Announcement

Collapse
No announcement yet.

AMD Radeon HD 8670D: Gallium3D vs. Catalyst

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

  • #16
    Originally posted by Michael View Post
    Any other game with automated test abilities works just fine
    That's kind of like saying, 99% of interesting tests are broken, but hey the other 1% all work.

    , but after you tell a Steam game to quit from its console, even when it was launched directly from the command-line, the screen goes away and returns to the desktop, but the process actually remains alive. That's an issue with the upstream (steam).
    Sure, i just don't understand why it matters. Why does PTS break if the process remains alive? Why can't you just use the kill command? If this isn't possible in PTS for some reason, it sounds to me like PTS has a poor design. Or rather, I'm sure the design seemed fine, but it fails to match up with what is needed to properly test proprietary games on Steam, which would seem to be a very important goal.

    Maybe the issue here is the automated bit, that requires you to do everything from the command line.
    Last edited by smitty3268; 07-05-2013, 07:42 PM.

    Comment


    • #17
      Originally posted by smitty3268 View Post
      That's kind of like saying, 99% of interesting tests are broken, but hey the other 1% all work.

      Sure, i just don't understand why it matters. Why does PTS break if the process remains alive? Why can't you just use the kill command? If this isn't possible in PTS for some reason, it sounds to me like PTS has a poor design. Or rather, I'm sure the design seemed fine, but it fails to match up with what is needed to properly test proprietary games on Steam, which would seem to be a very important goal.
      PTS doesn't break, it just sits there waiting for the process to be killed. Could add a kill command to the test profile but there isn't any sane way to do that unless just having it sit and wait for an arbitrary amount of time to kill the process, hoping the test completed in that amount of time.
      Michael Larabel
      http://www.michaellarabel.com/

      Comment


      • #18
        RAM Clocks

        Originally posted by ObiWan View Post
        1066 Mhz is DDR3 2133,

        DDR3 1066 would be 533 Mhz
        Michael please can you make clear to us what was the RAM clock in the tests you made for the APUs??
        Is it 1066Mhz or 2133Mhz as friend here said??
        Cause in an old article you proved that AMD APU's both CPU and GPU performance increase almost linearly with higher DDR3 RAM clocks!

        Comment


        • #19
          Originally posted by djdoo View Post
          Michael please can you make clear to us what was the RAM clock in the tests you made for the APUs??
          Is it 1066Mhz or 2133Mhz as friend here said??
          Cause in an old article you proved that AMD APU's both CPU and GPU performance increase almost linearly with higher DDR3 RAM clocks!
          It's at 2133MHz as controlled from the BIOS/UEFI but the Catalyst driver as was reported to the result table reports 1066MHz.
          Michael Larabel
          http://www.michaellarabel.com/

          Comment


          • #20
            Furthermore can you write the RAM clock at the hardware chart at each benchmark you make not just the amount, I mean something like 7068MB@1600Mhz or something...

            Comment


            • #21
              Originally posted by Michael View Post
              It's at 2133MHz as controlled from the BIOS/UEFI but the Catalyst driver as was reported to the result table reports 1066MHz.
              Thank you very much! So our APU was theoritically running at full throttle as far as it concerns the RAM speed! I don't speak for overclock speeds...

              Comment


              • #22
                Originally posted by djdoo View Post
                Furthermore can you write the RAM clock at the hardware chart at each benchmark you make not just the amount, I mean something like 7068MB@1600Mhz or something...
                All the charts are automatically created by PTS. If running phoronix-test-suite as root, on decent motherboards the DMI information exposes the memory DIMM configuration as well as the operating frequency, which in turn is parsed by PTS. However, I haven't seen any CAS latencies, etc available under Linux. Also wish it would work when not as root, since most users aren't benchmarking as root.
                Last edited by Michael; 07-05-2013, 08:00 PM.
                Michael Larabel
                http://www.michaellarabel.com/

                Comment


                • #23
                  Originally posted by Michael View Post
                  Any other game with automated test abilities works just fine, but after you tell a Steam game to quit from its console, even when it was launched directly from the command-line, the screen goes away and returns to the desktop, but the process actually remains alive. That's an issue with the upstream (steam).
                  So stop whining and do the tests manually like every Windows hardware review site on the planet does, it's not like there aren't hundreds of those posting far more in depth reviews then you do every week.

                  But I guess that would take too much time away from drinking over rated beers to bother doing what pays for that beer.

