:: Back to news index ::

MKZ GPU PhysX Benchmark version 2.2

with 16 comments

Our readers may remember Metal Knight Zero Online (MKZ – chinese online-FPS title by Object Software) benchmark, released in mid 2008 as part of GeForce PowerPack #1.

Now, new and enhanced MKZ Benchmark version 2.2, specially developed for Fermi GPUs launch in China, is available.

Compared to previous version, graphics is looking much nicer, game physics (destructible rigid body objects, particles, tearable cloth) is looking more solid and accurate.

MKZ Benchmark 2.2 download is available here (1.3 Gb)

Important note:

For English version you’ll need to select first line designated as ?? during installation.

Configuring and launching the game may be a little tricky.

Benchmark settings can be altered via MKZSettings.exe file, located in \Object Software\MKZBenchmark\system folder.

Now, to launch benchmark with saved settings, you’ll need to create shortcut for MetalKnightZeroClient.exe file (same folder) with following parameters “-openautomate -debug

You can use one of the shortcuts created automatically during installation, but don’t forget to delete “-RESOLUTION” parameter, to prevent graphics resolution settings override.

P.S. MKZ is now in closed beta state. Open beta will be available later this summer.

P.P.S. Don’t forget to check other recent PhysX benchmark – JX3

Again, thanks for Stefan for some hints

Written by Zogrim

April 28th, 2010 at 7:26 pm

Posted in PhysX Games

Tagged with ,

16 Responses to 'MKZ GPU PhysX Benchmark version 2.2'

Subscribe to comments with RSS

  1. “specially developed for Fermi GPUs”

    Does it mean that it’s not usable for other NV GPUs?



    29 Apr 10 at 9:01 am

  2. mareknr
    “specially developed for Fermi GPUs launch in China

    It’s running fine on my GTX275



    29 Apr 10 at 9:32 am

  3. I just ran this on my hybrid setup. I have two 5850’s in CrossfireX and a GTS 250 dedicated PhysX. Everything ran very smoothly. I ran it at 1920 x 1080 with everything maxed out to High. I was getting FPS: AVG 107, MAX 153, MIN 34. Hardware PhysX was working.


    xDee xDee

    29 Apr 10 at 9:34 am

  4. OK. Thanks. Do you know how it looks like with new FluidMark?



    29 Apr 10 at 10:35 am

  5. And anybody know if it’s possible to set secondary NV GPU for computing CUDA effect in Just Cause 2? I tried to find something but without any result…



    29 Apr 10 at 10:41 am

  6. Just Cause 2 doesn’t use the PhysX API. It uses “Havok” that runs in software mode.



    29 Apr 10 at 2:20 pm

  7. Johnny
    Just Cause 2 doesn’t use the PhysX API.
    mareknr was talking about special CUDA based graphical effects, like Bokeh filter and enhanced water surface.

    And anybody know if it’s possible to set secondary NV GPU for computing CUDA effect in Just Cause 2?
    Never saw any working solution



    29 Apr 10 at 2:46 pm

  8. I’m using a 5970 + 8800GT (G92, 112 SP) PhysX dedicated and the MKZ Benchmark 2.2 ran without problems with High settings @1920*1080. Also, have a i7-920 C0 @ 4.00Ghz
    My system’s results are:
    Avg: 74; Min: 43 Max: 118

    I can’t help to compare the results with xDee xDee’s setup.

    Considering that as far as the performance is concerned, 2×5850 are roughly a 5970 @ stock clocks, in this case, are the 128 SP from the GTS250 that make the difference; a tangible difference both in Max. and Avg. FPS, to be fair.
    The 43 FPS vs the 34 FPS are a curious Min. score, though.

    I know, these results are quite “relative” because, the systems are different but, I thought it was pertinent to mention this facts.



    29 Apr 10 at 3:10 pm

  9. Just Cause 2 uses CUDA for doing the water surface simulation and DoF (called Bokeh from japanese).
    Tecnically yes, CUDA enumerates the device, so you can select wich you want to execute the program, depends on developer to expose this option on the user interface.

