Announcement

Collapse
No announcement yet.

NVIDIA GeForce GTX TITAN Benchmarks On Linux

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

  • NVIDIA GeForce GTX TITAN Benchmarks On Linux

    Phoronix: NVIDIA GeForce GTX TITAN Benchmarks On Linux

    Here's some of the first OpenGL benchmarks of the ultra high-end $999 (USD) NVIDIA GeForce GTX TITAN running on Linux...

    http://www.phoronix.com/vr.php?view=MTMyNDY

  • #2
    Originally posted by phoronix View Post
    Sadly, with an uncooperative NVIDIA PR/marketing department

    Comment


    • #3
      TF2 with all the eyecandy cranked to the max @ 2560x1440 (vsync disabled) using Kano's benchmark script.

      3824 frames 14.299 seconds 267.43 fps ( 3.74 ms/f) 26.501 fps variability

      Now come on Valve, bring me some L4D2 loving.

      Comment


      • #4
        Useless benchmark

        With nothing to compare against, this benchmark is useless!

        No reference values.

        Comment


        • #5
          Originally posted by uid313 View Post
          With nothing to compare against, this benchmark is useless!

          No reference values.
          Through the Phoronix Test Suite you can see how your system compares by simply running phoronix-test-suite benchmark 1303113-FO-GEFORCETI80
          Feel free to use it as the reference on your own.

          Comment


          • #6
            Originally posted by deanjo View Post
            Through the Phoronix Test Suite you can see how your system compares by simply running phoronix-test-suite benchmark 1303113-FO-GEFORCETI80
            Feel free to use it as the reference on your own.
            So I am going to compare it against benchmarks on my computer which is an entirerly different system with a different CPU of a different generation, running a different Linux distribution compiled with a different version of the compiler, running a different version of the kernel, X.org Server, and device drivers?

            The results will say nothing and be completely useless for comparison.

            Comment


            • #7
              Originally posted by uid313 View Post
              So I am going to compare it against benchmarks on my computer which is an entirerly different system with a different CPU of a different generation, running a different Linux distribution compiled with a different version of the compiler, running a different version of the kernel, X.org Server, and device drivers?

              The results will say nothing and be completely useless for comparison.
              So instead you expect me to pull apart my system, on my own time, for zero reason of benefiting me except to appease your laziness. Since I cannot replicate your system anyways any comparison would be meaningless as well as the differences on my systems may not be parallel to differences you would see on your own. You want a comparison, it's about 2x + faster then the GTX-580 it replaced pretty much across the board.

              Comment


              • #8
                For heaven's sake, just write $1000. What's with the $999 nonsense? Seriously.

                Comment


                • #9
                  Originally posted by RealNC View Post
                  For heaven's sake, just write $1000. What's with the $999 nonsense? Seriously.
                  I'm Canadian, make that 1 kilodollar and I will agree.

                  Comment


                  • #10
                    Originally posted by RealNC View Post
                    For heaven's sake, just write $1000. What's with the $999 nonsense? Seriously.
                    But don't you see!? $999.99 is SO much cheaper then $1000! I have hated the 99 thing my whole life. It's like change, add sales taxes to your list price and round to the nearest dollar.

                    Comment


                    • #11
                      My regards for volunteering for this high-priced piece of silica.

                      But - as others already said, useless due to the lack of reference values.

                      What about OpenCL benchmarks? This expensive GPU is supposed to be more adequate for numerical research and number crunching than simple OpenGL-stuff and therefore I'd have expected some benchmarks regarding this. It would be nice to see the performance of OpenCL versus CUDA - which I suspect very bad for OpenCL, since the BLOB does by intention cripple the OpenCL.

                      The competitor in numerical applications/GPGPU would be AMD's GCN architecture in it incarnation HD7970 and for what I saw in the net, Kepler II/GK110 as TITAN is only by 30% faster than the most recent GCN on the GHz-editions of the HD7970. Power consumption on maximum load, minimum load, idle et cetera.

                      I'm sorry to say that, but having only a web-google-searchable tag "TITAN" or "benchmark" seems to be the only reason for this article to raise the hit marks on this site!

                      But anyway, my appreciation to the "benchmark apprentice" for buying the silica.

                      Comment


                      • #12
                        Originally posted by Eisnefaust View Post
                        My regards for volunteering for this high-priced piece of silica.
                        I bought it for personal uses.

                        But - as others already said, useless due to the lack of reference values.
                        As I said before you are perfectly capable of comparing it to your own setup which will be more relevant to you then numbers crunched on another system.

                        What about OpenCL benchmarks? This expensive GPU is supposed to be more adequate for numerical research and number crunching than simple OpenGL-stuff and therefore I'd have expected some benchmarks regarding this.
                        Oh I will be running a lot of GPGPU on the card, it was purchased primarily for CUDA development. Even then however I am going to be moving it into a Z77 board before to long as the board it is on with only PCI-e 2 stiffles the performance of computing the large data sets that I use.

                        It would be nice to see the performance of OpenCL versus CUDA - which I suspect very bad for OpenCL, since the BLOB does by intention cripple the OpenCL.
                        If there is a great difference between CUDA and openCL it would be from lack of optimization of the code for the device. The "crippling" of openCL remark is pure FUD however going from previous Cuda -- openCL ports experience with previous generations of CUDA/openCL cards.

                        The competitor in numerical applications/GPGPU would be AMD's GCN architecture in it incarnation HD7970 and for what I saw in the net, Kepler II/GK110 as TITAN is only by 30% faster than the most recent GCN on the GHz-editions of the HD7970. Power consumption on maximum load, minimum load, idle et cetera.
                        Wouldn't be able to tell you that as I do not have any GCN cards nor plan on purchasing any.
                        Last edited by deanjo; 03-12-2013, 09:09 AM.

                        Comment

                        Working...
                        X