here are new test results from
* EVM @5V 2A * Kernel letux-4.8-rc7 (not yet published) * Ethernet connected and active * USB-memory stick plugged in. * Runs ./high-load but not other activities (except networking daemons like nntpd, blinking LEDs)
1. omap5-uevm+pyra4.dtb, Pyra Mainboard V4 connected, Panel with SSD2858 connected (working incl. rotation)
root@letux:~# ./high-load Tue Sep 20 14:53:41 UTC 2016 51672 54868 50600 Tue Sep 20 14:53:43 UTC 2016 51672 54421 51000 Tue Sep 20 14:53:45 UTC 2016 65304 57996 55400 Tue Sep 20 14:53:46 UTC 2016 67860 60118 57800 Tue Sep 20 14:53:48 UTC 2016 70416 62017 59400 Tue Sep 20 14:53:49 UTC 2016 70842 63357 60200 Tue Sep 20 14:53:51 UTC 2016 72120 64698 61000 Tue Sep 20 14:53:53 UTC 2016 72120 64251 61000 Tue Sep 20 14:53:54 UTC 2016 72120 65145 61800 Tue Sep 20 14:53:56 UTC 2016 73398 65591 61800 Tue Sep 20 14:53:57 UTC 2016 74356 66485 62200 Tue Sep 20 14:53:59 UTC 2016 73824 66038 62600 Tue Sep 20 14:54:01 UTC 2016 74356 66932 63000 Tue Sep 20 14:54:02 UTC 2016 73824 67379 63400 Tue Sep 20 14:54:04 UTC 2016 74995 66932 63800 Tue Sep 20 14:54:05 UTC 2016 74356 66932 63800 Tue Sep 20 14:54:07 UTC 2016 74995 67379 63800 Tue Sep 20 14:54:09 UTC 2016 74995 68272 64200 Tue Sep 20 14:54:10 UTC 2016 74995 68272 64200 Tue Sep 20 14:54:12 UTC 2016 75954 68272 64600
It took 31 seconds to reach 75°C.
did hang after 48 minutes
Tue Sep 20 15:41:12 UTC 2016 99810 92399 86600 Tue Sep 20 15:41:14 UTC 2016 100236 91953 86200 Tue Sep 20 15:41:16 UTC 2016 99384 92399 86200 Tue Sep 20 15:41:17 UTC 2016 100236 92846 86600 Tue Sep 20 15:41:19 UTC 2016 100236 92846 86600 Tue Sep 20 15:41:21 UTC 2016 91290 90165 83800 Tue Sep 20 15:41:22 UTC 2016 89160 89272 82200 Tue Sep 20 15:41:24 UTC 2016 98532 91506 85000 Tue Sep 20 15:41:25 UTC 2016 98532 91953 85000 Tue Sep 20 15:41:27 UTC 2016 98532 91953 85000 Tue Sep 20 15:41:29 UTC 2016 98958 91953 85000 Tue Sep 20 15:41:30 UTC 2016 99810 91953 86200 Tue Sep 20 15:41:32 UTC 2016 99810 92846 86200 Tue Sep 20 15:41:33 UTC 2016 100236 91953 86200 Tue Sep 20 15:41:35 UTC 2016 100662 92399 86200 Tue Sep 20 15:41:37 UTC 2016 100236 92399 85800 Tue Sep 20 15:41:38 UTC 2016 100236 92846 85800 Tue Sep 20 15:41:40 UTC 2016 100662 92846 86600
Intersting are the temperature sawtooth cycles! With approx. 10-20 seconds cycle time. Which means that the internal over-temperature protection is activated, something cools down and heats up again.
One thing I had never seen is that temperature rises beyond 100°C and the thermo-shutdown does not happen any more!
Who is limiting the CPU temperature to 100°C? Some kernel mechanism?
