Announcement

Collapse
No announcement yet.

Parhelia ViewPerf

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

  • Parhelia ViewPerf

    The review is Norwegian, but the numbers are clear.

    The results are a bit strange, all the first tests are very bad, but in the last test the Parhelia really shines, is this a bad Open GL issue?

    Anyway, Parhelia will have a hard time claiming to be a card for professionals with this kind of performance.


  • #2

    impressive, but inconsistent with the other benchmarks, I don´t know that benchmark, however i have heard that the other cards can´t do more than 8 light-sources in hardware, while parhelia supports unlimited number of lights, could that be the reason?
    This sig is a shameless atempt to make my post look bigger.

    Comment


    • #3
      If i am not completley wrong abt this then vertex shaders are used a lot in this particular test.

      Parhelia has 4 of those while GF4 have 2 and R8500 have 1.

      In the case of R8500 this card produced some visual errors that improved the score. Parhelia had no errors btw..

      So the difference between GF4 and Parhelia is more like u maybe could expect out from the specs on the cards, but why this do not show in 3dmarks shader tests is strange.

      Comment


      • #4
        Matrox needs to complete the certified OpenGL driver before anyone passes judgement on it's ability in professional applications. The current driver set is throughput limited and Viewperf just makes that abundantly clear.
        <a href="http://www.unspacy.com/ryu/systems.htm">Ryu's PCs</a>

        Comment


        • #5
          Well they have to be improved hell of a lot then and would not this show in gaming performance too. If they improve viewspec performance at least 100%, what they have to do here, then i would expect gaming performance to rise quite a bit too.

          Say 100% improvment in gaming fps!! Then i would start to look at parhelia as my next grapich card again

          Comment


          • #6
            regarding 3dmark polytest, i noticed that gf4ti had a HUGE performance-drop going from 1 to 8 lights, while parhelia had a very small performance-drop, the 8-light results for both cards are pretty close, I think this is yet another case of optimizing for worstcase-scenarios(many lights).
            This sig is a shameless atempt to make my post look bigger.

            Comment


            • #7
              It doesn't need to beat a Quadro 4 or even a VP760. Why? Because it's not in the same price category. That's the same reason it doesn't need to have a 100% performance improvement.

              The least expensive Wildcat is $599. The Least expensive Quadro 4 is $700. The Parhelia is $399. All it needs to do is perform like a Quadro DCC for the most part. Given it's ability to anti-alias for almost free it should have real strength against even the more expensive cards in CAD applications.
              <a href="http://www.unspacy.com/ryu/systems.htm">Ryu's PCs</a>

              Comment


              • #8
                Exactly my point in earlier posts on other threads.

                Professional grade high res OpenGL for 400$ will seem like some kind of a sick joke to the competition.
                If [M]atrox manages doing it while keeping CPU usage to minimum (a la Wildcat), they'll have a sure winner for quite a while.

                Comment


                • #9
                  The Parhelia is markeded more as a professional card than the GF4, and it is this card the parhelia is comparing to on those graphs.

                  Anyway i find it very strange that the parhelia seem to have such a potential in some tests, the 1 Viewperf test and Multi-texturing in 3dmark. What makes the card suck such in all the other tests??

                  I hope and think drivers have quite a potential, even Matrox have used much effort on making them good all from the start

                  Comment


                  • #10
                    *IF* the card is capable of rendering *very* complicated scenes at high resolutions with a decent framerate (above 30fps) using OpenGL certified drivers it'll be the boon of the graphic industry.

                    Comment


                    • #11
                      regarding the shader performace, i think it is because they optimized it for long and complex shader programs, the advanced pixel shader test is actually faster than the simple pixelshader test in 3dmark2001se , and it does well in matrox shark-mark and nvidias chameleon-mark too. it is just that only synthetic shader benchmarks have the complexity matrox optimized for.
                      This sig is a shameless atempt to make my post look bigger.

                      Comment


                      • #12
                        I think the reason that one gets such good results is because it uses AA lines in one portion of the test.

                        Comment


                        • #13
                          could be, parhelias AA lines performance could be its big saviour, in that market.

                          AFAIK parhelia is the ONLY $400 card that has AA lines, it is disabled in the gf4ti and radeon8500(although you can "hack" it on), because that is the only thing that seperates the competitions gaming cards from their pro cards.
                          Last edited by TdB; 27 June 2002, 09:18.
                          This sig is a shameless atempt to make my post look bigger.

                          Comment

                          Working...
                          X