                  Comment


                  • #24
                    Originally posted by djdoo View Post
                    Thank you very much! So our APU was theoritically running at full throttle as far as it concerns the RAM speed! I don't speak for overclock speeds...
                    I've seen the 3D mark results of the 8670D being OCd to 1.2Ghz on DDR3 2.4Ghz, he said that at that speed the performance hit a wall due to lack of ram bandwidth, but did not say if he tried or even could push the ram speed higher for whatever reason.

                    Said results show that at those enough to keep pace with the Iris Pro 5200, with more ram speed it may even overtake it. Though many reviews are showing that the HD4600-Iris pro GPUs seem to have very high frame latency compared to the AMD APUs causing a noticeable stutter in gameplay.

                    Link
                    http://forums.anandtech.com/showthread.php?t=2323903

                    Comment


                    • #25
                      Originally posted by Kivada View Post
                      So stop whining and do the tests manually like every Windows hardware review site on the planet does, it's not like there aren't hundreds of those posting far more in depth reviews then you do every week.

                      But I guess that would take too much time away from drinking over rated beers to bother doing what pays for that beer.
                      Being a bit polite is not that bad these days, after all there are not many linux based benchmark sites around the web not to say there is not another one that offers a complete opensource benchmark suite like PTS for all linux users to run their own tests... So I find your comment at least unfair about Michael!

                      Comment


                      • #26
                        Originally posted by Kivada View Post
                        I've seen the 3D mark results of the 8670D being OCd to 1.2Ghz on DDR3 2.4Ghz, he said that at that speed the performance hit a wall due to lack of ram bandwidth, but did not say if he tried or even could push the ram speed higher for whatever reason.

                        Said results show that at those enough to keep pace with the Iris Pro 5200, with more ram speed it may even overtake it. Though many reviews are showing that the HD4600-Iris pro GPUs seem to have very high frame latency compared to the AMD APUs causing a noticeable stutter in gameplay.

                        Link
                        http://forums.anandtech.com/showthread.php?t=2323903
                        I never trusted Anandtech! If intel itself published their benchmarks they would have put lower values at the results!

                        Comment


                        • #27
                          Originally posted by Michael View Post
                          It's at 2133MHz as controlled from the BIOS/UEFI but the Catalyst driver as was reported to the result table reports 1066MHz.
                          Maybe you should ask the GPU-Z team to help you make PTS give more accurate information.

                          Comment


                          • #28
                            Originally posted by djdoo View Post
                            I never trusted Anandtech! If intel itself published their benchmarks they would have put lower values at the results!
                            Anand didn't do the test, some Chinese guy in their forum did. Though it's not like the results aren't reproducible. Just grab a mobo with a good Bios, some decent ram and a decent AM2/AM3/FM1 compatible CPU cooler, AMD has thankfully, unlike Intel kept the mount for their CPU coolers the same ever since the launch of AM2, start tinkering with the settings until you reach said results or possibly even better.

                            Since it's not a high end part or known to be a dirt cheap sleeper that can OC to absurd levels reliably on cheap air cooling there aren't allot of people out there trying to push the 6800K to it's absolute limits.

                            Comment


                            • #29
                              Originally posted by Kivada View Post
                              Maybe you should ask the GPU-Z team to help you make PTS give more accurate information.
                              It would likely require Linux kernel changes first to expose the DMI as non-root, otherwise it isn't too useful to begin with since most users aren't running as root. Just checked the code and in conditions where there is accurate DMI and PTS is run as root it will read the operating frequency, DIMM configuration, and RAM manufacturer / model (though not many systems get the manufacturer/model) but yeah in terms of memory latencies and other details, that would be nice but isn't a big ticket item and no commercial customers have requested such support either, but any patches are welcome.
                              Michael Larabel
                              http://www.michaellarabel.com/

                              Comment


                              • #30
                                Originally posted by Michael View Post
                                PTS doesn't break, it just sits there waiting for the process to be killed. Could add a kill command to the test profile but there isn't any sane way to do that unless just having it sit and wait for an arbitrary amount of time to kill the process, hoping the test completed in that amount of time.
                                i dont know much about php
                                cant you read when the process sends something like "benchmark complete" to stdout, then kill it ?

                                http://www.php.net/manual/en/function.proc-open.php
                                http://stackoverflow.com/questions/1...sing-proc-open

                                Comment

                                Working...
                                X