[SOLVED] clock_gettime problem (moving from ARM to MIPS)

elonica.pl · 07-14-2011, 03:31 AM

Hi,

I'm porting C++ application from ARM to MIPS System. For some scheduled work
clock_gettime(CLOCK_MONOTONIC,&res) is used.

On Mips this routine is 10 times slower then on ARM. Where could I check what is wrong ?

BR
Bartek

Mara · 07-17-2011, 02:14 PM

Multiple reasons are possible: syscall cost, general system cost. If the length is very long, you need to do more detailed profiling. Out of couriosity, what time scale we're talking about here?

elonica.pl · 07-18-2011, 05:24 AM

Hi,

Thank you for answer.
I've opened second thread that might be connected to this problem (is more detailed):
http://www.linuxquestions.org/questi...55#post4417855

Time scale is < 500 ms. I have 400 MHz clock on board which should handle it without problems.

elonica.pl · 07-19-2011, 07:37 AM

Ok. Problem solved. I've forced HZ = 1024.

It seems to be quite lot for embedded device (it has 400 MHz). But works with this setting.