You may wish to follow the issue 2794 discussion at http://code.google.com/p/android/issues/detail?id=2794 and consider the use of addCallbackBuffer() and setPreviewCallbackWithBuffer().
Regards On Jun 6, 8:53 pm, Jay Ha <nicejae...@gmail.com> wrote: > Hi. > > I'm writing an Augmented Reallity codes that should works in realtime, > > But the VM's garbage collection hinders my program's performance. > > I checked my program's execution time for each in OnPrevFrame(byte[], > Camera) > > and it execution time increases from around 50ms to 120ms at just > after each garbage collection. > > I copied parts of logs below. > (local time is just before and after of signal processing, and global > time is time stamps for every onPreviewFrame() calling ) > ----------------------------------------------------------------------------------------------------------------------------------------- > 06-06 18:42:26.340: WARN/matchTarget(725): global 151844ms, local > 167ms > 06-06 18:42:26.390: WARN/matchTarget(725): global 55ms, local 54ms > 06-06 18:42:26.510: DEBUG/dalvikvm(725): GC_FOR_MALLOC freed 20 > objects / 346432 bytes in 88ms > 06-06 18:42:26.510: WARN/matchTarget(725): global 119ms, local > 115ms > 06-06 18:42:26.600: WARN/matchTarget(725): global 83ms, local 75ms > 06-06 18:42:26.670: WARN/matchTarget(725): global 76ms, local 70ms > 06-06 18:42:26.740: WARN/matchTarget(725): global 66ms, local 63ms > 06-06 18:42:26.820: DEBUG/dalvikvm(725): GC_FOR_MALLOC freed 73 > objects / 464096 bytes in 55ms > 06-06 18:42:26.830: WARN/matchTarget(725): global 95ms, local 94ms > 06-06 18:42:26.890: WARN/matchTarget(725): global 57ms, local 56ms > 06-06 18:42:26.940: WARN/matchTarget(725): global 46ms, local 45ms > 06-06 18:42:27.000: WARN/matchTarget(725): global 59ms, local 59ms > 06-06 18:42:27.070: DEBUG/dalvikvm(725): GC_FOR_MALLOC freed 144 > objects / 466776 bytes in 56ms > 06-06 18:42:27.080: WARN/matchTarget(725): global 80ms, local 79ms > 06-06 18:42:27.150: WARN/matchTarget(725): global 71ms, local 65ms > 06-06 18:42:27.210: WARN/matchTarget(725): global 61ms, local 60ms > 06-06 18:42:27.250: WARN/matchTarget(725): global 41ms, local 39ms > 06-06 18:42:27.310: WARN/matchTarget(725): global 57ms, local 41ms > 06-06 18:42:27.350: DEBUG/dalvikvm(725): GC_FOR_MALLOC freed 207 > objects / 584128 bytes in 39ms > 06-06 18:42:27.420: WARN/matchTarget(725): global 116ms, local > 59ms > 06-06 18:42:27.470: WARN/matchTarget(725): global 50ms, local 49ms > 06-06 18:42:27.530: WARN/matchTarget(725): global 58ms, local 55ms > 06-06 18:42:27.620: DEBUG/dalvikvm(725): GC_FOR_MALLOC freed 210 > objects / 353680 bytes in 58ms > 06-06 18:42:27.630: WARN/matchTarget(725): global 96ms, local 95ms > 06-06 18:42:27.690: WARN/matchTarget(725): global 61ms, local 57ms > 06-06 18:42:27.740: WARN/matchTarget(725): global 57ms, local 55ms > 06-06 18:42:27.820: WARN/matchTarget(725): global 76ms, local 68ms > 06-06 18:42:27.870: WARN/matchTarget(725): global 49ms, local 46ms > 06-06 18:42:27.900: DEBUG/dalvikvm(725): GC_FOR_MALLOC freed 226 > objects / 469944 bytes in 34ms > 06-06 18:42:27.970: WARN/matchTarget(725): global 97ms, local 55ms > 06-06 18:42:28.000: WARN/matchTarget(725): global 39ms, local 38ms > 06-06 18:42:28.080: WARN/matchTarget(725): global 80ms, local 61ms > 06-06 18:42:28.170: DEBUG/dalvikvm(725): GC_FOR_MALLOC freed 220 > objects / 469184 bytes in 50ms > 06-06 18:42:28.170: WARN/matchTarget(725): global 86ms, local 84ms > 06-06 18:42:28.230: WARN/matchTarget(725): global 60ms, local 52ms > 06-06 18:42:28.280: WARN/matchTarget(725): global 52ms, local 51ms > 06-06 18:42:28.340: WARN/matchTarget(725): global 61ms, local 59ms > 06-06 18:42:28.420: DEBUG/dalvikvm(725): GC_FOR_MALLOC freed 193 > objects / 468200 bytes in 53ms > 06-06 18:42:28.430: WARN/matchTarget(725): global 82ms, local 81ms > 06-06 18:42:28.490: WARN/matchTarget(725): global 64ms, local 57ms > 06-06 18:42:28.550: WARN/matchTarget(725): global 61ms, local 60ms > 06-06 18:42:28.610: WARN/matchTarget(725): global 57ms, local 56ms > 06-06 18:42:28.690: DEBUG/dalvikvm(725): GC_FOR_MALLOC freed 205 > objects / 468832 bytes in 58ms > 06-06 18:42:28.700: WARN/matchTarget(725): global 88ms, local 86ms > 06-06 18:42:28.750: WARN/matchTarget(725): global 58ms, local 55ms > 06-06 18:42:28.810: WARN/matchTarget(725): global 54ms, local 53ms > 06-06 18:42:28.860: WARN/matchTarget(725): global 52ms, local 51ms > 06-06 18:42:28.990: DEBUG/dalvikvm(725): GC_FOR_MALLOC freed 226 > objects / 469544 bytes in 42ms > 06-06 18:42:28.990: WARN/matchTarget(725): global 134ms, local > 100ms > 06-06 18:42:29.060: WARN/matchTarget(725): global 62ms, local 52ms > 06-06 18:42:29.100: WARN/matchTarget(725): global 49ms, local 47ms > 06-06 18:42:29.160: WARN/matchTarget(725): global 51ms, local 50ms > 06-06 18:42:29.240: DEBUG/dalvikvm(725): GC_FOR_MALLOC freed 197 > objects / 468320 bytes in 52ms > 06-06 18:42:29.240: WARN/matchTarget(725): global 84ms, local 83ms > 06-06 18:42:29.290: WARN/matchTarget(725): global 51ms, local 43ms > 06-06 18:42:29.350: WARN/matchTarget(725): global 53ms, local 39ms > 06-06 18:42:29.410: WARN/matchTarget(725): global 69ms, local 44ms > 06-06 18:42:29.470: WARN/matchTarget(725): global 54ms, local 44ms > ----------------------------------------------------------------------------------------------------------------------------------------- > > Also I executed onPreviewFrame without any signal processing on it, > but > > the garbage collection is also carried showing debegging message > around 60ms it spends for memory release. > > Is this garbage collection come from releasing byte[] yuvs image data? > ( i uses 320x240 thumb nail ) > > then... this clealy comes from Java's limitation, so there will be no > hope for improvement unless > > I found a way to directly access camera device? do you think this is > feasible? > > Hope listen to others' comments. -- You received this message because you are subscribed to the Google Groups "Android Developers" group. To post to this group, send email to android-developers@googlegroups.com To unsubscribe from this group, send email to android-developers+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/android-developers?hl=en