    You can force programs to use a particular GPU by setting the Compute Mode of the other devices in “prohibited” mode (no compute programs may run on this GPU), on Linux there’s a tool called nvidia-smi for doing this but I don’t know if there’s a comparable tool for windows



    29 Apr 10 at 5:37 pm

  10. Johnny

    Here are my system specs: Asus Rampage II Gene, i7 930 @ 4.41GHz, 6GB G. Skill @ 1,754MHz 6-7-6-18-60-1T, 2 x Sapphire 5850’s @ 945/1220, GTS 250 @ 800/1915/1200, 6 x Samsung SpinPoint F3 1TB RAID 0 short stroked to 300GB volume, Corsair HX1000, Windows 7 Ultimate x64.

    The GTS 250 is running on the x4 PCI-E slot. I had to buy a 12″ ribbon PCI-E extender. And yes x4 is enough bandwidth for dedicated PhysX.

    My minimum score is rather low, but then again it can be caused by anything even a hiccup in the system maybe another program accessing the CPU or something. The minimum fps is just the lowest it hit ever during the benchmark as is the max fps. So really the min and max don’t really mean a damn thing.


    xDee xDee

    30 Apr 10 at 4:40 am

  11. Oh and my whole system is being cooled by a complete Koolance custom water cooling set up in 2 loops.


    xDee xDee

    30 Apr 10 at 4:41 am

  12. Not the same numbers I hit earlier, but you get the point.

    Link to pic of MKZ Bench 1920×1080 High:×1080High.jpg


    xDee xDee

    30 Apr 10 at 10:17 am

  13. Stefen: Thanks for advise. I’ll try to find some configuration tool, like you are talking about, under windows.



    30 Apr 10 at 11:36 am

  14. xDee xDee,

    Thanks for the input. Well, let’s just say our systems are pretty much alike as far as main components are concerned. I do have 6GB DDR3 Corsair.
    Just that I have EKWB as complete water solution instead of the beautiful Koolance waterblocks; in my Gigabyte X58-EXTREME is the EK X58 kit, the EK Supreme 1366 cools down the CPU and just installed the EK FC-5970 to cool the HD5970. Also, have 3x HDD: 2x WD 500GB RAID0 + F3 1TB for storage, and a Corsair TX 750W PSU.

    Indeed, I agree that the minimum FPS might very well be an “hiccup” or is related with the benchmark loading process (i.e. I get Min. 0 FPS everytime I ran a single performance test. Need to run “Loop” to escape this).
    You should run the Loop test to see if te min fps are more accurate.
    The main conclusion I take from our tests is that 128 SP in fact, does perform a lot better than 112 SP, at least it does more than I expected in this benchmark.
    I’m only in testing process and haven’t yet, decide what GFX I shall get to dedicate to PhysX.
    I’ve heard all around the net that a GT240 with 96 SP is enough for dedicated PhysX but, I’m realy not convinced that it get the job well done in a game like Metro 2003 maxed out, AFx16, 4xMSAA for example. Even with the 8800GT, I had FRAP’s scores of Min: 15, MAX 60 and AVG, 41 FPS



    1 May 10 at 9:44 pm

  15. Sorry I’m not going to be able to do anything on that comp for about a week or so. My cat knocked over my water cooling reservoirs and radiators and it cause a huge mess which sprayed some coolant on my motherboard causing a short. Luckily my motherboard is under warranty and Asus will be shipping me a new motherboard this week. My i7 930 may have been damaged too though. Hopefully not, but it is also still under warranty of course. I have a i7 920 as backup though.


    xDee xDee

    4 May 10 at 7:24 am

  16. Ok so I got my new motherboard back but it turns out my i7 930 is dead also. But I’ve contacted Intel already and they are sending me a new one. I surprised about how well Asus and Intel have treated me. Good customer service for once.


    xDee xDee

    6 May 10 at 9:57 pm

Leave a Reply

Copyright © 2009-2014. | About project | Privacy Policy
PhysX is trademark of NVIDIA Corporation