Announcement

Collapse
No announcement yet.

Results misleading

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

  • Results misleading

    [Admin: Moved to a separate thread, out of the "solutions for a slow PC thread"]

    I am sorry but the way you report numbers on your cpu and gpu charts is simply not a realistic image of the actual hardware. People considering purchasing this hardware have no way of knowing the many variables in the systems you test (such as motherboards, memory, cooling, background services, overclocking, SLI or Crossfire, operating systems etc.. etc.. etc...) that you give no information about.
    Your charts are a better indication of hardware potential in the hands of an expert with unlimited cash than the actual performance of a single CPU or GPU in the hands of an ordinary user.
    As one simple example todays report says that an nVidea gtx 285 outscores an nVidea gtx 295 by a wide margin of 2025 to 1682 points. Any dufuss knows that the 295 is by far the more powerful card with 2 GPUs to the one GPU of the 285. Clearly the 285 numbers are skewed upward by dual and triple SLI arrangements. I could easily cite 10 more exampes from the same chart without even moving on to the misleading information in the other charts.
    Maybe you should publish a brief hardware/software report in a list of ranked systems for each type of CPU and GPU before sending users off on a wild goose chase with the advice you are presenting here. Certainly you have this info and readers deserve to see more clarity on your part.
    Last edited by fastrabbit; Jun-26-2010, 03:37 PM. Reason: clarity

  • #2
    SLI and Crossfire only work if the application is supported by the driver. SLI and Crossfire do not work in PT because Nvidia and ATI will not add PT support to their drivers. A GTX 295 is two GTX 285 GPU's under clocked on a single card. Since PT can't test SLI performance and a GTX 295's core and memory run at a slower speed than a GTX 285 end result is a GTX 285 tests faster as it should.

    I do agree it would be nice to have a chart that compensates for the lack of SLI/Crossfire support in PT so people with limited knowledge can use the charts for its intended purpose, aka which card is faster. IMO SLI and Crossfire are only intended for power users with money burning a hole in their pockets anyway so it really does not matter if the charts are correct or not.
    Last edited by wonderwrench; Jun-26-2010, 07:50 PM.
    Main Box*AMD Ryzen 7 5800X*ASUS ROG STRIX B550-F GAMING*G.SKILL 32GB 2X16 D4 3600 TRZ RGB*Geforce GTX 1070Ti*Samsung 980 Pro 1 TB*Samsung 860 EVO 1 TB*Samsung 860 EVO 2 TB*Asus DRW-24B3LT*LG HL-DT-ST BD-RE WH14NS40*Windows 10 Pro 21H2

    Comment


    • #3
      Passmark benchmark charts are highly misleading

      Perfomance Test (PT) software and it's inability to communicate with nVidea and ATI drivers is only one issue that can be circumvented by other means.

      Belarc Advisor does a great job of reporting hardware/software setups without communicating to anything other than Windows. For starters, it can tell the cpu type, the type and number of graphics cards installed, the precise motherboard type, the windows build, the memory hardware as well as every piece of software residing on the system. A simple 10 column sort-enabled database with essential system information from the computers being tested isn't much to ask.

      Why can't Performance Test be made to do this? Perhaps a partnership with Belarc is in order if Passmark's code writers can't easily do it, althouth I believe they already can.

      It would be a tremendous asset to the industry and consumers if they could see which hardware combinations get the best results so they could buy those parts and build more powerful systems.
      Last edited by fastrabbit; Jun-26-2010, 09:38 PM. Reason: clarity

      Comment


      • #4
        SLI only works with a small number of applications. And doesn't work with PerformanceTest. Despite our efforts to make it work. There is nothing we can do in our code to make it work (we tried). nVidia need to update their drivers or SDK if is it to work, and nVidia don't seem interested in doing this.
        This has all been commented on
        http://www.passmark.com/forum/showthread.php?t=1435
        http://www.passmark.com/forum/showthread.php?t=668
        http://www.passmark.com/forum/showthread.php?t=2520

        With a couple of minor exceptions PerformanceTest reports correct system information. Maybe you missed looking at the System information tab in the main window? How do you think the graphs are drawn if we don't collect the CPU type?

        And in any case the collection of system information (which we do) is totally unrelated to the SLI based 295 card not scoring as high as you would like.

        Maybe you should publish a brief hardware/software report in a list of ranked systems for each type of CPU and GPU
        This isn't possible. Maybe you don't understand what the charts are displaying. We have collected around 250,000 benchmark results. The charts are an average. Each row in the charts might represent 1000s of systems all averaged together. We would need to add a quarter of a million web pages to our site to display all the systems. So it isn't practical.

        Also the CPUs are unaffected by the type of video card & disk you have. So if you are looking at CPU score, you don't need to know which video card was used at the time.

