Announcement

Collapse
No announcement yet.

Low 3D Graphics mark.

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Low 3D Graphics mark.

    Hi i have a old pc but a newer card.

    Q6600 Core 2 Quad @ 2.4 Ghz (not overclocked)
    8 GB DDR2 RAM
    680i motherboard

    570 GTX Video card (up-to-date driver - 306.23)


    I understand that this combination wont get the MOST out of my video card; but my scores are less than 25% of what the online scores are getting. I get a 800-850 score. These seem really low even considering the older cpu and mobo. Other people are having scores of 3,600 for the same card.

    What can be causing such horrible scores?


    UPDATE:
    Fix #1: Disabled "PCIe Spread Spectrum" in BIOS
    Fix #1 Result: +1,200 3D Points

    Fix #2: None yet. Still looking.
    Fix #2 Result: Hoping for a score close to 3,618. Realistically expecting less. Hopefully near 2,500-2900.
    Last edited by bi99uns; Oct-03-2012, 02:24 AM.

  • #2
    What are the individual scores for each of the 3D tests?

    Is the score for the CPUmark what you expect, or is the CPU also slow?

    See also this post for general hints and tips,
    http://www.passmark.com/forum/showth...-for-a-slow-PC

    Comment


    • #3
      well.. i have my last results still open.

      cpu: 3220
      2D: 434 ( 3.6 and 3.44 solid and trans vectors.. is this slow?) also not sure if 2D is for gpu or cpu.
      3D: 855 (BIG OUCH)
      Memory: 1248

      I am unsure how to copy and paste the specific scores for each section.

      each time i ran it, scores for the vid card were similar. About a 50-75 point deviation. it never go to even 1,000 points.

      Comment


      • #4
        CPUMark seems OK.
        So I would step through all the points in the forum post I linked to above. Maybe it is overheating on the video card or something like that.

        Comment


        • #5
          Originally posted by David (PassMark) View Post
          CPUMark seems OK.
          So I would step through all the points in the forum post I linked to above. Maybe it is overheating on the video card or something like that.
          i turned on evga PrecisionX just to see the temps when the test runs. Temp is around 57-60c

          Comment


          • #6
            UPDATE:

            I just significantly increased my score.

            There was a BIOS setting:
            "PCIe Spread Spectrum (SPP): [AUTO]"
            After reading online, I read that I should probably disable it.

            Old 3D Graphic Mark: 820 Avg.
            New 3D Graphic Mark: 2,000 Avg.

            However, based upon the benchmarks on this site, the card is still about 1,600 points off from what other people are getting. Its running at only 55% of what people are getting.

            I do expect to have a lower overall score considering that im a slower CPU and i have PCIex16, but it still seems a bit far off from the other benchmarks.

            I wonder how many people on these benchmarks are overclocking.
            Last edited by bi99uns; Oct-03-2012, 02:26 AM.

            Comment


            • #7
              Originally posted by bi99uns View Post
              well.. i have my last results still open.

              cpu: 3220
              2D: 434 ( 3.6 and 3.44 solid and trans vectors.. is this slow?) also not sure if 2D is for gpu or cpu.
              3D: 855 (BIG OUCH)
              Memory: 1248

              I am unsure how to copy and paste the specific scores for each section.

              each time i ran it, scores for the vid card were similar. About a 50-75 point deviation. it never go to even 1,000 points.
              The graphic scores using passmark are useless imo because there is no standardization. Try something like heaven. Passmark is heavily dependent on software (windows and nivida settings). For example my PC has i7-3930k + 2x GTX 670 + 64Gb PC-19200 mem. I can get get swings of 50% in both passmark V7 & V8. If you want to see a higher score, in control panel -> performance settings -> visual effects -> change to performance rather than quality. That should increase your 2d scores. You can even lower the resolution of your monitor and it will increase scores... For 3d go into the Nivida settings and again change it all to performance and try playing around with a couple things. I have gone from 2200 on the bouncy ball to 7000 by just making and deleting a custom rule in the Nivida settings. I am sure there are other tricks to get even higher scores that i dont know about if that is all you care about but at the end of the day it doesn't reflect the day to day performance of the card.

