Multi Monitor Performance with Single PC (980X / NV470)

[whitav8r]

I wanted to report the results of some tests that I was fortunate to be able to run on a new PC with a 980X 6 core CPU running at 4.0 Ghz (CL7 memory timing) with an Nvidia 470 connected to a Matrox triple Head 2 Go. Three monitors were connected to the TH2Go for three separate windowed, independent views with a large field of view (210 deg) - the center large screen at Zoom=.33 which provides +-45 degrees and the two side views centered at directly sideways with +-30 degrees (Zoom = .7). Two additional monitors were running a helicopter instrument panel and a GPS display. I realize that using three independent PCs could provide even more performance but the goal here was just to see the performance with a single PC system. The Scenery settings were pretty modest but still fine for a good scene particularly when using the FlyTampa Hong Kong scenery. It was found that Vsync could be left off (providing higher frame rates) when using the TH2Go since it provides the final sync and removes the picture tearing. This system would be usable for a home cockpit that wants to keep it simple. Smoothness was also very important (particularly for hovering and high rate tail rotor turns)and internal/external limiters were shown to provide some improved fluidity. Framerates listed below were without any limiting. The monitors were also driven straight without a Th2Go using a second video card but UsePools=0 caused some artifacts and RejectThreshold had to be utilized - frame rates were somewhat lower.

(3 x 1400x900 views) (Aniso=8x,AA=8xQ)

Dense Scenery - Normal Autogen - test with yaw rate
NOTE: Scenery Settings Med, 38,38,1,Mid 1.x, no traffic
Field of View (90,60,60 - .33,.7,.7) Weather=Building Storms, TOD=day/dusk
Increased Load - Light Traffic, Very Dense Scenery, More res, Water Low 2.x, Dusk

----------------------Baseline --- Inc Load
-----------------------day/dusk
Scenery Location

Megascenery Seattle
Downtown------ 60/45 --- 40
KSEA 34R------60/50 --- 41
KBFI 31L------- 54/44 --- 41
KRNT 33-------- 52/44 --- 47
KSEA ramp------ 53/40 --- 40
FlyTampa HongKong
Airport Gates 54/37---32
helipad 1------ 70/50 --- 41
helipad 2------ 71/51 --- 40
Megascenery Socal
KLAX 25L-------62/46 --- 40
KLAX 25R------- 64/55 --- 39
KLAX at tower-- 68/60 --- 46
KSAN 27-------- 87/85 --- 50
KPSP G1-------- 100/80 --- 55
KCRQ 6----------90/70 --- 50
Aerosoft Nice, France-- 55/53 --- 40
Aerosoft Monte Carlo--- 62/55 --- 41
Megascenery Oahu
Downtown------- 80/65 --- 55
mtns----------- 85/80 --- 54
Megascenery NewYork
KLGA------------50/40--- 35
Downtown------- 50/40--- 31

Tweaks Applied:
UsePools=0, Matrox TH2Go ( Vsync=0), FFTF=0.05, display.cfg, from Bojote/Needham list.

[Sean Nixon]

That's one heck of a post for it being your first! Thank you.

[vyper883]

I wanted to report the results of some tests that I was fortunate to be able to run on a new PC with a 980X 6 core CPU running at 4.0 Ghz (CL7 memory timing) with an Nvidia 470 connected to a Matrox triple Head 2 Go. Three monitors were connected to the TH2Go for three separate windowed, independent views with a large field of view (210 deg) - the center large screen at Zoom=.33 which provides +-45 degrees and the two side views centered at directly sideways with +-30 degrees (Zoom = .7). Two additional monitors were running a helicopter instrument panel and a GPS display. I realize that using three independent PCs could provide even more performance but the goal here was just to see the performance with a single PC system. The Scenery settings were pretty modest but still fine for a good scene particularly when using the FlyTampa Hong Kong scenery. It was found that Vsync could be left off (providing higher frame rates) when using the TH2Go since it provides the final sync and removes the picture tearing. This system would be usable for a home cockpit that wants to keep it simple. Smoothness was also very important (particularly for hovering and high rate tail rotor turns)and internal/external limiters were shown to provide some improved fluidity. Framerates listed below were without any limiting. The monitors were also driven straight without a Th2Go using a second video card but UsePools=0 caused some artifacts and RejectThreshold had to be utilized - frame rates were somewhat lower.

(3 x 1400x900 views) (Aniso=8x,AA=8xQ)

Dense Scenery - Normal Autogen - test with yaw rate
NOTE: Scenery Settings Med, 38,38,1,Mid 1.x, no traffic
Field of View (90,60,60 - .33,.7,.7) Weather=Building Storms, TOD=day/dusk
Increased Load - Light Traffic, Very Dense Scenery, More res, Water Low 2.x, Dusk

----------------------Baseline --- Inc Load
-----------------------day/dusk
Scenery Location

Megascenery Seattle
Downtown------ 60/45 --- 40
KSEA 34R------60/50 --- 41
KBFI 31L------- 54/44 --- 41
KRNT 33-------- 52/44 --- 47
KSEA ramp------ 53/40 --- 40
FlyTampa HongKong
Airport Gates 54/37---32
helipad 1------ 70/50 --- 41
helipad 2------ 71/51 --- 40
Megascenery Socal
KLAX 25L-------62/46 --- 40
KLAX 25R------- 64/55 --- 39
KLAX at tower-- 68/60 --- 46
KSAN 27-------- 87/85 --- 50
KPSP G1-------- 100/80 --- 55
KCRQ 6----------90/70 --- 50
Aerosoft Nice, France-- 55/53 --- 40
Aerosoft Monte Carlo--- 62/55 --- 41
Megascenery Oahu
Downtown------- 80/65 --- 55
mtns----------- 85/80 --- 54
Megascenery NewYork
KLGA------------50/40--- 35
Downtown------- 50/40--- 31

Tweaks Applied:
UsePools=0, Matrox TH2Go ( Vsync=0), FFTF=0.05, display.cfg, from Bojote/Needham list.

If you are going to use the "UsePools=0" , add this entry to the [Graphics] section of your fsx.cfg, and see if it helps with the artifacts:

[Graphics]
HIGHMEMFIX=1
ForceWindowedVSync=1

Try this with vsync on. The misconception here about vsync is that it affects performance. Nothing could be furter from the truth. With vsync on, you are basically removing the horizontal tearing effect, when panning left to right. What is happening though is that vsync tries to lock FPS to your screens refresh rate, which makes it appear that perf is suffering, when your fps exceeds the monitors refresh rate. Try using an external frame limiter, and lock the FPS to your screens refresh rate (60 fps= 60hz) and see if that makes things smoother for you.

[whitav8r]

Thanks much for the tips. I've seen the STALE_BUFFER_THRESHOLD=214748364 in someone's fsx.cfg but didn't realize what it might be able to do. (By the way, the comment about Stale... showed up in a direct message to me but not on the forum - how can that be?) Also, have you or anyone else compared framerates between using a TH2Go and just three ports of two video cards such as a 470/450 combo? My experience was that it was somewhat slower (10%) than with the Matrox box but, of course, you can completely control your resolutions with separate video cards. With regard to Vsync, my experience was that if you don't have to use it (due to usage of the Th2Go), you have more flexibility. I did try the external limiter and found some fluidity with it but I just wanted to report the unlimited data. The most significant point of the report is simply the very wide field of view (210 degrees) results being pretty good. Thanks again!