• Testing a Z77 Computer – Professionally Built for Flight Simulation


    Gaiiden

    Review by Doug Horton. As indicated in the previously published article, and provided again as a refresher, a friend had been away from flight simming for awhile, and when he regained interest in the hobby, he recognized that his computer was quite outdated, with several out of date components, as well as a few software issues, including:

    • Processor – dual core operating at 2.4 GHz
    • Motherboard – older processor socket, unable to host newer processors
    • Memory – older DDR2 operating at 800 MHz – about half speed of newer memory
    • Graphics card – five year old technology, limited to DirectX 10 and two monitors per card
    • Flight Simulator X installed with a very large number of add-on aircraft and scenery files – very slow to load
    • FSX configuration file tweaked with many questionable parameter changes
    • Older version of X-Plane installed, with interest in updating to version 10
    • Operating three monitors, connected to older graphics card with older technology

    Though I offered to recommend components and help assemble a new system, my friend decided to purchase a new computer from Jetline Systems, which offered a model with most of the components I recommended. Importantly, Jetline would also test the system and save BIOS profiles for default operation and overclocking the processor up to 4.5 GHz. This article reports on testing a sample Jetline computer that is identical to the computer purchased by my friend. You can read about the components in my previous article on AVSIM.

     

    Jetline Systems specializes in building, testing, and selling computers that are intended to provide high performance for flight simulation programs. Jetline was founded in late 2006, coincident with the release of Flight Simulator X, by veteran U.S. Air Force flight engineer Greg Sanderson. His vision was to create a high performance PC company that would directly support the flight simulation pilot community, with research and accumulated knowledge of what hardware is really best for “top of the line” flight simulation computers.

     

    Testing - Background

     

    Most computer flight simulator users have learned that framerate (frames displayed per second) is a very important concern when a new system is purchased, along with use of compatible state of the art components. My benchmarking test results, published in several articles over the past four years, verify that for a given set of components and program settings, processor core speed is the most significant factor in producing acceptable framerates. In particular, for given components and settings, framerate performance is directly proportional to core speed. Accordingly, many users overclock their processors to drive higher framerates, or perhaps operate flight simulator programs with higher display settings and acceptable framerates – more on this subject below.

     

    Benchmarking Framerate Performance

     

    I’ve been using Gary Dunne’s free FSXMark07 utility for benchmarking over the past four years. It’s simple, dependable, and results are repeatable. There’s a similar utility that updates Gary’s utility with a different test aircraft, test flight, and settings, but it does not credit Gary for his earlier work, and it biases results higher by setting airline and GA traffic to zero, which is unrealistic for most users, so I’m ignoring the newer utility. Also, I can track results back four years by staying with FSXMark07, and that’s important when tracking trends in performance. File FSXMark07.zip can be downloaded from the file library at www.Avsim.com. Follow the simple instructions to get started with the simple procedure. You may also wish to consult my article about benchmarking, which was recently published in ASVIM Online.

     

    T_BenchStart.jpg

    Single monitor view at beginning of FSXMark07 test flight

     

    T_BenchEnd.jpg

    Single monitor view after five minutes of FSXMark07 test flight

     

    Single Monitor Tests

     

    I began by testing the Jetline computer with a single monitor, a 40” LCD/LED TV at 1920 x 1080 resolution, with an EVGA GTX 660 Ti graphics card. I then set the motherboard BIOS to “Optimized Default.” I presumed this would allow the processor to run at its nominal speed of 3.5 GHz up to 3.9 GHz. Given the graphics load of FSX, I expected that the processor would actually run close to its highest Turbo Boost speed of 3.9 GHz. Subsequently, I overclocked the processor, using BIOS settings recommended by Jetline Systems, to run the processor at 4.2 GHz, and then 4.5 GHz. For the three selected settings, the results were:

     

    BIOS Setting FPS
    “Optimized Default” 45.8
    Overclock to 4.2 GHz 54.0
    Overclock to 4.5 GHz 57.3

     

    The results of these three tests provided surprising results! By visual inspection, the results indicate that the framerate for the Optimized Default setting is much less than proportional for the presumed speed of 3.9 GHz. This suggests that the processor is not achieving 3.9 GHz for the five-minute test flight. To verify this conclusion, I added a fourth test; namely, modifying the overclocking settings to set the Turbo Boost for 3.9 GHz. At this point, the results, sorted by framerate, were:

