TechSpot

Experienced 3DMark users opinions needed

By red1776
Sep 30, 2009
  1. Hi gang,
    i will endeavor to give you the readers digest version here :) I happened to meet the local owner of a custom PC shop here, and he told me that he has many customers who want a machine to run intense games like crysis,stalker, etc but on a budget. i brought him over for several cocktails, and a go on my machine (crysis crusher). the upshot is that he was impressed to the point of commissioning me to build these things when the need arises
    I am putting together a benchmark/presentation on this(or very similar) configuration. I myself am only interested in frame-rates from actual games play and real world application, however i am aware of the popularity of the 3DMark suite(s). i have read varying opinions on these benchmarks ranging from "the best thing since sliced bread" to "worthless and designed to get you to upgrade" to "nothing to do with how your machine will perform gaming". i have not purchased 3Dmark 06, or Vantage (yet) and have used them in only their 'free trial' state. The results I got are a bit difficult to parse, but then i don't have unrestricted access not having bought the full version. my framerates in actual gaming are fantastic, but I'm not sure about the 3Dmark scores.
    I was hoping someone well versed in Futuremark products could tell me weather my scores are at all reflecting the performance of my machine or not. I have also noticed in sifting through scores that some on the list have things turned off, i assume that is to increase their score, adding to my inquiry about how accurate the scores are while i am running the tests with the presets on. I don't wish to be deceptive, but if they do not accurately reflect performance, I certainly don't want to use them.

    I attached a 06 screen-shot and Crysis Warhead frame rates can be seen here http://www.techspot.com/vb/attachment.php?attachmentid=51563&d=1250553089, and my specs are listed in the corner as usual.
    Thanks :)


    The scores:
    3DMark 06: 22186 (using 3x HD 4850's)
    3DMark 06: 20088 (using 3x HD 4830's)
    3DMark Vantage P 15186 (3x HD 4850's)
     

    Attached Files:

  2. klepto12

    klepto12 TechSpot Paladin Posts: 1,115   +9

    red i would not base 3d mark as a good gaming performace app i mean a core i7 920 at 3.8 with a 4870 x2 gets around 24,000 3d marks but you get higher fps in games that that setup does :)
     
  3. red1776

    red1776 Omnipotent Ruler of the Universe Topic Starter Posts: 5,219   +157

    Yeah thats what i have been seeing (I think) , its kind of a mess though when you start sifting through ORB. Thanks Klep :) i thought you might be the first to pony up an opinion! :)
     
  4. Adhmuz

    Adhmuz TechSpot Paladin Posts: 1,648   +521

    With my system I score P14156 in 3DMark Vantage (You know where to find specs) and I get no where near the FPS you get in Crysis at 1920x1080 somewhere around 25-35 maybe 30-40. The 3DMarks have become things people who want to set world records use because thats all they are good for.

    PS I really like how the "Common" system has an i7 920 and a GTX295
     
  5. red1776

    red1776 Omnipotent Ruler of the Universe Topic Starter Posts: 5,219   +157

    Thats a great point Ad, I think whats going on here is that way too much weight is put on the CPU, more so in Vantage. you can see the change buttons for the common system there. the i7 920 is the one that came up as default LOL, i did think it was interesting that I beat it by 3000+ points, but then in Vantage???? not a very linear test i think.
     
  6. Adhmuz

    Adhmuz TechSpot Paladin Posts: 1,648   +521

    Actually to come clean on that result, the reason I scored as high as I did is because of PhysX support. Thats the only place I see it do so much. With PhysX turned off my score is a good 1500-2000 points lower, but I never turn it off so it makes my score kinda realistic.
     
  7. klepto12

    klepto12 TechSpot Paladin Posts: 1,115   +9

    red im just saying that a stock 920 and even a gtx 295 cant compete with your setup so 3d mark is crap for gaming rigs imo it should be based off of real world tests only i mean my setup scores 14000 3d marks in 06 and that is 1 4850 but i only have a duel core if i had a quad my score would be around 17 to 18k just because of the cpu
     
  8. red1776

    red1776 Omnipotent Ruler of the Universe Topic Starter Posts: 5,219   +157

    I have doing some more looking into it, and I think your exactly right Klep, if I had this paired with an i7 my score would be through the roof....but my FPS would not be any better as many benchmarks attest to.
     
  9. Ritwik7

    Ritwik7 TechSpot Chancellor Posts: 1,672   +9

    I had been reading up on 3DMark06 before I sent you my score. It seems that one cannot significantly judge as to how the machine will perform during gaming just from these scores. Also, as Adhmuz mentions, nVidia cards get high scores due to PhysX support.

    LOL! And I was sad that I didn't beat it's score! :D
     
  10. Ritwik7

    Ritwik7 TechSpot Chancellor Posts: 1,672   +9

    Also, 3DMark06 tests lay emphasis on both GPU and CPU. However, 3DMark05 stresses the GPU more supposedly.
     
  11. klepto12

    klepto12 TechSpot Paladin Posts: 1,115   +9

    exactly red i think 3d mark is just a way for people to say look at e pen$$ if you know what im saying. i would rather have your rig that cost tons less than an i7 920 with a gtx 295 when your system kicks its but in gaming :)
     
  12. Adhmuz

    Adhmuz TechSpot Paladin Posts: 1,648   +521

    Okay, now that I'm on my Sig Rig I can go through my results since I first started out with a P4 and a single 8800GT. I'll list some short specs and results.

