Announcement

Collapse
No announcement yet.

9-Way AMD Radeon Comparison On Ubuntu With Catalyst 14.1 Beta

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

  • 9-Way AMD Radeon Comparison On Ubuntu With Catalyst 14.1 Beta

    Phoronix: 9-Way AMD Radeon Comparison On Ubuntu With Catalyst 14.1 Beta

    For those curious how AMD's Catalyst Linux performance is doing as we get 2014 underway with the first Catalyst 14.1 beta, here are benchmarks from nine different AMD Radeon graphics cards under Ubuntu Linux and running this latest publicly available driver when looking at both the OpenGL graphics and OpenCL compute performance.

    http://www.phoronix.com/vr.php?view=19862

  • #2
    Nice review, it looks like the R290 is finally starting to deliver some of it's available performance under Linux as well as Windows.

    Would it be possible when producing a driver review like this to include the most recent previous benchmark result for each of the cards as well as the new driver?

    This would make it much easier to see what is changing instead of having to try and reference two articles to determine the differences.

    Comment


    • #3
      Originally posted by Herem View Post
      Nice review, it looks like the R290 is finally starting to deliver some of it's available performance under Linux as well as Windows.

      Would it be possible when producing a driver review like this to include the most recent previous benchmark result for each of the cards as well as the new driver?

      This would make it much easier to see what is changing instead of having to try and reference two articles to determine the differences.
      At some point when more drivers are out I may do such a comparison with the R9 290 and other CIK hardware... But this article, as explained in the article, was just about sharing some new AMD Linux benchmarks prior to a big Linux hardware review in the coming days.
      Michael Larabel
      http://www.michaellarabel.com/

      Comment


      • #4
        HD6450 would be doing so much better if it had a 128-bit interface instead of the 64-bit interface. Even if it was the GDDR5 version, the 64-bit interface cripples it pretty severely.

        Also could we maybe get a 9way comparision with the same hardware running the open source radeon vs 14.1 catalyst beta?

        Comment


        • #5
          Originally posted by Herem View Post
          Nice review, it looks like the R290 is finally starting to deliver some of it's available performance under Linux as well as Windows.
          R9 290, not R290.

          Comment


          • #6
            Indeed

            Originally posted by synaptix View Post
            HD6450 would be doing so much better if it had a 128-bit interface instead of the 64-bit interface. Even if it was the GDDR5 version, the 64-bit interface cripples it pretty severely.

            Also could we maybe get a 9way comparision with the same hardware running the open source radeon vs 14.1 catalyst beta?
            Yes, please, a Radeon vs catalyst 13.12 vs catalyst 14.1

            Comment


            • #7
              Originally posted by Michael View Post
              At some point when more drivers are out I may do such a comparison with the R9 290 and other CIK hardware... But this article, as explained in the article, was just about sharing some new AMD Linux benchmarks prior to a big Linux hardware review in the coming days.
              I believe that he actually made feature request for Phoronix.com to automatically fetch data from (openbenchmarking?) the past, and providing that for reference point.

              As opposed to request for mooooorrrrreeeee benchmarks to be made (Though we surely will read all those that You work on right now)

              Comment


              • #8
                Originally posted by Michael View Post
                At some point when more drivers are out I may do such a comparison
                Could you alternatively maybe just add a link with a comparison with some recent benchmark on openbenchmarking? It's alway so cumbersome to search a fitting benchmark there...

                Or, also alternatively, could you add a search functionality to openbenchmarking to search for benchmarks with the same set of tests as the current one? I'm sometimes trying to compare results but then they just have 2 or 3 benchmarks in common and it doesn't really work out...

                Comment


                • #9
                  I'm mainly just interested in seeing the performance gains of the 7950. I don't even own anything of the HD7000 series, but the 7950 I think would be the best to show improvements.

                  Also, did hell freeze over or something? I'm surprised nobody is complaining about "irrelevant tests" (finally)

                  Comment


                  • #10
                    Any ideas what's the reason for the bad performance of 7950 in GpuTest Volplosion?
                    Actually it's the full 7xxx line and up (excluding R9 290).

                    Thanks for these benchmarks, Michael!

                    Comment


                    • #11
                      Is there some asinine reason we can't see a full AMD FX-8350 pairing with the AMD GPGPUs running against the Intel/AMD GPGPU combo?

                      How about progress of CodeXL 1.3 showing the full swath of OpenCL leveraging the CPU/GPGPU combo of AMD?

                      Comment


                      • #12
                        Originally posted by Marc Driftmeyer View Post
                        Is there some asinine reason we can't see a full AMD FX-8350 pairing with the AMD GPGPUs running against the Intel/AMD GPGPU combo?

                        How about progress of CodeXL 1.3 showing the full swath of OpenCL leveraging the CPU/GPGPU combo of AMD?
                        A bit of an over-reaction eh? There is a reason AMD isn't used in 99% of all benchmark websites, though I wouldn't consider it asinine. If you want the most accurate results, you use the best system to your disposal. Considering intel beats AMD in most tests, it makes sense to only use them. It also makes sense why you'd want an overkill amount of RAM and a single SSD (RAID could have an impact on results).

                        I'm not bashing AMD, I myself own a FX-6300 which I use almost every day. But like every test, you need a control group, and that's what intel falls under.

                        Comment


                        • #13
                          Thanks!

                          Thanks for this long awaited test. By chance my R9 290 arrived yesterday. Great play times ahead! Yay! ^^
                          Lockheed

                          Comment


                          • #14
                            Originally posted by Marc Driftmeyer View Post
                            Is there some asinine reason we can't see a full AMD FX-8350 pairing with the AMD GPGPUs running against the Intel/AMD GPGPU combo?

                            How about progress of CodeXL 1.3 showing the full swath of OpenCL leveraging the CPU/GPGPU combo of AMD?
                            It would be nice to see the results of an AMD CPU/GPU pairing, because I found some benchmarks where the performance on my AMD FX-8320 bottlenecked due to the single-core performance not being up at the level of Intel CPUs.
                            My CPU/GPU config is the AMD FX 8320 and R9 270X, the CPU appeared to be a bottleneck on a few of the benchmarks; OpenArena, Xonotic, etc (sorry don't remember them all. I couldn't save the benchmarks using the suite due to some of the tests that couldn't find the OpenCL libraries on my Fedora setup). Other benchmarks such as the Unigine Heaven/Valley the results were comparable to those with the Intel CPU/R9 270X, I presume this is because they are very demanding on OpenGL and most work is being done by the GPU or it could also be that they make use of the additional cores?!
                            Additionally, I have an overclocked GPU Sapphire R9 270X Toxic 2GB; here both the core and unusually the memory too is overclocked resulting in an ~10% boost. So when I said the Unigine results where comparable they were actually a tad better on my system.

                            Comment

                            Working...
                            X