I did some more testing and am now calculating mean delta times for each second as well as the standard deviation of the delta times. Here's what i get for the space racer above:
01-27 15:11:42.274: DEBUG/AFX(2504): fps: 56.82312, delta: 0.016143799, mean delta: 0.01756897, stddev. delta: 0.0026706476, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:43.274: DEBUG/AFX(2504): fps: 54.986576, delta: 0.016479492, mean delta: 0.018019613, stddev. delta: 0.0038374658, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:44.290: DEBUG/AFX(2504): fps: 50.92457, delta: 0.03729248, mean delta: 0.019476827, stddev. delta: 0.0064565083, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:45.298: DEBUG/AFX(2504): fps: 53.9243, delta: 0.015136719, mean delta: 0.019114176, stddev. delta: 0.00692833, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:46.306: DEBUG/AFX(2504): fps: 52.600296, delta: 0.012298585, mean delta: 0.018918864, stddev. delta: 0.0053105806, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:47.306: DEBUG/AFX(2504): fps: 52.914413, delta: 0.01171875, mean delta: 0.018682353, stddev. delta: 0.005447507, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:48.337: DEBUG/AFX(2504): fps: 51.17586, delta: 0.030395508, mean delta: 0.019417318, stddev. delta: 0.0067782626, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:49.337: DEBUG/AFX(2504): fps: 53.930874, delta: 0.031311035, mean delta: 0.019150289, stddev. delta: 0.0061283736, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:50.345: DEBUG/AFX(2504): fps: 54.986576, delta: 0.017272947, mean delta: 0.01829478, stddev. delta: 0.0044108317, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:51.360: DEBUG/AFX(2504): fps: 55.01943, delta: 0.026489256, mean delta: 0.018075053, stddev. delta: 0.0037128378, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:52.376: DEBUG/AFX(2504): fps: 55.93651, delta: 0.026336668, mean delta: 0.017899577, stddev. delta: 0.003501554, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:53.384: DEBUG/AFX(2504): fps: 56.657646, delta: 0.016845703, mean delta: 0.017754618, stddev. delta: 0.0029748674, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:54.384: DEBUG/AFX(2504): fps: 55.919792, delta: 0.016937256, mean delta: 0.017803447, stddev. delta: 0.002930975, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:55.391: DEBUG/AFX(2504): fps: 54.774338, delta: 0.016662598, mean delta: 0.018299866, stddev. delta: 0.0052955016, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:56.407: DEBUG/AFX(2504): fps: 51.882175, delta: 0.016479492, mean delta: 0.018996684, stddev. delta: 0.005628621, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:57.415: DEBUG/AFX(2504): fps: 55.93684, delta: 0.0223999, mean delta: 0.018104045, stddev. delta: 0.0036228807, #listeners: 1, #meshes: 35, #textures: 2 01-27 15:11:58.415: DEBUG/AFX(2504): fps: 51.946102, delta: 0.011444091, mean delta: 0.019133504, stddev. delta: 0.0067592976, #listeners: 1, #meshes: 35, #textures: 2 The delta time varries between 3-6ms from frame to frame! That explains the stuttering pretty well. Now i wonder why i get such inconsistent delta times. I removed rendering the background tiles so i only render the ship which is a lousy 50 triangles big, no textures or lighting. Very strange indeed. On 27 Jan., 14:23, Mario Zechner <[email protected]> wrote: > @Ralf > that is very strange behaviour i wouldn't expect from a Nexus One. I > have to admit that i only own a Droid so i can't confirm it for the > N1. Would you mind putting an apk with your test application up > somewhere, for example athttp://www.file-pasta.com, and add log cat > output so we can observe the behaviour of your app on our devices? > Additionally, if it's not to much of a hazzle you could also share the > source code so we can have a look at it. > > On 27 Jan., 12:42, Ralf Schneider <[email protected]> wrote: > > > I definitely have this problem on the Nexus One. > > > MyOpenGLperformance test (using JNI/NDK pumped by java GLSurfaceView) is > > drawing 3 rotating Quads(8-8-8 texture) - all covering the whole screen - > > with blending. > > And I get very constant 60 FPS. As soon as I touch the screen: Frame rate > > drops to 43 FPS until i release the finger, than it gets back to 60 FPS. > > > I think the garbage collection is the remaining reason for the slow down on > > Android > 2.0. > > > The problem is less visible in a more complex scene where the initial frame > > rate is around 35 FPS. Touching the screen in this case does not change the > > frame rate significantly. > > > 2010/1/27 Robert Green <[email protected]> > > > > @MrChaz > > > From my tests on 1.5 and 1.6, Thread.sleep() does not even come close > > > to adequately fixing the problem. I have tried ranges from 16-1000 > > > and touch'n'hold still causes my game to drop from 30FPS down to 15FPS > > > or lower on the G1. *The game suffers no such problems on the Droid > > > and Nexus One*. I've also taken out everything but the call to sleep -- You received this message because you are subscribed to the Google Groups "Android Developers" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/android-developers?hl=en