     

    BIOS Setting FPS
    “Optimized Default” 45.8
    Overclock to 3.9 GHz 51.5
    Overclock to 4.2 GHz 54.0
    Overclock to 4.5 GHz 57.3

     

    Next, I graphed the results, using linear regression analysis to calculate the effective core speed for the Optimized Default condition, which was about 3.3 GHz, as shown in accompanying graph. The graph includes data labels, and in the upper right corner, it shows the derived regression equation, which I used to calculate the core speed. It also includes the value of R2 – a measure of “goodness of fit” of the line – with 1.0 being a perfect fit. Clearly the regression line is a nearly perfect fit.


     

    T_FPS-Test1.jpg

    Spreadsheet graph of framerate vs. processor core speed,

     

    With the Optimized Default BIOS settings, I presumed that the processor would operate close to its specified top speed of 3.9 GHz, but instead, it operated at about 3.3 GHz. What was happening? In consulting with Jetline Systems, and using the free GPU-Z utility to peek at the instantaneous core speeds during the FSXMark07 test flights, we concluded that there are probably two factors at work:

    • The Default Optimized BIOS settings, in concert with Windows 7 settings, also include power-saving settings, which allow the core speed to vary between about 1.6 and 3.9 GHz, depending on the processor load, instead of 3.5 to 3.9 GHz.
    • The graphics load during the five-minute FSXMark07 flight varies considerably, and it’s insufficient to keep the processor operating at 3.9 GHz.

    Alternatively, using BIOS settings to specify (indeed, force) an overclocked processor clock speed of 3.9 GHz, with a Windows 7 power setting for Performance, kept the processor at this speed, and it increased FSXMark07 framerate test results by slightly over 12%.

     

    Testing with Three Monitors

     

    Using three 1920 x 1080 LCD/LED TVs as monitors, I set up Windows 7 for three monitors and then set up NVIDIA 2D Surround with the GTX 660 Ti graphics card. This provided a total resolution of 5760 x 1080. I then repeated the tests for the overclocked processor speed of 4.5 GHz. The result was that the average framerate, for five FSXMark07 trials, decreased from an average of 57.3 to an average of 46.5. That’s not bad, considering that the pixel rendering load increased by a factor of three, to produce a display resolution of 5760 x 1080.

     

    For those who have not yet run FSXMark07 benchmarking, here are screenshots of the beginning and ending virtual cockpit view. Note that zoom is set for about 0.70 for correct perspective on three monitors at 45 degree angles.

     

    T_ThreeMonStart.jpg

    Three monitor view at beginning of FSXMark07 test flight

     

    T_ThreeMonEnd.jpg

    Three monitor view after five minutes of FSXMark07 test flight

     

    A framerate of 46.5 for the FSXMark07 Global High configuration is very playable, except with shadows enabled and with high AI traffic settings, particularly near airports with heavy scenery loads. This suggests the question: “What display configuration settings could be changed to increase the framerate with three monitors?” There are many combinations of settings that could be tried, and I made a few changes that I knew would increase the framerate, while not making much difference in the virtual cockpit view. Here are the settings I changed from the FSXMark07 Global High configuration, using the Options, Settings, Display interface, noting that two of the changes were made to enhance the simulation experience:

    • Graphics Tab: Turn off lens flare.
    • Aircraft Tab: Turn off aircraft shadows on ground and on itself.
    • Scenery Tab: Set water to Low 2.x, and move Scenery Complexity slider from Dense to Very Dense – enhancing the scenery display.
    • Weather Tab: Turn on Download Winds Aloft – to increase realism for other flights, but  has no effect on framerate performance because the FSXMark07 test flight uses preset static weather
    • Traffic Tab: Change AI traffic sliders for Ships and Ferries, and Leisure Boats, from 35% to 15%.

    The result is quite surprising! Making these simple changes resulted in the FSXMark07 framerate benchmarking test, with three monitors, increasing from a framerate 46.5 to 71.8! This is a profound result, in that a few simple display configuration changes, two of which improve the simulation experience, could be used to increase the framerate by 54%.

