mitch_de Posted March 21, 2013 Author Share Posted March 21, 2013 COOL , first bench results (lots) for new AMD 7950 Mac by Rob - barefeats!! http://www.barefeats.com/gpu7950.html Link to comment Share on other sites More sharing options...
Slice Posted March 22, 2013 Share Posted March 22, 2013 I will give you my result when I return home. But I see blackosx also beat me in Furmark test So 5770 > 6670 really 5770 = 6770 > 6670 ! @Alex009988: deactivate AA for ATI graphics. OpenGL Extension Viewer just crashes every time I try to run the OpenGL 3 tests. Any idea why that happens? Every other graphics test I've tried works fine. Update OpenGL Extension Viewer. Or just download more recent version. Link to comment Share on other sites More sharing options...
Maniac10 Posted March 22, 2013 Share Posted March 22, 2013 Update OpenGL Extension Viewer. Or just download more recent version. It is the latest from the AppStore, version 4.0.8 (7). Really strange but I suspect it's just the app, I still haven't bumped into another one crashing because of OpenGL. @blackosx: you seem to have the same graphics card I have, can you run OpenGL Extension Viewer's core tests? Link to comment Share on other sites More sharing options...
mitch_de Posted March 22, 2013 Author Share Posted March 22, 2013 I tested OGLV also with Core setting / Nvidia 9600 GT. Crashes also. I am sure with OS X up to 10.8.2 it didn't crash using that core / 3.1,2 OpenGL tests - only 3.3+ wasnt benched because my gpu hasnt the features. Perhaps some prob with 4.0.8 & 10.8.3 - homepage shows 4.0.10 - http://www.realtech-...ew/download.php but the App Store wasnt already updated (stays at version 4.0.8) - no other way to get the download Same happend with version 4.0.9 - Feb 2013 - never came up in the App Store. Perhaps Apple has some problem with that? Or works Apple very slow updating freeware stuff in the app Store - no money to get ?! Link to comment Share on other sites More sharing options...
blackosx Posted March 22, 2013 Share Posted March 22, 2013 @blackosx: you seem to have the same graphics card I have, can you run OpenGL Extension Viewer's core tests? v4.0.8 from the App Store crashes for me also if I run the core tests. v4.0.0 works with the core tests. EDIT: v4.0.0 file from link removed from my dropbox 1 Link to comment Share on other sites More sharing options...
Ion Darie Posted March 22, 2013 Share Posted March 22, 2013 Hello guys, Here are my results, the specs of my "mac" are in the signature. [bASIC] Quality= Medium Stereo 3D = Disabled Anti-aliasing = Off Resolution = 1280x720 [EXTREME] Quality= Ultra Stereo 3D = Disabled Anti-aliasing = Off Resolution = 1280x720 Link to comment Share on other sites More sharing options...
Maniac10 Posted March 22, 2013 Share Posted March 22, 2013 v4.0.8 from the App Store crashes for me also if I run the core tests. v4.0.0 works with the core tests. Thanks, that one works. So it was the app after all, I suspected that much. @mitch_de: I've noticed the crashing issue with OGLV since 10.8.2 but it was probably due to the version from the AppStore (it's dated 27/09/2012). Link to comment Share on other sites More sharing options...
Slice Posted March 24, 2013 Share Posted March 24, 2013 v4.0.8 from the App Store crashes for me also if I run the core tests. v4.0.0 works with the core tests. EDIT: v4.0.0 file from link removed from my dropbox Hmm, my version also is 4.0.0. The best version? Link to comment Share on other sites More sharing options...
mitch_de Posted March 24, 2013 Author Share Posted March 24, 2013 Version diff is MOSTLY only adding new Mac Modell identfier and (less) changes in the html result layout. So no major code changes for the benchparts itself. For result i think version doesnt matter - so as longs as the bench results stay compareable its no prob to use older version. Link to comment Share on other sites More sharing options...
k3nny Posted April 10, 2013 Share Posted April 10, 2013 Results with build 12E30 and unmodified presets Basic and Extreme, both with AA: According to barefeats the 7950 reaches an average of 33 fps with preset Extreme, so my result seems right. Link to comment Share on other sites More sharing options...
fau7i Posted April 12, 2013 Share Posted April 12, 2013 Guys, does enabling iGPU make any sense in benchmarking? Link to comment Share on other sites More sharing options...
mitch_de Posted April 12, 2013 Author Share Posted April 12, 2013 What is iGPU? Link to comment Share on other sites More sharing options...
k3nny Posted April 12, 2013 Share Posted April 12, 2013 fau7i means the integrated GPU, in his case the Intel HD 4000 which comes with the i5-3570k. Link to comment Share on other sites More sharing options...
fau7i Posted April 13, 2013 Share Posted April 13, 2013 Yes, exactly as both GPUs stated in the result. Link to comment Share on other sites More sharing options...
yoyostring Posted April 28, 2013 Share Posted April 28, 2013 My scores: Windows gives a much better result: Windows OpenGL: Updated: Mac Scores after switching to Mac Pro 3,1 from iMac 8,1 (Huge Improvement!) Link to comment Share on other sites More sharing options...
