ATI HD4870 vs X1900. What the????

Discussion in 'Mac Pro' started by stefanski, Jun 23, 2009.

  1. stefanski macrumors member

    stefanski

    Joined:
    Apr 11, 2004
    Location:
    Australia
    #1
    I just got myself a HD4870 1GB, and replaced my ageing X1900. All is well. However, if I run XBENCH, I get some strange results. The X1900 is just as fast and beats the 4870 in OpenGL! How come? What's happening?

    I know XBENCH is supposed to be a bit old but shouldn't the results still be better for the 4870? I have other benchmarks to run but if there is somebody on here who's done the same thing, could you let me know your results?

    Ta


    This is the result for the 4870:

    Quartz Graphics Test 239.08
    Line 205.75 13.70 Klines/sec [50% alpha]
    Rectangle 282.98 84.48 Krects/sec [50% alpha]
    Circle 232.73 18.97 Kcircles/sec [50% alpha]
    Bezier 225.03 5.68 Kbeziers/sec [50% alpha]
    Text 264.62 16.55 Kchars/sec
    OpenGL Graphics Test 217.34
    Spinning Squares 217.34 275.70 frames/sec

    User Interface Test 401.98
    Elements 401.98 1.84 Krefresh/sec

    Here are the results for the X1900:

    Quartz Graphics Test 238.42
    Line 206.76 13.77 Klines/sec [50% alpha]
    Rectangle 282.40 84.31 Krects/sec [50% alpha]
    Circle 230.17 18.76 Kcircles/sec [50% alpha]
    Bezier 225.70 5.69 Kbeziers/sec [50% alpha]
    Text 261.88 16.38 Kchars/sec
    OpenGL Graphics Test 278.74
    Spinning Squares 278.74 353.60 frames/sec

    User Interface Test 413.15
    Elements 413.15 1.90 Krefresh/sec
     
  2. jeanlain macrumors 6502a

    Joined:
    Mar 14, 2009
    #2
    Xbench is no good at testing graphics cards. Use a recent game.
     
  3. Andrew Henry macrumors 6502a

    Joined:
    Mar 4, 2008
    #3
    Real world results will be no where near that. The 4870 will stomp the x1900 into the ground.
     
  4. stefanski thread starter macrumors member

    stefanski

    Joined:
    Apr 11, 2004
    Location:
    Australia
    #5
    Yes, yes... As I said, I know that xbench isn't the best choice. I am going to test with different benchmarks tonight but it means I have to put the x1900 back in to run those tests on the old card as well. Not something I'm looking forward to but for the sake of knowing the difference, I'll bite the bullet. Should make for an interesting blog.

    >The 4870 will stomp the x1900 into the ground.

    That's what I would have said, too. But that was before I ran xbench. :D
     
  5. MacVidCards Suspended

    Joined:
    Nov 17, 2008
    Location:
    Hollywood, CA
    #6
    Xbench is a joke.


    Xbench tested a GF2 versus a Radeon 9800 years ago. It showed the GF2 to be "superior"

    Never used it again.

    Quit wasting your time.
     
  6. stefanski thread starter macrumors member

    stefanski

    Joined:
    Apr 11, 2004
    Location:
    Australia
    #7
    Well, I downloaded Cinebench 10.

    I am too much of a lazy bum to reinstall my X1900. So I scanned the net for test results and found a guy who has the same setup as I do.

    The results for the X1900 are below:

    ****************************************************

    Rendering (Single CPU): 3171 CB-CPU
    Rendering (Multiple CPU): 18242 CB-CPU

    Multiprocessor Speedup: 5.75

    Shading (OpenGL Standard) : 4819 CB-GFX

    ****************************************************

    Then I did the same for the HD4870:

    ****************************************************

    Rendering (Single CPU): 3233 CB-CPU
    Rendering (Multiple CPU): 18673 CB-CPU

    Multiprocessor Speedup: 5.77

    Shading (OpenGL Standard) : 5579 CB-GFX

    ****************************************************

    My results for the 4870 seem to be in line with everybody else who's running it on a 1st Gen Mac Pro.

    As far as I can see, the 4870 is only 2-3% faster in rendering but about 16% in the OpenGL test which is very good. But I have seen results where the OpenGL result is above 5000 for the X1900 so the difference is smaller.

    I say there is a lot of room left to improve the 4870 drivers further and I'm very much interested in the results on Snow Leopard. At least the 4870 will be fully supported while the X1900 will miss out on the features.

    Comments?
     
  7. Freyqq macrumors 68040

    Joined:
    Dec 13, 2004
    #8
    try a real graphics benchmark like 3dmark06. The difference should be enormous...like 2-3x faster.
     
  8. mattical69 macrumors newbie

    Joined:
    Jul 14, 2009
    #9
    Hey guys, read your post and thought you might be able to help. Do any of you have an untouched ATI X1900 XT 512MB that came OEM from Apple w/ your Mac Pro 1,1??? If so I need your firmware!!!

    Background:
    I need to unbrick my :apple: OEM Apple ATI X1900 XT 512MB from my Mac Pro 1,1 3.0GHz BTO. I've already upgraded to the XFX AIT HD4870 1GB using the atiflash firmware update method also on this forum. However, in doing my flash of the 4870 I mistakenly used the PCIe ID of the X1900 XT during the flash process and killed the OEM Apple firmware. I thought I had backed up both BIOS/firmware ROMs from both cards, but evidently I didn't.

    So, if you guys have the OEM Apple ATI X1900 XT 512MB card untouch from the Apple firmware I'd love some help getting the ROM file.

    Thanks in advance for your help/reply. :)

    Cheers!
     
  9. mattical69 macrumors newbie

    Joined:
    Jul 14, 2009
    #10
    UPDATE: OEM Apple Firmware FOUND!

    I also fished this out to a few old Apple colleagues who were able to send me the firmware in the last 24 hours. I've successfully unbricked my card and can now at least benchmark and do the work I need while still having my rockin' XFX HD-4870 for the other fun stuff.

    If anyone should require this firmware in the future just reply here or PM me. I'm keeping an archive of the firmware v140 and v202 from Apple. Now I'm curious to also run some benchmark tests on my config...
     
  10. macz1 macrumors 6502

    macz1

    Joined:
    Oct 28, 2007
    #11
    Why testing with benchmarks? Open a game, set everything to "very high" and enjoy! Just use the card!
    The performance difference (in windows at least) should be more than big enough such that you don't need any benchmarks to see that the new card is faster...
    Unless you really want to compare numbers...
     

Share This Page