Re: GMP 3.1.1 Alpha(21164) Linux(Debian 2.2) problems

[Kevin Ryde]

Fat Phil <address@hidden> writes:
>
> Problems and fortunately a solution to one too.
> 
> Firstly, on my Alpha, /proc/cpuinfo has its info in a different form, it
> uses "BogoMIPS", rather than "bogomips". Hence in tune/time.c, the
> following changes (or similar) need to be applied: (orig_time.c being
> the distributed code)

Yes, you're quite right, we noticed this too, it'll be fixed in the
next release.  In fact the next release will also have a scheme to
measure the frequency if there's no value from the kernel.


> The second one I've not had a chance to look at yet (hey, I only
> downloaded it an hour ago!).
> 
> address@hidden:/usr/local/src/gmp-3.1.1/tune$ ./speed -s 349059 -C
> mpn_add_n
> overhead 9.03 cycles, precision 10000 units of 1.88e-09 secs, cycle
> 1.9e-09
>             mpn_add_n
> 349059        45.8807
> address@hidden:/usr/local/src/gmp-3.1.1/tune$ ./speed -s 349060 -C
> mpn_add_n
> overhead 9.03 cycles, precision 10000 units of 1.88e-09 secs, cycle
> 1.9e-09
>             mpn_add_n
> Segmentation fault

The data for these measurements is allocated on the stack, so you'll
have about 3*349060*8 ~= 8Mbytes in that test.  Perhaps you need a
higher "ulimit -s".