I repeated 1) and 2) at least 10 times, but BoingCube did not fail once, I can't reproduce the bug, sorry.
No idea what is really going on, but it might be your gfx-card (or your setup) unless more people can confirm the bug with different setups, or something else not sure... :S
I'll try "some tricks" in the comming updates to go around this bug, hopefully it will help.
Anyway, thanks Raziel for your time reporting the bug
*** BoingCube benchmark *** Name: -benchmark10 (Counts iterations per ten seconds.) Version: 1.30 (20151219) Mode: Windowed NON-vsynced Hardware acc.: Buff: ON (hardware) Gfx: ON (hardware) Iterations: 1523 frames Duration: 10003 ms (10.003 seconds) Rendering time: 199 ms (0.199 seconds) [lower is better] D. frame-rate: 152.25432370289 FPS (acquired displayed frames/second) Result: 7653.2663316583 FPS (rendering frame-rate) [higher is better]
$ BoingCube -benchmark -gofullscreen
*** BoingCube benchmark *** Name: -benchmark10 (Counts iterations per ten seconds.) Version: 1.30 (20151219) Mode: Fullscreen NON-vsynced Hardware acc.: Buff: ON (hardware) Gfx: ON (hardware) Iterations: 1431 frames Duration: 10002 ms (10.002 seconds) Rendering time: 256 ms (0.256 seconds) [lower is better] D. frame-rate: 143.07138572286 FPS (acquired displayed frames/second) Result: 5589.84375 FPS (rendering frame-rate) [higher is better]
Edit: Hmm, when comparing to Severin's HD 5450, in fullscreen, why does mine have a faster rendering time (256 ms vs 271) but achieves fewer frames (1431 vs 1585) and hence a lower FPS (5590 vs 5849)?
And the same is - more or less - the case in windowed mode.
Here's my results with version 1.30, on Sam440ep-flex 800MHz Radeon 9250 with AOS4.1 FE.
1.> boingcube -benchmark
*** BoingCube benchmark *** Name: -benchmark10 (Counts iterations per ten seconds.) Version: 1.30 (20151219) Mode: Windowed NON-vsynced Hardware acc.: Buff: ON (hardware) Gfx: ON (hardware) Iterations: 618 frames Duration: 10011 ms (10.011 seconds) Rendering time: 145 ms (0.145 seconds) [lower is better] D. frame-rate: 61.732094695835 FPS (acquired displayed frames/second) Result: 4262.0689655172 FPS (rendering frame-rate) [higher is better]
1.> boingcube -benchmark -gofullscreen
*** BoingCube benchmark *** Name: -benchmark10 (Counts iterations per ten seconds.) Version: 1.30 (20151219) Mode: Fullscreen NON-vsynced Hardware acc.: Buff: ON (hardware) Gfx: ON (hardware) Iterations: 563 frames Duration: 10001 ms (10.001 seconds) Rendering time: 186 ms (0.186 seconds) [lower is better] D. frame-rate: 56.294370562944 FPS (acquired displayed frames/second) Result: 3026.8817204301 FPS (rendering frame-rate) [higher is better]
1.> boingcube -benchmark -noaccgfx -noaccbuffer
*** BoingCube benchmark *** Name: -benchmark10 (Counts iterations per ten seconds.) Version: 1.30 (20151219) Mode: Windowed NON-vsynced Hardware acc.: Buff: OFF (software) Gfx: OFF (software) Iterations: 556 frames Duration: 10012 ms (10.012 seconds) Rendering time: 819 ms (0.819 seconds) [lower is better] D. frame-rate: 55.533359968038 FPS (acquired displayed frames/second) Result: 678.87667887668 FPS (rendering frame-rate) [higher is better]
With 1.20 I got frame rate: 54, rendering frame rate: 624
The last one with -noaccgfx -noaccbuffer options is like the previous versions with "only" software graphics. Now hardware acc. graphics is used by default.
Could it be the resolution you're using, especially for windowed mode as the gfx card still has to churn out the rest of the display?
I'm using 1920x1080x32
or morelikely something you have running in the background
Duh, yes, of course. Most likely both of the above.
My resolution is 2560×1600×32.
Here's a new set of results, where I have rebooted without running WBStartup:
$ BoingCube -benchmark
*** BoingCube benchmark *** Name: -benchmark10 (Counts iterations per ten seconds.) Version: 1.30 (20151219) Mode: Windowed NON-vsynced Hardware acc.: Buff: ON (hardware) Gfx: ON (hardware) Iterations: 1576 frames Duration: 10013 ms (10.013 seconds) Rendering time: 140 ms (0.14 seconds) [lower is better] D. frame-rate: 157.3953859982 FPS (acquired displayed frames/second) Result: 11257.142857143 FPS (rendering frame-rate) [higher is better]
$ BoingCube -benchmark -gofullscreen
*** BoingCube benchmark *** Name: -benchmark10 (Counts iterations per ten seconds.) Version: 1.30 (20151219) Mode: Fullscreen NON-vsynced Hardware acc.: Buff: ON (hardware) Gfx: ON (hardware) Iterations: 1562 frames Duration: 10004 ms (10.004 seconds) Rendering time: 133 ms (0.133 seconds) [lower is better] D. frame-rate: 156.13754498201 FPS (acquired displayed frames/second) Result: 11744.360902256 FPS (rendering frame-rate) [higher is better]
Hmm, that's quite an improvement there. I guess the resolution isn't really a big factor compared to whatever I had running in the background (e.g. AmiDock, LimpidClock, Odyssey, just to mention a few that have graphics implications).