I wanted to do some quick time measurements of some things on a driver
running of the 405GP and I eneded up using get_tbl(). It seems like
get_cycles should be the more standard way to get a tick count but
it seems to return 0. I don't get the __ftr_fixup stuff in timex.h
so I am not sure what it is trying to do differently from get_tbl.
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
I wanted to do some quick time measurements of some things on a driver
running of the 405GP and I eneded up using get_tbl(). It seems like
get_cycles should be the more standard way to get a tick count but
it seems to return 0. I don't get the __ftr_fixup stuff in timex.h
so I am not sure what it is trying to do differently from get_tbl.
The __ftr_fixup in timex.h adds a reference to this code in the
feature fixup section. This allow the code that access the TB to
be nop'ed out on CPUs that don't implement it (601)
Ben.
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
andrew may wrote:
I wanted to do some quick time measurements of some things on a driver
running of the 405GP and I eneded up using get_tbl(). It seems like
get_cycles....
Hmmm....these look like they should work exactly the same.....
It would help me if you could disassemble a little code snippet
around the place you use these so we could have a look.
Thanks.
-- Dan
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/