12 Replies Latest reply on Mar 29, 2015 1:46 AM by Cornugon

    3D Mark API Overhead test

    Cornugon

      Well I've done the overhead test which tests DX11, 12 and Mantle. But since the Draw calls for Mantle are that astronomically higher than those of DX11, is there something wrong with DX11 performance?

       

        • 3D Mark API Overhead test
          backFireX64

          No it is normal and shows the tremendous benefit of mantle and the dx12 api.

          See this article:

          http://www.guru3d.com/news-story/quick-test-directx-12-api-overhead-benchmark.html

            • 3D Mark API Overhead test
              Cornugon

              Ah nice, didn't see that article yet. And good to know DX12 is somewhere near Mantle's result. Promising indeed

                • 3D Mark API Overhead test
                  kingfish

                  I was under the impression that mantle benefited the mid range cards and processors. Doesn't look like that's the case. I downloaded an ran the new Mark...posted the reults in benchmarking. The Firestorm test is really demanding. I would be curious to see what a single 290x would score on that..could you run that Cornugon?

                    • 3D Mark API Overhead test
                      Cornugon

                      Originally posted by: stumped  I was under the impression that mantle benefited the mid range cards and processors. Doesn't look like that's the case. I downloaded an ran the new Mark...posted the reults in benchmarking. The Firestorm test is really demanding. I would be curious to see what a single 290x would score on that..could you run that Cornugon?</p>

                       

                      </end quote></div>

                      Already did a while back, so here's my single R9 290X score on a i7 4770:

                • 3D Mark API Overhead test
                  black_zion
                  Glad I sold my 5970 and bought this 7970, only cost me net about $50.
                    • 3D Mark API Overhead test
                      backFireX64

                      Originally posted by: black_zion  Glad I sold my 5970 and bought this 7970, only cost me net about $50.</end quote></div></p>

                      Sure, that is a more permanent "fix" i guess.

                      And a good choice also, cause you got rid of all the crossfirex side effects at the same time.

                        • 3D Mark API Overhead test
                          kingfish

                          BF4 on my computer runs a solid 60fps with mantle. BF 4 runs a solid 75 fps using DX 11. I wonder if mantle implements vert sync? Or, is the initial info on mantle correct...the benefits are more noticable on mid range graphics and CPU's? Same game settings and CCC settings.

                            • 3D Mark API Overhead test
                              Cornugon

                              Originally posted by: stumped  BF4 on my computer runs a solid 60fps with mantle. BF 4 runs a solid 75 fps using DX 11. I wonder if mantle implements vert sync? Or, is the initial info on mantle correct...the benefits are more noticable on mid range graphics and CPU's? Same game settings and CCC settings.</p>

                               

                              </end quote></div>

                              I'm going to check it out myself. I still have a week 'gametime' so why not use it now.

                              In my case I found that Dragon Age: Inquisition ran smoother using Mantle, aside from the ridiculous loadtimes, Thief (yeah I know different engine) has a higher fps too.

                                • 3D Mark API Overhead test
                                  kingfish

                                  Here are my in-game settings. All I did was change DX 11 to mantle. These never changed.

                                   

                                    • 3D Mark API Overhead test
                                      Cornugon

                                      I've tested this. I loaded up the map 'Test Range' and walked towards the vehicles in view. That short piece I got 110-120fps using DX11 and 130-140fps using Mantle. So even a 'high end' CPU gains an advantage.

                                      My graphics settings were at 2560x1440 (using DSR ofcourse), Ultra settings, MSAAx4, and a FOV of 70 (I hate the fish-eye view at high FOVs, but I don't like the in-your-face look of a low FOV either, so this is something which I'm comfortable at).

                                      Concerning the forced VSync using Mantle, did you try setting the CCC-option (yeah you know, the one that never works) to Off? Or 'Off, unless app specifies otherwise'?

                                        • 3D Mark API Overhead test
                                          kingfish

                                          Yes...I had it set to 'off unless'..normally I have it 'on unless' . Strange.

                                            • 3D Mark API Overhead test
                                              Cornugon

                                              Originally posted by: stumped  Yes...I had it set to 'off unless'..normally I have it 'on unless' . Strange.</p>

                                               

                                              </end quote></div>

                                              Yeah strange indeed. I ran into this 'issue' when trying to compare Thief Mantle results.

                                              We should file a report: 'CCC VSync setting actually works when using Mantle. Please fix this ASAP so that it no longer works, like intended, or I'm going nVidia'.

                                              Granted it only partially works, if set to 'On, unless...' it doesn't listen to the game's setting when set to off in-game, but still...