Slice Posted April 28, 2013 Share Posted April 28, 2013 Windows gives a much better result: Because this test is native for Windows while make OpenGL emulation through DirectX in OSX. Link to comment Share on other sites More sharing options...
mitch_de Posted April 28, 2013 Author Share Posted April 28, 2013 My scores: Basic give the same score as Extreme. There must be a bottleneck somewhere... Windows gives a much better result: Again, The enigine runs native OpenGL and not DIrectX wrapper > OpenGL. You may try to set OpenGL (API Setting) in WIndows to compare OpenGL vs OpenGL As every complex (more real game) CPU does matter, so C2D is the bottleneck for your fast gpu (compared to C2D). My much slower 9600 GT has much more FPS diff between BASIC and EXTREME using C2D, because in this case GPU is the bottleneck. Link to comment Share on other sites More sharing options...
yoyostring Posted April 28, 2013 Share Posted April 28, 2013 I added the Windows OpenGL scores to my post above. I am just sort of surprised that my OSX scores are so much worse than the other scores on this topic despite the decent GPU. So far the only game that I have so far that is significantly CPU limited on this rig is Batman: Arkham City, most games run over 60FPS on high quality settings. (Windows though, I haven't done much OSX gaming on this system) But, yeah the mobo and CPU are due for an upgrade. Link to comment Share on other sites More sharing options...
mitch_de Posted April 29, 2013 Author Share Posted April 29, 2013 You see, OpenGL Win is less fps that DirectX Win. How much depends on OpenGL driver speeds - other gpu may have different diff. CPU/Systembus speeds has often effect, at least on newer highend games. Games magz test gpus always with fastest possible i7 + fast DDR3 ram to avoid much cpu dependences. Some magz disabled (for test) some of the cpu cores of Quad CPU or i7 and most new games run much slower (with fast gpu) running 2 cores vs 3 or 4 cores. Thats the negative side of such game near featured benches vs. more syntetic benches, which have much less cpu load than such benches or real games. Like grass is moved by the cpu of course .... Link to comment Share on other sites More sharing options...
yoyostring Posted April 30, 2013 Share Posted April 30, 2013 After switching my smbios to MacPro 3,1 from iMac 8,1 my scores improved dramatically! (I updated the above screen shots) Even my GeekBench score improved from 3000 to 4200 which is better than I get in Windows (3600) Link to comment Share on other sites More sharing options...
mitch_de Posted April 30, 2013 Author Share Posted April 30, 2013 Yep, you see now diff between Windows OpenGL and OS X OpenGL is very less! No major speed diff = Unique Engine runs native OpenGL and not any Direktx wrapper for OpenGl. Otherwise the OS X OpenGL result would be much lower. Link to comment Share on other sites More sharing options...
mitch_de Posted May 4, 2013 Author Share Posted May 4, 2013 Also http://www.g-truc.net , wich test monthly OpenGL featues (by an set of OpenGL samles using newest (as OpenGL spezif tells) features) compared Unique Engine OS X and there is no significant speed diff between Win OpenGL and OS X Opengl - sometimes (gpu type) OS X is little better, sometimes Win. Direktx vs OpenGL is an other case Also interesting , they test monthly the different OpenGL features in "real programming life". Apple stopped adding OpenGL features at OpenGL 3.2. All newer features are not supported or only a few of them = incomplete = not usable. http://www.g-truc.ne...-0566.html#menu April 2013 test ( compared to march 2013, you see AMD implemented / debugged some of the 4.3 features = more % in that samples case) % means how many of the OpenGL test code works or not. March 2013 test: Compared to older monthly tests, Win OpenGL makes big steps to get much more OpenGL features working bugfree (or even impletemt them compared to apple). Apples % stays same of many months (doesn't matter 10.8.1, 10.8.2 or 10.8.3 : same OpenGL 3.2 barrier ) That OS X OpenGL 3.2 "barrier" doesn't mean there are no OpenGL 3.3, 4.0 features in the Apple OpenGL - OpenGl ExtensionsViewer can show that - but they are uncompleted ( Using newer OpenGL features (sure the bug free ones ) means : Some features an be used , like tessellation or not. But much more interesting, new features reduce cpu usage load (using OpenGL commands) much or use faster working OpenGL commands. Its like using SSE(2/3/4) commands of the cpu or using workarounds to do same stuff but using instead one sse command lots of "normal" cpu commands in an row. 1 Link to comment Share on other sites More sharing options...
haxzion Posted May 7, 2013 Share Posted May 7, 2013 10.8.4 + GTX660 Link to comment Share on other sites More sharing options...
colorfuel Posted October 9, 2013 Share Posted October 9, 2013 1280x720 / 1920x1080 no AA Medium Windowed .... AMD 6870 12D68 Classical CPU bottleneck : Link to comment Share on other sites More sharing options...
Slice Posted October 9, 2013 Share Posted October 9, 2013 mitch_de OK, you convinced me that the test uses real OpenGL. But there is still great CPU dependency influences on final score. 1 Link to comment Share on other sites More sharing options...
Recommended Posts