Announcement

Collapse
No announcement yet.

How can I compare single vs. dual cpu systems?!

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

  • How can I compare single vs. dual cpu systems?!

    Hi there,

    I am trying to put a computer together for video editing. This site helps a lot when you want to look up, what cpu is "faster"! So thanks for that! What I am not sure about is, is I can simply add up benchmark scores?! For example: If one i7 has a score of 10000 and two xeons have a score of 8000 each, does this automatically mean the two xeons are faster because they have a combined score of 16000?! From my understanding this is not how it really works, does it?! I believe it depends on the software application as well, right?!

    I am flying Radio Controlled Helicopters for years now and just switch to Multicopters. There are plenty of websites out there, that allow you, to enter the components you are going to use, and the website will calculate all kinds of things for you?! Amp draw, watts, run time and so forth!? Is there a website or program that allows you to do the same?! It would be nice if you could upload a small clip or so and the website or program would allow you to change different Graphics cards, RAM and cpu and then calculate how long it would take to render the file... blah blah blah, ha ha. I hope you know what I mean?!

    thanks,

    Georgios

  • #2
    Originally posted by neavissa View Post
    does this automatically mean the two xeons are faster because they have a combined score of 16000?!
    No. Adding CPUs doesn't scale in a linear fashion. Other things like RAM become the bottle neck. And as you say the software also needs to be coded in such a way that it can use all the available cores.

    This Multi-CPU performance chart might also help,
    http://www.cpubenchmark.net/multi_cpu.html

    There is no web site that I know about that allows you to calculate the time it would take to render a video file using an arbitrary encoder, on arbitrary hardware with an arbitrary video. There are too many variables.

    Comment


    • #3
      Originally posted by David (PassMark) View Post
      No. Adding CPUs doesn't scale in a linear fashion. Other things like RAM become the bottle neck. And as you say the software also needs to be coded in such a way that it can use all the available cores.
      I thought so... too bad.

      This Multi-CPU performance chart might also help,
      http://www.cpubenchmark.net/multi_cpu.html
      Sure it helps! Thanks!

      There is no web site that I know about that allows you to calculate the time it would take to render a video file using an arbitrary encoder, on arbitrary hardware with an arbitrary video. There are too many variables.
      Well, take a look at this site and you get an idea with how many variables they deal with!

      http://www.ecalc.ch/xcoptercalc.htm?ecalc&lang=en

      I am no programmer, but I would think it is possible to create a website like this. You do not have to start with every possible combination of components out there, but you can start with the most common once... Of course you would have to come up with some kind of formula, to calculate the difference in performance. Maybe you start with gathering information from existing PC and see if there is some sort of formula you can come up with?!

      For example: (just making something up here)...

      Intel Core i7-3960X Extreme Edition Sandy Bridge-E 3.3GHz (3.9GHz Turbo)

      24GB 1300mhz

      7200 WD hard drive

      GTX Titan 6Gb

      and so on...

      If this person downloads a specific file, loads it into lets say Adobe Premiere CS6 and applies a SPECIFIC effect or transition and saves it. Simply clock how long it takes to render the file and you have your first "benchmark"?! Hopefully you find someone that uses the same system exempt more or less RAM?! You have him to the same test and now you have two PC to compare to.

      Anyhow, I am just having a brain fart I guess?! LOL

      thanks for the fast reply and your time!

      Gerogios

      Comment


      • #4
        This is more or less what our charts already do.
        Except that we have removed a bunch of variables / options, like having a user defined video file. We run pre-defined algorithms on pre-defined data that we believe can be extrapolated into a measure of general computing performance.

        Comment


        • #5
          Originally posted by David (PassMark) View Post
          This is more or less what our charts already do.
          Except that we have removed a bunch of variables / options, like having a user defined video file. We run pre-defined algorithms on pre-defined data that we believe can be extrapolated into a measure of general computing performance.
          Maybe my whole idea of knowing BEFORE hand, how fast an ENTIRE system is going to be, is not that great?! I guess what people do is, they look at the benchmark for a given CPU, a given GPU and that's it?! You can probably not go that wrong with picking the highest scoring CPU and the highest scoring graphics card and slapping them together?! I have no idea how RAM works?! Probably not linear either?! So 48GB is not going to render your project twice as fast as 24GB would do... Anyhow, RAM is not that expensive, so you might as well slap as much in there as you can afford...

          thanks again,

          G

          Comment


          • #6
            Originally posted by neavissa View Post
            Maybe my whole idea of knowing BEFORE hand, how fast an ENTIRE system is going to be
            If you want to have a look at system wide performance then the overall Passmark rating is a good indication. You can look up results from other machines in the PerformanceTest software. It still won't tell you how long it takes to encode a arbitrary video however.

            Comment


            • #7
              Originally posted by David (PassMark) View Post
              If you want to have a look at system wide performance then the overall Passmark rating is a good indication. You can look up results from other machines in the PerformanceTest software. It still won't tell you how long it takes to encode a arbitrary video however.
              I see. Thanks for all the info and help!

              Comment


              • #8
                Great thread! I know it's old but it answers my exact question. I built a Plex server last week in hopes of scoring somewhere in the high 20,000's or even 30,000. So I looked carefully at the charts and choose a CPU that had a Passmark of 16k then purchased two of them used.

                The system is up and running and I'm getting a score of 21k. So that was a head scratcher. My CPU's are Engineering samples of the E5-2658 v3. Which means the clock speed is slightly lower than the production chips. Comparing it so comparable dual E5-26xx 12 core tests, it looks like I'm in the ballpark of what I should be expecting from this system.

                Love the software!

                Comment


                • #9
                  If a single CPU scores 16,000, it isn't too surprising that two slower versions of that CPU score 21,000 in total.
                  Doubling the CPUs doesn't double the power of the machine.

                  Comment


                  • #10
                    Originally posted by David (PassMark) View Post
                    If a single CPU scores 16,000, it isn't too surprising that two slower versions of that CPU score 21,000 in total.
                    Doubling the CPUs doesn't double the power of the machine.
                    That's what I gathered. I don't know too much about engineering samples but the only difference I can see is that the clock speed of my chips is 2.0GHz (2.6 Turbo) per core and the production model is 2.3GHz (2.9 Turbo) per core. Every other spec is the same. So.... I can make this work.

                    Comment

                    Working...
                    X