        Comment


        • #5
          Sorry but ....

          After reading your links, I feel like I am being mean in my critism of some of your results with video card performance tests given that those results may be a result of nVidia and ATI not playing nice with you. I also hate being brutally honest but 3DMark06 gets the sli/cf/single card ratings in the right order in the tom's hardware video card chart.

          Yes, I understand that your cannot create a quarter million web pages, but that is really not what I am asking for. Having read Passmark CPU result charts for several years now, I like them, (and I will continue to use them) but as an sli/cf using game player and system builder, I wish for information that enables me to make better hardware purchasing decisions.

          The main thing missing from your top 20 system chart is the motherboard being used. If you could detect more hardware and pubish columns like motherboard and memory types in the top 20 web page (and perhaps increase it to the top 100) you would have one up on the competition .... and people like me would not have to hunt for so many current unbiased reviews all of the time.

          Finally, high-end gamers are probably among the more frequent visitors to the PassMark site so slagging sli (and by default crossfire) won't get you anywhere. I am also grateful for the opportunity to say what I have said in this space. You may or may not be able to do what I have suggested, but at least you are listening.

          Comment


          • #6
            SSD Benchmarks useful

            It's helpful to have such a complete list of the popular high end SSDs on the market.

            After having some serious stuttering issues trying to play Crysis on my new Kingston SSDNow 64GB, I got a little turned off by some of the hype surrounding these devices. I even advised against putting SSDs into our factory floor data collection servers based upon my experience, the large number of small read/writes we did and the immaturity of the technology. My friend's jobs were possibly on the line had they experienced issues similar to mine.

            The more information people have, the better ... and hype sucks.

            Comment


            • #7
              You have posted a complaint about SSD drives in a thread about graphics cards and SLI.

              Generally games don't use the hard drive much after the initial level load. So it isn't obvious how a SSD might cause frame rate stuttering. Maybe the cause was elsewhere.

              The SSDs I have used have all been great (except for one particularly cheap one). Both for for gaming and work.

              Comment


              • #8
                I am certain it was the SSD

                Originally posted by passmark View Post
                You have posted a complaint about SSD drives in a thread about graphics cards and SLI.

                Generally games don't use the hard drive much after the initial level load. So it isn't obvious how a SSD might cause frame rate stuttering. Maybe the cause was elsewhere.

                The SSDs I have used have all been great (except for one particularly cheap one). Both for for gaming and work.
                To be frank, I did not say gaming ... I said Crysis (Warhead to be more explicit) which I hoped would perform well with my expensive new SSD. My regular system has a 9800gx2 with a Q6600, 4MB of pc6400 and a terrabyte Seagate HDD on Windows 7 - 32 bit.

                Everything stayed the same including the screen resolution except I gave the Kingston SSDNow 64MB a fresh fully updated install of Win7-32 (dual Win7 install on the same computer), then installed Crysis Warhead, even checking with an internet script later to be certain the trim instruction was enabled and working.

                Warhead played well for about 90 minutes then began stuttering and freezing finally becoming completely unplayable. I temporarily fixed the problem by reducing the paging file to zero as I suspected random read write problems were occurring because the trim instruction could not keep the paging file fully functional while Crysis was hogging all of the system resources as it is famous for doing. (Yes, I read AnandTech). For whatever reason the game never played well even after I returned the paging file to a normal size and then an increased size.

                None-the-less when I rebooted into my original Win7 installation on my terrabyte Seagate HDD everything was fine and Crysis Warhead performed flawlessly using exactly the same other hardware and settings. GPU temperature was similar in both cases and not a factor.

                I am using the Kingston drive an a new P55 build with 8GB of memory and I will try installing Crysis 2 when it comes out in March 2011 to see if the extra memory and a 480gtx graphics card helps the Kingston SSD drive out.

                Until then, I stand by my guns with respect to Crysis and this SSD.

                As well the reasoning for which I advised against installing SSDs on our factory floor data servers includes the fact that we use Windows 2003 Server which as you know does not have trim (it's risky without trim and our CPUs run at 30-50% load with steady hard drive usage most of the time). Downtime also costs $30,000 per hour in our plant.
                Last edited by fastrabbit; Sep-04-2010, 04:39 AM.

                Comment


                • #9
                  I did not say gaming ... I said Crysis Crysis
                  Last time I checked Crysis was a well known game.

                  I temporarily fixed the problem by reducing the paging file to zero..
                  There is no reason to think stuttering in a game is due to paging when you have 4GB of RAM. Even less reason to think it is due to the lack of trim support. You more or less proved this yourself, when you set the paging file back to normal and it didn't fix the problem.

                  Comment


                  • #10
                    Okay I guess I'm just an idiot and you are a genius. Forget it buddy.

                    Comment

                    Working...
                    X