              Comment


              • #8
                Im not worried about tricks to getting a better score on paper. I am concerned about the video card's combination with my older motherbaord and CPU, and what parameters might be hindering performance. I'm trying to see if there are any settings in bios or overclocking methods that can possitively affect the gpu's performance.

                Comment


                • #9
                  In reply to the comments from Dbo,
                  The purpose of a benchmark is to reflect the performance of the system. If there are settings in the operating system and the device driver that allow a trade off to be made between performance and display quality then it makes sense that graphics benchmark is impacted by these settings. This allows people to measure the impact of these setting changes on their system.

                  Comment


                  • #10
                    Originally posted by bi99uns View Post
                    Im not worried about tricks to getting a better score on paper. I am concerned about the video card's combination with my older motherbaord and CPU, and what parameters might be hindering performance. I'm trying to see if there are any settings in bios or overclocking methods that can possitively affect the gpu's performance.
                    That's kinda what i am trying to get at. Its hard to compare one graphics score to another because you don't know what other settings are in play that go into that average you trying to get close to. You card might be working perfectly fine and maxed out or it might not.

                    For your reference when I OC my cards running in SLI to +70mhz core & +400mhz memory (1209mhz core, 1702mhz mem) I go from 142.3 FPS to 152 FPS averagerunning heaven with the following settings. So maybe a you might see a 7% bump from OC'ing depends on the card.
                    Settings

                    Render: direct3d11
                    Mode: 1680x1050 fullscreen
                    Shaders: high
                    Textures: high
                    Filter: trilinear
                    Anisotropy: 16x
                    Occlusion: enabled
                    Refraction: enabled
                    Volumetric: enabled
                    Tessellation: extreme
                    Last edited by Dbo; Oct-03-2012, 11:40 PM.

                    Comment


                    • #11
                      Depending on the popularity of the individual video cards, the average result might be made up of results from over 10,000 different machines. Even for the less popular cards the average is typically made up from 100s of machines.

                      There are dozens of factors that influence the benchmark numbers. Any single number anybody comes up with to represent is a best an approximation. You can be pretty sure that the average machine out of 100 samples isn't overclocked however. Only the top 20% of CPUs tend to be overclocked, and the number is surely less for overclocked video cards. (except for those models that are always sold as factory overclocked units).

                      Comment


                      • #12
                        Earlier today, i overclocked my CPU to 3.0 Ghz. GPU is factory oc'd. I just gained +650 3D Points! My first overclock.

                        Current 3D score: 2670
                        Im still not there, but im at 73% of the other scores.. This is where i expected i would be. So far it has been a great improvement, I must say. Changing my performance settings in windows did nothing for me.

                        I plan on trying to suck out another .2-.5 Ghz out of this CPU, this should increase the scores even a bit more.

                        Comment


                        • #13
                          I did some searching and filtering in the baseline database.

                          There are 99 baseline results from machines with the same CPU and video card.

                          The median 3DMark result for these 99 machines is 2470.

                          So given your older CPU, RAM and motherboard hardware you are doing OK.

                          I guess this means that the average person with a GTX 580 has a better CPU, etc.. than your machine.

                          Comment


                          • #14
                            Originally posted by David (PassMark) View Post
                            I guess this means that the average person with a GTX 580 has a better CPU, etc.. than your machine.
                            Do you mean GTX 570?

                            Comment


                            • #15
                              Yes sorry about the typo.
                              Average 3DMark result for a GTX 570 with a Q6600 is ~2470.
                              (Based on PerformmanceTest V7 results)

                              So you are slightly above average for your CPU/GPU combination. But below average compared to everyone with a GTX 570.

                              Comment

                              Working...
                              X