185 - Min: 10 - Max: 62 and that was overclocked to 3.9 GHz!!!!!!

 

 

Even my runs at 3.960GHz with 2x2Gigs of PC8500 RAM at 1100MHz, only averaged: 44.439fps in XP. . .so simply running Windows 7, and using both CPU cores gave me .8 avg fps better performance at a stock 3.06 Ghz than at nearly 4.0 GHz in 32 bit Windows XP!!

 

I can't wait to see what it does at 3.96 GHz!!! Well, yeah I can, I need to sleep now that it's 6:30 am!!!! :music_whistling:

 

[Edit to add the results from the 3.96 GHz runs]

 

Windows XP 32 E8400 @ 3.960GHz 2x2Gigs of PC8500 RAM at 1100MHz, 8800GTS (G92) at stock clocks vSynch and triple buff on on 178.24 drivers

Frames: 21281 - Time: 480107ms - Avg: 44.439 - Min: 10 - Max: 63

 

Windows7 (64) core 0+1 E8400 @ 3.960GHz 2x2Gigs of PC8500 RAM at 1100MHz, 8800GTS (G92) at stock clocks 178.24 drivers vSynch ON

Frames: 26647 - Time: 480074ms - Avg: 55.506 - Min: 27 - Max: 62

 

Windows7 (64) core 0+1 E8400 @ 3.960GHz 2x2Gigs of PC8500 RAM at 1100MHz, 8800GTS (G92) at stock clocks 178.24 drivers vSynch OFF

Frames: 28235 - Time: 479994ms - Avg: 58.823 - Min: 32 - Max: 93

 

Impressive results no? Under the same conditions I had an increase of 11.1 frames per second better on average when overclocked to 3.96 1100MHz DDR2, and 14.4 frames per second better when vSynch and triple buffering are forced off in the driver control panel (the normal way in which benchmarks are run).

Perhaps most impressive is the minimum fps. . .they are almost as good as the average fps at stock clocks!!!! My track really killed the fps in a flew places on XP, but with 7, and both CPU cores, that's gone!!!!!

Also, I am using drivers that are not technically compatible with WIndows 7 in the interest of running the same driver as I'd used previously. Next I'll try the driver MS suggests, and perhaps the 185 series beta that shows big gains in many titles.

 

 

Strange, I thought since Vista came out that VSYNC is forced on all the time and no way to turn it off...

[miguez]

Strange, I thought since Vista came out that VSYNC is forced on all the time and no way to turn it off...

 

Hey Slayer,

 

You can just go to the nVidia Control Panel (normally installed with nVidia ForceWare drivers) and turn it off. I do that in Vista all the time.

[TX-EcoDragon]

Impressive performance gain when using two cores in windows 7. Can't wait to see what happens with 3 or 4 cores (how about an Intel i7?).

Guys, so far nobody has posted graphics settings stating if mirrors are on or off in their benchmark results. That's important for an apples to apples comparison.

 

Downloading the track. I'll post my results later on

 

Did I leave that out??? I'll edit it - mirrors were on for all runs but one (which I probably didn't post) I didn't see any real fps change with them on vs off on my machine, at least when running XP.

 

Also, keep in mind that I'm not comparing 7 to Vista - Vista to 7 gains might be minimal, though from what I've read, it does seem that 7 is much less resource consuming than Vista was.

Edited January 12, 2009 by TX-EcoDragon

[TX-EcoDragon]

Sorry, but where is the track file in there?

There are a lot of files but no *.trk.

 

I would like to compare your track. To me it seems like your Windows XP is to slow. You oced your 8400 to 3,9 GHz. I have my 8600 clocked @ 4,2 GHz and I have average FPS with the original stress test track around 80 - 90 with 1680x1050 and the same settings as yours. I dont think that my 280GTX is the only reason that doubles the FPS. And my AF/AA settings are actually at 16/4 (you used 8/2).

 

As far as I know BS could only use 1 core (correct me if I am wrong). So the best that can happen is that other cores using the all other programms that are running in the background. I dont know how that could result in doubling the FPS with Windows 7.

 

Yes, in XP it really only uses one core, not so in Vista/7.

As far as differences, well, are you running the Russian version or the English version? Mine is Russian, and I've heard that the English version has different flares and smoke that are much easier on the system. Are you running your machine on XP? Clearly we can't compare 7 or Vista with both cores running to XP. I only did one session of testing at resolutions other than 1920*1200, and while there was not much difference, the averages of course go up at lower resolutions. What about vSynch (most of my runs are with vSynch on which keeps maximum fps from going above ~60, and will also lower overall performance)? Certainly we can be sure that BS scales well with CPU clock speeds so your 4.2 is certainly going to be a bit faster, especially if you reach that magic point where the CPU stops being the bottleneck. At that point, the difference between our video cards will matter!

 

Either way, you can't really compare performance without running the same benchmark. The data from that track is pretty darn different than usual in game performance given the number of objects in the area being so high.

 

I haven't found this "original stress test track" so I used the one I posted, and recorded during the periods I previously mentioned.

 

And no, there is only one track file in that download I linked to, it's called battlebenchmark.trk if you see something else, that's not my link. Just follow that link, enter the 3 letter code that it shows you, wait the 45 or whatever seconds, press "free download" and then the track should download.

 

I don't know how much the video card will change things in BlackShark - I know that in my other sims, anything more than my card doesn't do much for performance, and in FSX, oddly, my card is much faster than the latest cards. Also interesting is that in FSX, adding higher AA settings may actually raise fps in some cases - strange but true. Who knows where BS fits into things.

Edited January 12, 2009 by TX-EcoDragon

[Slayer]

Hey Slayer,

 

You can just go to the nVidia Control Panel (normally installed with nVidia ForceWare drivers) and turn it off. I do that in Vista all the time.

 

What I'm refering to is this off the Nvidia website:

 

"Due to architectural changes in the new Windows Vista Window Display Driver Model (WDDM), the graphics driver can no longer disable vsync from its own driver or Control Panel. Selecting this option from the NVIDIA Control Panel will have no affect on DirectX applications. For applications that use Direct3D on Vista, use the vertical sync setting within the application. We are adjusting the help text in the NVIDIA Control Panel to make this clearer to our customers."

 

In other words unless they program an option in the game to enable/disable vsync, that setting will do absolutely nothing for games with Direct3d in Vista...there is no setting to disable Vsync in DCS:BS

Edited January 13, 2009 by Slayer

[zokier]

What I'm refering to is this off the Nvidia website:

 

"Due to architectural changes in the new Windows Vista Window Display Driver Model (WDDM), the graphics driver can no longer disable vsync from its own driver or Control Panel. Selecting this option from the NVIDIA Control Panel will have no affect on DirectX applications. For applications that use Direct3D on Vista, use the vertical sync setting within the application. We are adjusting the help text in the NVIDIA Control Panel to make this clearer to our customers."

 

In other words unless they program an option in the game to enable/disable vsync, that setting will do absolutely nothing for games with Direct3d in Vista...there is no setting to disable Vsync in DCS:BS

 

well, the results tell otherwise. maybe that applies to d3d10 apps?

[BakaNi]

well, the results tell otherwise. maybe that applies to d3d10 apps?

 

You can disable vsync with dx10 apps, the statement by nvidia is wrong/outdated.

 

It might be some semantic point, such as you cannot disable vsync globally but instead we just disable it on a per program basis.

[Maximus_G]

...So if you are posting any comments, identify if you are using the current Russian version, or the forthcoming English version. This track does show some different views between the Russian and English versions.

 

Soooo, does this track display _the same_ results in _release_ 1.0 versions of the game, Russian and English?