After reading all this, I just round the temps to the nearest degree, which I think is good enough for my purposes.I found that doing this in bash was so slow that it triggered the CPU scaling governor and possibly influenced the temps, so I changed to badly written, but very lightweight, C:No need to reopen the sysfs file in the loop - seeking back seems to work fine.
Also, for interest, you can get the ARM and CORE freq's with one single call to vcgencmd
"vcgencmd measure_clock arm core".
Code:
Time CPU Core Vcore Temp....Min....Max Health10:31:43 2800 910 0.9996 48 48 48 OK10:31:53 500 500 0.720 44 44 48 OK10:32:03 500 500 0.720 44 44 48 OK10:32:13 500 500 0.720 44 44 48 OK10:32:23 500 500 0.720 45 44 48 OK10:32:33 500 500 0.720 45 44 48 OK10:32:43 2800 910 0.9996 62 44 62 Fan210:32:54 2800 910 0.9996 66 44 66 Fan210:33:04 2800 910 0.9996 69 44 69 Fan310:33:14 2800 910 0.9996 70 44 70 Fan310:33:24 2800 910 0.9996 70 44 70 Fan310:33:34 2800 910 0.9996 72 44 72 Fan3Code:
// TEMPERATURE "12345\n"read(temp, in, 3);lseek(temp, 0, SEEK_SET);cur_temp = (in[0] - 48U) * 10U + in[1] - 48U;if( in[2] > '4' ) ++cur_temp;if( cur_temp < min_temp ) min_temp = cur_temp;if( cur_temp > max_temp ) max_temp = cur_temp;p[6] = chr(cur_temp / 10);p[7] = chr(cur_temp % 10);p[13] = chr(min_temp / 10);p[14] = chr(min_temp % 10);p[20] = chr(max_temp / 10);p[21] = chr(max_temp % 10);Also, for interest, you can get the ARM and CORE freq's with one single call to vcgencmd
"vcgencmd measure_clock arm core".
Statistics: Posted by jahboater — Mon Feb 05, 2024 10:42 am