2. omap5-uevm+pyra4.dtb, Pyra Mainboard V4 disconnected, Panel with SSD2858 connected (not working)
Did run for 2 hours w/o hanging.
root@letux:~# ./high-load Tue Sep 20 16:17:52 UTC 2016 49968 53975 50200 Tue Sep 20 16:17:54 UTC 2016 54228 53975 50600 Tue Sep 20 16:17:56 UTC 2016 64878 57996 55400 Tue Sep 20 16:17:57 UTC 2016 67008 60677 57000 Tue Sep 20 16:17:59 UTC 2016 67008 60677 57000 Tue Sep 20 16:18:00 UTC 2016 68286 62017 57800 Tue Sep 20 16:18:02 UTC 2016 69990 62464 59000 Tue Sep 20 16:18:03 UTC 2016 70416 62911 59400 Tue Sep 20 16:18:05 UTC 2016 70842 63357 60200 Tue Sep 20 16:18:07 UTC 2016 70842 63804 60600 Tue Sep 20 16:18:08 UTC 2016 71694 64251 60600 Tue Sep 20 16:18:10 UTC 2016 72120 64698 60600 Tue Sep 20 16:18:11 UTC 2016 72120 64698 61400 Tue Sep 20 16:18:13 UTC 2016 72972 65145 61400 Tue Sep 20 16:18:15 UTC 2016 73398 65145 62200 Tue Sep 20 16:18:16 UTC 2016 72972 66485 61800 Tue Sep 20 16:18:18 UTC 2016 74356 66038 62200 Tue Sep 20 16:18:19 UTC 2016 73824 66038 62600 Tue Sep 20 16:18:21 UTC 2016 73824 66932 62600 Tue Sep 20 16:18:23 UTC 2016 74356 66932 63000 Tue Sep 20 16:18:24 UTC 2016 74356 66485 63400 Tue Sep 20 16:18:26 UTC 2016 74356 67379 63800 Tue Sep 20 16:18:27 UTC 2016 74356 67825 64200 Tue Sep 20 16:18:29 UTC 2016 74995 67379 63400 Tue Sep 20 16:18:31 UTC 2016 74995 67825 63800 Tue Sep 20 16:18:32 UTC 2016 75528 67379 64200
Here it took 40 seconds to reach 75°C.
ca. 17 minutes after starting high-load:
Tue Sep 20 16:34:24 UTC 2016 98532 91506 85400 Tue Sep 20 16:34:25 UTC 2016 98958 91953 85400 Tue Sep 20 16:34:27 UTC 2016 98958 91059 84600 Tue Sep 20 16:34:28 UTC 2016 98958 91059 85000 Tue Sep 20 16:34:30 UTC 2016 99384 91953 85400 Tue Sep 20 16:34:32 UTC 2016 98958 91506 85400 Tue Sep 20 16:34:33 UTC 2016 98958 91059 85000
Interestingly, the CPU-Temperature remained almost constant here! And did not (yet) run into protection. No sawtooth.
Some time later sawtooths started but slower than in scenario. Ca. 1 minute.
2 hours later...
Tue Sep 20 18:13:00 UTC 2016 99384 91953 85800 Tue Sep 20 18:13:02 UTC 2016 98958 92399 85800 Tue Sep 20 18:13:04 UTC 2016 99810 91953 85400 Tue Sep 20 18:13:05 UTC 2016 99384 91953 85800 Tue Sep 20 18:13:07 UTC 2016 99810 91953 85800 Tue Sep 20 18:13:08 UTC 2016 100236 91953 85800 Tue Sep 20 18:13:10 UTC 2016 99810 92399 85800 Tue Sep 20 18:13:12 UTC 2016 100236 91953 85800 ^Ckill 2566 2567
root@letux:~#
So this means that really running MIPI/DSS makes the OMAP draw more power. So that it heats up faster.
And generally this draw less power than scenario 1.