    Runs:
    P4 630 @ 3.0 and a single 8800GT: Score of P5653, GPU Score 5140, CPU Score 8071
    E6420 @ 3.2 and single GTX260: P12055, GPU Score 9988, CPU Score 31792 <- PhysX
    E6420 @ 3.36 and single GTX260 Small OC: P12734, GPU Score 10569, CPU Score 33035 <- PhysX
    Q6600 @ 2.4 and SLI 8800GT: P11284, GPU Score 9512, CPU Score 25563 <- PhysX
    Q6600 @ 2.7 and SLI 8800GT: P12718, GPU Score 10745, CPU Score 28310 <- PhysX
    Q6600 @ 3.0 and SLI 8800GT @ 702/1015/1702: P13527, GPU Score 11325, CPU Score 32424 <- PhysX (This was my best score before going i7)
    Q9550 @ 4.0 and single GTX260: P11092, GPU Score 10070, CPU Score 15946 <- PhysX OFF
    Q9550 @ 4.0 and single GTX260: P12447, GPU Score 10066, CPU Score 42865 <- PhysX
    i7 920 @ 4.0 and SLI 8800GT: P13812, GPU Score 11227, CPU Score 44650 <- Physx
    i7 920 @ 4.0 and SLI 8800GT: P12884, GPU Score 11172, CPU Score 23844 <- Physx OFF
    i7 920 @ 4.0 and SLI 8800GT @ 702/999/1728: P14156, GPU Score 11523, CPU Score 44995 <- PhysX (Current best run)

    GTX260 runs courtesy of Floofy_Fox

    Now thats my contribution of data it shows pretty well that 3DMark Vantage is not CPU orientated. Going from a Q6600 @ 3.0 GHz to an i7 920 @ 4.0 GHz gave me an increase of 629 points. If thats not enough of an indication I'm not sure what could be. Let me know what you think.
     
  13. red1776

    red1776 Omnipotent Ruler of the Universe Topic Starter Posts: 5,219   +157

    now im more confused lol, i have been looking at the scores on ORB and its not making much sense.
     
  14. Adhmuz

    Adhmuz TechSpot Paladin Posts: 1,648   +521

    Sorry about that, those are just my results for the past 3 motherboards, 2 CPUs and other RAM and GPU configurations.
     
  15. red1776

    red1776 Omnipotent Ruler of the Universe Topic Starter Posts: 5,219   +157

    no , not at all. i am comparing my 3DMark scores and frame rates, and trying to find some sense of lineage between them...maybe that's my mistake.
     
  16. klepto12

    klepto12 TechSpot Paladin Posts: 1,115   +9

    lol look at the score of the cpu though from the p4 to the quad there is a big difference that was what i was trying to say duel cores and tri cores will lower the overall score it is very cpu based unlike what you said otherwise it would score the same wit hthe same gpus in the system.
     
  17. Adhmuz

    Adhmuz TechSpot Paladin Posts: 1,648   +521

    The only reason the Quad is scoring so high is because I always had PhysX enabled, on the P4 it was not turned on. Even if it was turned on it would only be using a single 8800GT vs all the quad test with PhysX on and dual 8800GTs. If I could go back and run the test again on the P4 with SLI 8800GT and PhysX I'm almost 100% sure the results would be much improved.

    Look at what happens when I turn off PhysX with the i7, the CPU score almost cuts in half.
     
  18. Adhmuz

    Adhmuz TechSpot Paladin Posts: 1,648   +521

    Added to my results.
     
  19. klepto12

    klepto12 TechSpot Paladin Posts: 1,115   +9

    i still think the quad has alot to do with the overall score especially with 3d mark 06
     
  20. Adhmuz

    Adhmuz TechSpot Paladin Posts: 1,648   +521

    I never ran 3DMark 06 with my quad, I did do a run with my i7 and it was only 18527, lower than Red's score with his triple core.
     
  21. klepto12

    klepto12 TechSpot Paladin Posts: 1,115   +9

    i dont know for sure but i dont like 3d mark as a gauge for gaming in general.
     
  22. Ritwik7

    Ritwik7 TechSpot Chancellor Posts: 1,672   +9

    Hey Adhmuz, those results confuse me quite a bit. Are they Vantage results or 3DMark06?
     
  23. red1776

    red1776 Omnipotent Ruler of the Universe Topic Starter Posts: 5,219   +157

    Those were Vantage scores Rit
     
  24. Ritwik7

    Ritwik7 TechSpot Chancellor Posts: 1,672   +9

    @red : Have you tried Vantage? If so, what are your results?

    I'm still on DX9 so I can't do anything with Vantage...

    Here is a screenshot of my 06 scores.
     
  25. Adhmuz

    Adhmuz TechSpot Paladin Posts: 1,648   +521

    Those results seem accurate, I'm on SLI 8800GT (9800GT is just a rename, Darn you Nvidia! "Shakes fist in air") heres my score
     
Topic Status:
Not open for further replies.

Similar Topics

Add New Comment

You need to be a member to leave a comment. Join thousands of tech enthusiasts and participate.
TechSpot Account You may also...