Benchmarking FSX with Three Models of 500 Series Graphics Cards

Review by Doug Horton. Many users spend hard-earned money for new hardware in the vain hope that it will improve FSX framerate performance. Those readers who’ve read my recent article on AVSIM about framerate benchmarking understand how easy it is to run meaningful, statistically-based framerate benchmarks. If you haven’t read it, I recommend doing so before continuing with this article.

The series of tests described in this article is significant, because we were able to obtain three different models of NVIDIA based GTX 500 series graphics cards from EVGA and NVIDIA. The cards have differing specifications, such as clock speeds, number of cores, memory size, memory bandwidth, and price points. These cards have been succeeded by NVIDIA 600 series and recently, NVIDIA 700 series cards, but the information is still quite meaningful.

As an example, in my recently published article on testing a professionally built Z77 computer system, GTX 660 Ti and GTX 680 cards provided the same benchmarking performance.

We tested each card using the FSXMark07 framerate benchmarking procedure by flight simmer Gary Dunne, to assess how each affected FSX framerate performance. With this information, readers will understand that some hardware components improve FSX performance and some do not, sometimes in spite of high costs of various components.

Comparative Specifications

Here are comparative specifications for the three sample graphics cards, with the theoretically best value of each performance parameter in bold blue type:

* all cards are 4.376” high

** if registered within 30 days; additional warranty extension for fee

*** if registered within 30 days

**** from EVGA (rebate amount if offered)

Graphic Card Descriptions

EVGA GeForce GTX 560 FTW+ 1024 MB GDDR5

The “60” cards from the NVIDIA GTX 200, 400, and 500 series were designed to be low to middle price range products. The “FTW+” identifier indicates that this is not the base card in the 560 line, but rather, it is factory overclocked about 15%, compared to the NVIDIA entry-level reference model. For example, the FTW+’s core clock is 931 MHz, compared to the reference speed of 810 MHz, and the shader clock speed is 1862 MHz, compared to the reference speed of 1620 MHz.

With its siblings, this card supports the following key features: DirectX 11; and NVIDIA CUDA, PhysX, PureVideo, 2-way SLI, and 3D Vision Surround technologies. It will feed two monitors, with DVI and/or mini-HDMI connectors. With a second, identical card connected and enabled in SLI configuration, the two cards will feed three monitors. Note that from previous testing, FSX does not benefit in performance from two cards in NVIDIA SLI or AMD Crossfire configuration.

EVGA GTX 560 FTW+ graphics card and packaging

Edge connectors of EVGA GTX 560 FTW+ graphics card

EVGA GeForce GTX 560 Ti 448 Cores FTW 1280 MB GDDR5

With this card, which was released in late 2011, more recently than the other models tested for this article, EVGA has boosted the basic GTX 560 card from 336 to 448 processing cores, upgraded the memory to 1.2 GB, and they’ve added full support for NVIDIA 3-Way SLI, and as suggested in the product press release, “making an excellent card even better.” EVGA also provides very flexible connector options on this card, with one each full-size HDMI and DisplayPort connectors, as well as two DVI-I connectors, though as with its siblings, one card supports one or two monitors.

Interestingly, this card has lower core, memory, and shader clock speeds than the GTX 560 FTW+ card described above, but with the increased number of cores. We’ll see from testing if this makes a difference in benchmarking. I like the full size connector layout, in comparison to the other two sample cards, because users who connect to their monitors with DisplayPort or HDMI connectors, can use standard cables, rather than purchasing cables with mini connectors on one end and regular size connectors on the other end.

EVGA GTX 560 Ti 448 graphics card and packaging

Edge connectors of EVGA GTX 560 Ti 448 graphics card

EVGA GeForce GTX 580 1536 MB GDDR5

This high-end graphics card was released in late 2010, and at that time, it was claimed to be the fastest graphics processing unit (GPU) in the market. It includes 512 cores and 1.5 GB memory. EVGA produces several versions of this card, including cards with factory overclocking and up to 3.0 GB memory. At the high priced end of the product line, EVGA also offers several models that are set up for liquid cooling. Similar to the first card above, this card provides one mini HDMI and two DVI connectors, which are capable of feeding two monitors.

EVGA GTX 580 graphics card and packaging

Edge connectors of EVGA GTX 580 graphics card

Test Results

Though the three sample cards have variations of clock speeds, numbers of cores, and other specifications, benchmark testing with “big picture” tests showed that they perform in proportion to their prices. I think that’s good news, as no one would like to have lower performance for more money. As with previous benchmark testing of FSX with the framerate benchmark FSXMark07 standardized configuration settings and sample five-minute flight, there was no statistically significant difference in the performance of FSX with the three cards. As testing before has verified, FSX framerate performance in standardized testing is almost entirely dependent on processor core speed, assuming at least four cores are available. On the other hand, there is increased performance for the more expensive cards, when tested to common benchmarking programs.