3. omap5-uevm.dtb, Pyra Mainboard V4 disconnected, Panel with SSD2858 connected (not working)
root@letux:~# ./high-load Tue Sep 20 18:33:03 UTC 2016 41874 45150 41800 Tue Sep 20 18:33:05 UTC 2016 41874 45150 43100 Tue Sep 20 18:33:07 UTC 2016 56358 48166 47000 Tue Sep 20 18:33:09 UTC 2016 58914 50847 49000 Tue Sep 20 18:33:10 UTC 2016 60618 52187 50600 Tue Sep 20 18:33:12 UTC 2016 61470 53081 50200 Tue Sep 20 18:33:14 UTC 2016 61896 53975 51000 Tue Sep 20 18:33:15 UTC 2016 62748 53528 52200 Tue Sep 20 18:33:17 UTC 2016 63174 54868 52600 Tue Sep 20 18:33:19 UTC 2016 64026 54868 53800 Tue Sep 20 18:33:21 UTC 2016 64026 57102 54200 Tue Sep 20 18:33:22 UTC 2016 64452 57102 54200 Tue Sep 20 18:33:24 UTC 2016 65304 57102 54200 Tue Sep 20 18:33:26 UTC 2016 65304 57102 55000 Tue Sep 20 18:33:28 UTC 2016 65304 57549 55000 Tue Sep 20 18:33:29 UTC 2016 65304 57996 55400 Tue Sep 20 18:33:31 UTC 2016 66156 58443 55900 Tue Sep 20 18:33:33 UTC 2016 66156 59448 55900 Tue Sep 20 18:33:34 UTC 2016 66582 58889 55900 Tue Sep 20 18:33:36 UTC 2016 66582 59448 56500 Tue Sep 20 18:33:38 UTC 2016 67434 59448 55900 Tue Sep 20 18:33:40 UTC 2016 67434 59448 56500 Tue Sep 20 18:33:41 UTC 2016 67434 60677 57800 Tue Sep 20 18:33:43 UTC 2016 68286 61570 58200 Tue Sep 20 18:33:45 UTC 2016 67860 61123 58200 Tue Sep 20 18:33:47 UTC 2016 67860 61570 58600 Tue Sep 20 18:33:48 UTC 2016 68286 61123 58600 Tue Sep 20 18:33:50 UTC 2016 69564 61570 58600 Tue Sep 20 18:33:52 UTC 2016 68286 62464 58600 Tue Sep 20 18:33:54 UTC 2016 69564 62464 58600 Tue Sep 20 18:33:55 UTC 2016 68712 62017 59000 Tue Sep 20 18:33:57 UTC 2016 70842 62464 59800 Tue Sep 20 18:33:59 UTC 2016 70416 62464 59400 Tue Sep 20 18:34:00 UTC 2016 70416 63357 59400 Tue Sep 20 18:34:02 UTC 2016 70842 62911 59800 Tue Sep 20 18:34:04 UTC 2016 70416 62911 59400 Tue Sep 20 18:34:06 UTC 2016 70842 63804 60200 Tue Sep 20 18:34:07 UTC 2016 71694 63357 61000 Tue Sep 20 18:34:09 UTC 2016 71268 64251 60600 Tue Sep 20 18:34:11 UTC 2016 71694 64251 61000 Tue Sep 20 18:34:12 UTC 2016 71268 63357 60600 Tue Sep 20 18:34:14 UTC 2016 72120 64698 61000 Tue Sep 20 18:34:16 UTC 2016 72546 64251 61400 Tue Sep 20 18:34:18 UTC 2016 72972 64698 61800 Tue Sep 20 18:34:19 UTC 2016 73398 65591 61800 Tue Sep 20 18:34:21 UTC 2016 72972 65145 62200 Tue Sep 20 18:34:23 UTC 2016 73398 66038 61800 Tue Sep 20 18:34:24 UTC 2016 73398 65591 62200 Tue Sep 20 18:34:26 UTC 2016 73398 66038 62200 Tue Sep 20 18:34:28 UTC 2016 73824 66038 63000 Tue Sep 20 18:34:30 UTC 2016 73398 66485 63000 Tue Sep 20 18:34:31 UTC 2016 74356 66038 62600 Tue Sep 20 18:34:33 UTC 2016 74356 66038 63000 Tue Sep 20 18:34:35 UTC 2016 73824 66485 63000 Tue Sep 20 18:34:37 UTC 2016 73824 66485 63000 Tue Sep 20 18:34:38 UTC 2016 74356 66932 63800 Tue Sep 20 18:34:40 UTC 2016 74356 66932 64200 Tue Sep 20 18:34:42 UTC 2016 74995 66485 63800 Tue Sep 20 18:34:43 UTC 2016 75528 67379 63800
So it takes 100 seconds to reach 75°C.
