31 Mar 2014

 

In this article I take a look at the relative performance differences between the Nvidia 670 FTW, 780 SuperClock and the 780 ti as it pertains to Flight Simulator X (FSX).  I'll also include some non-fsx benchmarks for comparing how these three cards stack up.  This project began as an attempt to get a boost in performance in FSX from my original card which was the 670.  In the end I settled on the 780ti.

To set a baseline for the test results there needs to be an understanding of the system I am testing with.  At the time I did these tests my system was as below:

 

 

                     Test Machine Specs and Software Used

Motherboard

Asus P8P67 Deluxe Rev 3.0

CPU

I7-2600K 3.4 GHz at 4.6 GHz with offset +0.045

Memory

16GB of 4x4GB 1600 MHz Ripjaws F3-12800CL9S-4GBRL Cas9

Hard drive

OS/FSX: Sata III 7200 RPM 64MB (FSX on its own drive)

Cooling

Corsair H70

Case

Vantec VA8000BWS

PSU

Silverstone OP1000-E 1000watt

OS

Windows 8 x64

Flight Sim

Flight Simulator X with acceleration

FSX Benchmark

FRAPS

3D benchmarks

3Dmark Vantage and 3Dmark11

Here is a comparison of the specs of each card before we dig into the specifics of the performance of each card.

 

EVGA GTX670 FTW (02G-P4-2678-KR)

EVGA GTX780 FTW (03G-P4-3784-KR)

EVGA GTX780ti SC (03G-P4-2883-KR)

Core Clock

1006 MHz

980 MHz

980 MHz

Boost Clock

1084 MHz

1033 MHz

1046 MHz

CUDA Cores

1344

2304

2880

Effective Memory Clock

6208 MHz

6008 MHz

7000 MHz

Memory Size

2GB

3GB

3GB

Memory Interface

256-bit

384-bit

384-bit

DirectX Support

11.0

11.1

11.1

Ports

1xHDMI; 2x DP;    
1 x DVI-D; 1x DVI-I

1xHDMI; 2x DP;
1 x DVI-D; 1x DVI-I

1xHDMI; 2x DP;
1 x DVI-D; 1x DVI-I

Max Displays

4

4

4

PSU Required

500 Watt

600 Watt

600 Watt

Card Dimensions

10” x 4.38”

10.5” x 4.376”

10.5” x 4.38”

Lets begin by taking a look at each card and their corresponding synthetic benchmark results.

The EVGA GTX670 FTW (02G-P4-2678-KR)

This particular GTX 670 was chosen because it had solid performance, was quiet and was not a reference card.  Reference cards tend to have the minimum specs for the class they come from.  This card also uses the same PCB design as the 680 class card.

 

 

Synthetic Benchmark Results

 

3DMark11 Performance Test Results

3Dmark Vantage Test Results

 

 The EVGA GTX780 FTW (03G-P4-3784-KR)

 Next up we jump into the 780 tier of cards and see how things stack up in this area.

 

 Synthetic Benchmark Results

 

 

 

 


The EVGA GTX780ti SC (03G-P4-2883-KR)

 

 

 

 

 

As you can see in the image above, the P8P67 Deluxe sports Pcie v2.0 x16 slots, not v3.0, later on ill show an example as to how much this makes a difference to synthetic tests such as these, though largely there won't be a huge difference, as its difficult even with the 780ti to saturate the pci express lanes.

Here is a summary of the 3D scores.

 

3D Synthetic Benchmark Scores

 

3DMark11

3DMark Vantage

EVGA GTX670 FTW (02G-P4-2678-KR)

P9045

P39450

EVGA GTX780 ACX (03G-P4-2784-KR)

P12390

P49131

EVGA GTX780ti SC (03G-P4-2883-KR)

P12784

P51334

FSX Benchmark Testing

 To do the FSX benchmarking I have FRAPS set to benchmark on pressing the P key and for a 60 second duration for this first test.  I took the AA clouds test flight situation from Kosta's site, which you can find here.   Its a good way to show how strong a system you have, though maybe not fully accurate of the type of frame rate you will actually get in most cases in FSX.  View it as a gauge only.

These tests were all done with DX10 mode.  DX10 mode works very well through the use of guides such as this one as well as the DX10 Scenery Fixer tool by Steve Parsons.  I have done side by side comparisons and there is a performance jump using DX10.

In conjunction with the fixer, Nvidia Inspector is used to set the proper level of AA.  To avoid tons of shimmering, AA Transparency Sparse Grid Super Sampling is used.  Note that at very high resolution such as this, it can kill performance with clouds.  I generally run at 2x or 4x at most, even with my system, which as now been upgraded to a 4930k processor as well (a whole other story).