Heaven Benchmark

According to its developer UNIGINE Corp, “Heaven Benchmark is a DirectX 11 GPU benchmark based on advanced UNIGINE™ engine. It reveals the enchanting magic of floating islands with a tiny village hidden in the cloudy skies. Interactive mode provides emerging experience of exploring the intricate world of steampunk.” UNIGINE also claims that Heaven Benchmark is the first DirectX 11 benchmark in the world, with the original version released on the date of the Microsoft Windows 7 launch in October 2009.

User interface for Heaven Benchmark 4.0

The free version of the Heaven Benchmark, currently version 4.0 Basic Edition, is available at www.unigine.com). It’s a visual treat to run, taking the viewer through an amazing tour of a floating-in-air “steampunk” village, with accompanying music. Check out the meaning of “steampunk” at Wikipedia.org.

Screenshot of steampunk airship from Heaven Benchmark

It’s possible to run Heaven Benchmark hands-off, with or without benchmarking enabled by the F9 key, letting the program run, while you admire the steampunk scenery. Also, you can toggle the F3 key, to intermittently turn DirectX 11 tessellation off and on. This action allows you to see the remarkable difference in the textures of a cobblestone walk, stone wall, tile roof, dragon sculpture, and more, with or without tessellation. Heaven Benchmark produces an average framerate number and a numerical score, and in testing, both indices increased in approximate proportion to card prices.

3DMark^® 11

According to developer Futuremark, “3DMark 11 is the latest version of the world’s most popular benchmark for measuring the 3D graphics performance of gaming PCs. 3DMark 11 uses a native DirectX 11 engine designed to make extensive use of all the new features in DirectX 11, including tessellation, compute shaders, and multi-threading.” The program generates a Benchmark score and ratings for Graphics, Physics, and Combined performance. The three cards performed similarly for Physics, while the Benchmark score and ratings for Graphics and Combined performance increased in approximate proportion to prices of the cards.

Flight Simulator X and Prepar3D

We test FSX framerate performance with the simple FSXMark07 procedure developed by user Gary Dunne, as explained in my article referenced above. This simple procedure is available in file “FSXMark07.zip,” which can be downloaded from www.Avsim.com.

The FSXMark07 zip file contains simple instructions, and includes files for a saved autopilot flight in the default Bombardier CRJ 700 over the Seattle, Washington area of the U.S. Pacific Northwest. The filenames are FSXMark07.flt *.fssave, *.pln, and *.wx files. The flight is standardized by loading the included FSXMark07 - Global High.cfg file, which sets 38 parameters in the FSX.cfg file. I also manually set Anti-aliasing to On and Filtering to Anisotropic, for consistency with previous testing and for a better look.

Starting position of FSXMark07 benchmark flight

The FSXMark07 benchmark flight in the CRJ 700 begins in the air, paused, at 1500 feet, auto-throttle set for 250 knots, and autopilot set to navigate and hold altitude on a flight plan that loads with the flight. Framerate is tracked by the freeware FRAPS program, which is available at www.FRAPS.com. The flight is run for five minutes during which FRAPS counts frames and logs results in a separate file for each run.

After recommended settings are made in FRAPS, when the “P” key is pressed, the flight begins and FRAPS begins counting frames. After the flight has progressed over a great variety of scenery, FRAPS stops counting frames and writes a data file the minimum, maximum, and average framerates for the flight. Following a few more instructions is necessary but very simple. We usually repeat the procedure for a total of about 3-5 trials, and then calculate the combined average of the separate run averages. We’ve also found that if we run FSX in DirectX 10 Preview mode, the framerates are more consistent.

*Average of five FSXMark07 trials for each graphics card

Summary

As highlighted above, in spite of better results with industry benchmarks for graphics cards, the results from running the FSXMark07 benchmark with all three graphics cards with FSX confirmed what we have found before – no statistically different framerate performance for graphics cards produced since FSX was released in 2007. This confirms that framerate continues to be principally related to processor speed. This is bad news for those purchasing more expensive graphics cards that will not produce a higher framerate in FSX! The good news is that modestly priced graphics cards “get the job done” with FSX. On the other hand, those running multiple high resolution monitors may have improvements in some FSX characteristics.

Framerate test results for Prepar3D version 1.3 (1.4 is now current) were similar, with no difference in performance among the three tested graphics cards. Fortunately, forboth FSX and Prepar3D, satisfactory framerate performance can be achieved for most users by compromising on demanding display settings, such as water, shadows, 3D clouds, auto-generated scenery, and AI traffic; and/or purchasing higher speed processors or overclocking processors.