Tue Sep 20 19:54:37 UTC 2016 99384 91953 85400 Tue Sep 20 19:54:39 UTC 2016 98958 91506 85800 Tue Sep 20 19:54:40 UTC 2016 99384 91953 85800 Tue Sep 20 19:54:42 UTC 2016 99384 91953 85400 Tue Sep 20 19:54:44 UTC 2016 100236 91506 85400 Tue Sep 20 19:54:46 UTC 2016 99384 91506 85000 Tue Sep 20 19:54:48 UTC 2016 90438 90165 83000 Tue Sep 20 19:54:49 UTC 2016 98106 90165 84600 Tue Sep 20 19:54:51 UTC 2016 98106 90612 84600 Tue Sep 20 19:54:53 UTC 2016 98532 91506 85400 Tue Sep 20 19:54:54 UTC 2016 99384 91059 85000 Tue Sep 20 19:54:56 UTC 2016 98958 91506 85400 Tue Sep 20 19:54:58 UTC 2016 98958 91506 85400 Tue Sep 20 19:55:00 UTC 2016 98958 91059 85400 Tue Sep 20 19:55:01 UTC 2016 98958 91059 85400 Tue Sep 20 19:55:03 UTC 2016 98532 91506 85000 Tue Sep 20 19:55:05 UTC 2016 98532 91953 85400 Tue Sep 20 19:55:07 UTC 2016 98958 91953 85400 Tue Sep 20 19:55:08 UTC 2016 98958 91506 85400 Tue Sep 20 19:55:10 UTC 2016 99384 91059 85400 ^Ckill 2592 2593
root@letux:~#
Did not hang within 1.5 hours.
4. omap5-uevm.dtb, Pyra Mainboard V4 connected, Panel with SSD2858 connected (not working because missing in DT files)
root@letux:~# ./high-load Wed Sep 21 04:39:46 UTC 2016 49116 52634 49400 Wed Sep 21 04:39:48 UTC 2016 49542 53081 48600 Wed Sep 21 04:39:50 UTC 2016 64026 57102 54200 Wed Sep 21 04:39:52 UTC 2016 67008 58889 55900 Wed Sep 21 04:39:53 UTC 2016 67860 60677 57400 Wed Sep 21 04:39:55 UTC 2016 69564 61570 58200 Wed Sep 21 04:39:57 UTC 2016 70842 62464 59800 Wed Sep 21 04:39:59 UTC 2016 70416 63357 59400 Wed Sep 21 04:40:00 UTC 2016 71268 63357 59400 Wed Sep 21 04:40:02 UTC 2016 71694 63804 60200 Wed Sep 21 04:40:04 UTC 2016 71268 63804 61000 Wed Sep 21 04:40:05 UTC 2016 72120 64698 60600 Wed Sep 21 04:40:07 UTC 2016 72120 64698 61000 Wed Sep 21 04:40:09 UTC 2016 72546 64698 61400 Wed Sep 21 04:40:10 UTC 2016 73398 65145 61800 Wed Sep 21 04:40:12 UTC 2016 72972 65145 61800 Wed Sep 21 04:40:14 UTC 2016 73398 65591 62200 Wed Sep 21 04:40:16 UTC 2016 73824 66932 62600 Wed Sep 21 04:40:17 UTC 2016 73824 65591 63000 Wed Sep 21 04:40:19 UTC 2016 73824 66485 63000 Wed Sep 21 04:40:21 UTC 2016 74995 66932 63000 Wed Sep 21 04:40:23 UTC 2016 73824 66932 63800 Wed Sep 21 04:40:24 UTC 2016 74356 67379 63800 Wed Sep 21 04:40:26 UTC 2016 74356 67825 63400 Wed Sep 21 04:40:28 UTC 2016 75954 67825 64200 Wed Sep 21 04:40:29 UTC 2016 75528 67379 63800 Wed Sep 21 04:40:31 UTC 2016 75954 68272 64200 Wed Sep 21 04:40:33 UTC 2016 76380 68272 64200
So that is 42 seconds until 75°C.