     

    Performance with GTX 680 Graphics Card

     

    For those who are inclined to presume that a newer, more expensive graphics card will increase their framerate performance, be aware that many reviewers have concluded that FSX is “CPU-bound,” with the more robust and expensive graphics cards not making much difference in framerate performance. To confirm this again, I repeated the three monitor tests with an EVGA 680 card in place of the EVGA GTX 660 Ti. Not surprisingly, with the same configuration settings, there was no difference in FSXMark07 framerate test results. Clearly, investment in processor core speed is the most important consideration in hardware selection, including modest overclocking of the processor, though, as shown above, a few changes in display configuration settings can have much greater results than overclocking!

     

    T_EVGA680.JPG

    Testing of the EVGA GTX 680 graphics card, in place of the EVGA GTX 660 Ti, indicated that with three monitors, there framerate performance was identical with the two cards.

     

    Overclocking Settings

     

    For the ASUS motherboard model P8Z77-V Pro, here are the BIOS settings recommended by Jetline Systems, for overclocking the i7-3770K processor to 4.5 GHz:

    • During computer startup (POST), press F2 or DEL to display the UEFI BIOS setting interface.
    • To ensure a consistent starting point, press the F5 function key to invoke the Optimized Default BIOS settings.
    • Press the F7 key to change to Advanced settings.
    • Use the left and right keyboard arrows to select the AI Tweaker tab, and make the following changes:
    1. AI Overclock Tuner: set to X.M.P.
    2. Asus MultiCore Enhancement: set to Disabled (see below).
    3. 1-Core Ratio Limit: set to 45 (4.5GHz).
    4. CPU Offset Voltage: set to +0.065
    5. DRAM Voltage: will be automatically set by X.M.P profile.
    • Now Select the DIGI+ Power Control and make the following changes:
    1. CPU Load-Line Calibration: set to Medium.
    2. CPU Power Phase Control: set to Optimized.

    Note: as related to the recommended settings, I chose to explore two settings further:

    • Why should MultiCore Enhancement be disabled, as it sounds like a desirable setting? It’s not clear what this setting is supposed to do, but with it enabled, FSXMark07 results for five trials were inconsistent. With this setting was disabled, results were more consistent, so disabling is indeed recommended.
    • Jetline had suggested disabling Hyper-threading, which limits the processor to using its four physical cores and disables its four virtual cores. Yes, FSX only uses the four physical cores, but X-Plane 10 uses all available cores, and the forthcoming Prepar3D version 2.0 might also use more than four cores. Benchmarking tests with Hyper-threading enabled or disabled resulted in no difference in FSX framerate, so I leave this setting enabled.

    Summary

     

    Here are key conclusions from testing Jetline’s sample flight simulation computer with FSX:

    • The Optimized Default BIOS settings were not optimal for FSX framerate performance with the tested motherboard and processor. Setting the processor to its upper specified clock speed, without overclocking, produced higher framerates than the Optimized Default BIOS settings.
    • It was relatively simple to connect three monitors to either of two NVIDIA 600 series graphics cards, using the Windows 7 multi-monitor interface and the NVIDIA 2D Surround interface.
    • Operating with three monitors, instead of one, reduced the framerate, but less than proportionately to the increased total resolution.
    • With or without overclocking processor speed, and with one or three monitors, reducing a few display settings provided a significant increase in framerate performance – more so than overclocking.
    • Changing to a more robust, higher cost graphics card did not increase framerate performance with FSX. On the other hand, newer graphics cards such as the NVIDIA 600 (and now 700) series allow connection of three or four monitors without the previous need for two or more cards in SLI or Crossfire configurations, or the need for external adapters.

    Overall, test results for the selected hardware are excellent, and Jetline has done a great job with component selections for this mid-priced computer. At time of writing, Jetline was conducting testing in preparation for moving from a Z77 board, i7-3770K processor, and choice of GTX 600 series graphics cards, to a newer generation Z87 board, i7-4770K processor, and choice of GTX 700 series cards. I’m expecting to review their test data, and we might be reconfiguring the sample computer used for this review, to match the new recommendations, in which case I’ll expect to repeat some of the above testing for a future review.

     

    Reviewer’s Note: Jetline Systems is a supporting advertiser of AVSIM Online and provided the sample computer system that was tested for this article. The author has no financial interest in Jetline Systems.



      Report Article


    User Feedback


    There are no comments to display.



    Create an account or sign in to comment

    You need to be a member in order to leave a comment

    Create an account

    Sign up for a new account in our community. It's easy!

    Register a new account

    Sign in

    Already have an account? Sign in here.

    Sign In Now