Hi

A did a FSMark11 test with 3 monitor setup res 5040x1050.

my goal is to run 3 monitors with qute high settings, 1920x1080 or 2560x1440 monitors.

one thing is the GPU limiting and then CPU limithing factors.

Saab340 , Dazz,Tech or Word Not Allowed can explain my eng is so bad.

 

1920x1080 shall work but but would 2560x1440 work can the GPU be the limit, or both

on 1920x1080 can be it the GPU or?

 

Kind Regards/ Hasse

 

Settings by the book , res: 5040x1050

A resolution of 5040x1050 = 5.292.000 pixels

Every pixel needs to be calculated.

 

A resolution of 5760x1280 = 7.372.800 pixels

That is a whopping 38.5% more pixels...

And every pixel needs to be calculated.

 

You cannot compare this to someone using 1 monitor with a resolution of 1920 x 1280.

That are just 2.457.600 pixels.

 

 

The CPU, working in conjunction with software applications, sends information about the image to the graphics card. The graphics card decides how to use the pixels on the screen to create the image. It then sends that information to the monitor through a cable.

 

If the CPU gets overloaded with data the information sent about the image can be delayed or corrupt.

On the other hand if the GPU gets overloaded artifacts can occur, the inmage cannot be built fast enough or a black screen appears.

 

So it can be difficult to find out what the limiting factor is of your setup.

 

The higher the level of AA and Af , the more the graphcis card gets taxed.

You cannot compare this to someone using 1 monitor with a resolution of 1920 x 1280.

That are just 2.457.600 pixels.

 

Wy not? FSMark11 is almost CPU limited, if you put in a 690,680 or a 670 it be a diff of max 1-2fps

the runs ar at 1680x1050=1 764 000 pixels a get a 55.5 avg and with 300% more pixels a get avg 45.5fps about 21% lower fps

can it be with 5760x1080 417% more pixels get 30% lower avg fps that give 39fps ?

 

Thats on the CPU limited part and a get the same avg as a SB @4,6-4,8 with 580 on res 1680x1050 with 5040x1050.

Cant do much more on the CPU part max 2fps to 57avg

 

Hasse

Simply benchmark it at two different clock speeds. See how it scales. If going from say 3GHz to 5GHz scales 1:1 or close it's obviously CPU limited

<br />Simply benchmark it at two different clock speeds. See how it scales. If going from say 3GHz to 5GHz scales 1:1 or close it's obviously CPU limited <br />

 

a done that and its almost CPU limited.

The point is if you shall run FSX with multiple monitors you cant only change the GPU card.

I have seen to much of that, You MUST have the raw CPU power also.

IF the system only can do 40-45fps avg in FSMark11 you can not change to a 680 690 from a 570 and think now

we ar ready for multimonitor setup and be dissiponted because it only deliver an avg 28fps and min less than 18.

 

To test you must run both FSMark11 and Dazz GPUMark to get a hint how it gone work.

not a subjective a have 70fps on fl350.

Yeah, I guess those massive resolutions add a lot of extra strain on the CPU. Probably every effect is magnified, increasing AA is going to quickly get to the point where the GPU chokes too, probably 2xSGSS is already too much for a GTX680? not sure.

 

Our good old FSX will always be a resource hog, even more so if you throw a gazillion pixels at it

Yeah, I guess those massive resolutions add a lot of extra strain on the CPU. Probably every effect is magnified, increasing AA is going to quickly get to the point where the GPU chokes too, probably 2xSGSS is already too much for a GTX680? not sure.

 

Our good old FSX will always be a resource hog, even more so if you throw a gazillion pixels at it

 

Thats why I only use 4xS on my system and not 8xS and 2x Sparse Grid SS.

Using 8xS with 4x Spoarse Grid SS is just too much with detailed scenery.

 

However I am curious if the GTX690 will make a difference on my system.

<br />However I am curious if the GTX690 will make a difference on my system. <br />

 

Me too, have you test SLI ? i have a second option one more 680 if it not work a can sell one card

@westman,

Could you make it clear that you are running the 3 monitors in single, very widescreen view (maybe +-45 degrees of field of view), NvSurround mode - as opposed to three separate views? While this goes as fast as possible (used to do this with a Matrox TH2GO), the distortion on the side views is considerable.

 

whitav8

I run the FSMark on NV Surround , but in my sim i run three separate views left -53 0 53 total FOV 160 degrees thats harder than Nvsurround but no fisheye.

 

The setup is like wideveiw

 

 

Me too, have you test SLI ? i have a second option one more 680 if it not work a can sell one card

 

Already the seller and agreed on the price, but from that moment on I have nog heard from him...

 

Hi

A did a FSMark11 test with 3 monitor setup res 5040x1050.

my goal is to run 3 monitors with qute high settings, 1920x1080 or 2560x1440 monitors.

one thing is the GPU limiting and then CPU limithing factors.

Saab340 , Dazz,Tech or Word Not Allowed can explain my eng is so bad.

 

1920x1080 shall work but but would 2560x1440 work can the GPU be the limit, or both

on 1920x1080 can be it the GPU or?

 

Kind Regards/ Hasse

 

Settings by the book , res: 5040x1050

Hasse, what happens if you downclock your GTX680 at that high pixelcount screen resolution? You can normally downclock to 75% of stock speed. How much FPS drop does that yield in FSXmark11?

Lars did a fast test.

 

CPU @5150 GPU 1110 45.0 avg GPU 660 43.0 avg

CPU @4635 " 39.0 avg " 38.5 avg

CPU @4120 " 36.0 avg " 35.9 avg

Lars did a fast test.

 

CPU @5150 GPU 1110 45.0 avg GPU 660 43.0 avg

CPU @4635 " 39.0 avg " 38.5 avg

CPU @4120 " 36.0 avg " 35.9 avg

Thanks a lot Hasse,

So the ROPs on the GTX 480 are obviously enough for FSX with 5M pixel count with a maximum clocked IvyBridge without any additional AA.

 

So the lowered FPS by going 3x1680x1050 is purely down to the CPU.

And given that you have tested the GTX680 @ 60% speed and we still only possibly see a slight effect at 45FPS we can assume that with 3x1920x1080 it will still be your CPU holding you back, and possible even at 3x2560x1440. That is without additional AA. I guess even the GTX680 would start struggeling with a bit of SuperSampling of 11Mpixels.

 

Never the less, the GTX670 all of a sudden looks less interesting compared to a GTX660ti in FSX if you are on a single 1920x1200 display (or less). The ROPs are clearly enough for double that pixel count.

 

Thanks once again

That is very interesting.

 

That would mean that the amount of Shaders is the most important here and not the clock speed.

That is also confirmed in SLI with external view. The FPS are allmost doubled then ( with double the amount of Shaders )