Wed Sep 21 06:44:53 UTC 2016 100236 91953 85800 Wed Sep 21 06:44:55 UTC 2016 90438 90165 83400 Wed Sep 21 06:44:56 UTC 2016 89160 88378 81800 Wed Sep 21 06:44:58 UTC 2016 98106 90165 84200 Wed Sep 21 06:45:00 UTC 2016 98532 91059 85000 Wed Sep 21 06:45:01 UTC 2016 98532 90612 85000 Wed Sep 21 06:45:03 UTC 2016 99384 91953 85400 Wed Sep 21 06:45:05 UTC 2016 98958 91953 85400 Wed Sep 21 06:45:06 UTC 2016 98958 91506 85400 Wed Sep 21 06:45:08 UTC 2016 98958 91953 85800 Wed Sep 21 06:45:10 UTC 2016 99810 92399 85800 Wed Sep 21 06:45:12 UTC 2016 99384 91506 85400 Wed Sep 21 06:45:13 UTC 2016 99810 91059 85800 Wed Sep 21 06:45:15 UTC 2016 99810 91953 85400 Wed Sep 21 06:45:17 UTC 2016 99810 91059 85400 Wed Sep 21 06:45:18 UTC 2016 99810 91953 85400 Wed Sep 21 06:45:20 UTC 2016 99810 91953 85800 Wed Sep 21 06:45:22 UTC 2016 99384 91953 85400 Wed Sep 21 06:45:24 UTC 2016 99384 91953 85400 Wed Sep 21 06:45:25 UTC 2016 99810 92399 85400 Wed Sep 21 06:45:27 UTC 2016 99384 92399 85400 Wed Sep 21 06:45:29 UTC 2016 99384 91953 85800 Wed Sep 21 06:45:31 UTC 2016 100236 91953 85000 Wed Sep 21 06:45:32 UTC 2016 89586 89719 82600 Wed Sep 21 06:45:34 UTC 2016 88308 87485 81800 Wed Sep 21 06:45:36 UTC 2016 97254 90165 83800 Wed Sep 21 06:45:38 UTC 2016 98106 90612 84200 Wed Sep 21 06:45:39 UTC 2016 98106 91059 84600 Wed Sep 21 06:45:41 UTC 2016 98958 91506 85000 Wed Sep 21 06:45:43 UTC 2016 98532 91059 85400 ^Ckill 2590 2591
root@letux:~#
Also did run for 2 hours. With approx. 30-40 seconds sawtooth.
A little unexpected is that it appears to suffice to connect the Pyra mainboard (which is not configured by DT but powered by Palmas) to heat up the OMAP faster.
5. Discussion
So this clearly shows that the heat production in the CPU depends on the device tree file and if the mainboard is connected or not.
We have 31 sec with pyra mainboard and pyra DT 40 sec without pyra mainboard and pyra DT 42 sec with pyra mainboard and uevm DT 100 sec without pyra mainboard and uevm DT
This is of course to be expected since the uevm.dtb does not know anything about panels and Pyra peripherals and therefore does not enable anything - and does not have to process polling loops etc.
A key conclusion is that the OMAP5 temperature is not a problem per se.
But if we assume that the scenario 1 has the highest energy consumption, it explains why it is more risky to run into power supply limits. So again we have not gained a really new insight that a too weak power supply makes trouble.
BR, Nikolaus