Here are the relevant settings for the inspector that I used when doing these tests:

Also important is Vsync, to prevent tearing and other issues.  Its not visible at the bottom of this image, but basically you would set this to half your refresh rate if your LCD were 60 Hz (so 1/2), then set FSX to be locked internally at 60 Hz.  This brings us to the relevant fsx.cfg settings which are listed below (of course the video card is different in each test):

[GRAPHICS]
AC_SELF_SHADOW=0
AIRCRAFT_REFLECTIONS=1
AIRCRAFT_SHADOWS=1
ALLOW_SHADER_30=1
COCKPIT_HIGH_LOD=1
D3D10=1
EFFECTS_QUALITY=2
ForceWindowedVSync=0
GROUND_SHADOWS=0
HIGHMEMFIX=1 
IMAGE_QUALITY=0
LANDING_LIGHTS=1
 
MultiSampleQuality=4
MultiSamplesPerPixel=4
 
NUM_LIGHTS=8
See_Self=1
TEXTURE_MAX_LOAD=2048
TEXTURE_QUALITY=3
SHADER_CACHE_PRIMED_10=1693500672
SHADER_CACHE_PRIMED=1693500672
 
[Display]
BLOOM_EFFECTS=1
SKINNED_ANIMATIONS=1
TEXTURE_BANDWIDTH_MULT=120
TextureMaxLoad=9
TransitionTime=4.000000
UPPER_FRAMERATE_LIMIT=30
WideViewAspect=True
 
[TERRAIN]
AUTOGEN_DENSITY=3
BATCH_AUTOGEN2=1
DETAIL_TEXTURE=1
LOD_RADIUS=7.500000
MESH_COMPLEXITY=90
MESH_RESOLUTION=24
//SWAP_WAIT_TIMEOUT=2  
//TERRAIN_MAX_AUTOGEN_BUILDINGS_PER_CELL=3000     //897
//TERRAIN_MAX_AUTOGEN_TREES_PER_CELL=4500  //1100
TEXTURE_RESOLUTION=29
WATER_EFFECTS=4
 
 
[SCENERY]
//ObjectsToBatchPerFrame=50 //new
DAWN_DUSK_SMOOTHING=1
IMAGE_COMPLEXITY=4
LENSFLARE=1
//MAX_ASYNC_BATCHING_JOBS=3
//SmallPartRejectRadius=1  //4 b4
 
[Main]
DisablePreload=1
FIBER_FRAME_TIME_FRACTION=0.15 //0.05  
 
[TrafficManager]
AIRPORT_SCENERY_DENSITY=1
AirlineDensity=0
FreewayDensity=7
GADensity=0
IFROnly=0
LeisureBoatsDensity=0
ShipsAndFerriesDensity=0
 
[Weather]
CLOUD_COVERAGE_DENSITY=8
CLOUD_DRAW_DISTANCE=3
DETAILED_CLOUDS=1
DisableTurbulence=0
DownloadWindsAloft=0
DynamicWeather=0
THERMAL_VISUALS=0
WindshieldPrecipitationEffects=1
I have a beefy CPU in either case, so I use bufferpools set to zero.  Also noting that the two entries for MultiSampleQuality and MultiSamplesPerPixel have to be set in sync with the NI settings used.  This is a much longer discussion than for this article, but basically the entry for MultiSamplesPerPixel should match the Transparency Supersampling entry, in this case it was 4 x Sparse Grid Super Sampling (4xSGSS), therefore "4", while the other entry matches the "Antialiasing Setting" of 4x in this case.

 

With my setup, such things as traffic did not have an impact on the frame rate.  7% vs 0% the difference was minimal.  With that, this brings us to the actual test results for each card.
 

FSX Cloud AA Flight Benchmarks

EVGA GTX670 FTW (02G-P4-2678-KR)

7.3 fps

EVGA GTX780 ACX (03G-P4-2784-KR)

9.25 fps

EVGA GTX780ti SC (03G-P4-2883-KR)

11.15 fps

I was able to eventually do some testing with P3D v2.0 and then P3D v2.1.  Since the settings don't exactly mirror that of FSX (you get a few extra bells and whistles), I'm leaving any results from that comparison out of this article.  I will say that for all things held as equal as possible.. i find that P3D v2.0 and even 2.1 (which 2.1 loses a few fps over 2.0), performs a bit better than FSX and is now my platform of choice.

Given the choice between these cards, I did end up sticking with the 780ti as I figured it would give me the best bang for the dollar in both Flight Sims and for PC gaming, particularly 3D PC gaming.


Use this QR code in a QR reader application on mobile to open quickly on a mobile device

blog comments powered by Disqus

Flight-Stew Now Live

Be sure to visit our newest site, Flight-Stew, now active!

Recently Added

Subscribe

Poll

Which format do you prefer for buying games on Xbox One and PS4?


Show Results

[forcestyle][nozoom][nocaption]

Donate

       Support Tech-Stew
   [nozoom]
        Make a Donation

Recent Comments