Toon Moene wrote:
Richard Guenther wrote:
It's a matter of making the cbrt builtin available - I have a patch
for this,
Oh, BTW, my own version of this patch (plus your work in the area of
sqrt) had the following effect on a profile breakdown (every routine
above 2 %) of the forecasting code in HIRLAM (http://hirlam.org) - just
look at the %; the times are not comparable, as you can see from the
number of calls of each routine:
Before:
% cumulative self self total
time seconds seconds calls Ks/call Ks/call name
14.29 755.79 755.79 116160 0.00 0.00 verint_
12.66 1425.23 669.43 __ieee754_powf
5.50 1716.19 290.97 116160 0.00 0.00 bixint_
4.96 1978.62 262.42 121 0.00 0.03 sl2tim_
4.94 2239.98 261.36 15730 0.00 0.00 hlcondcv_
4.26 2465.30 225.32 15730 0.00 0.00 hlvcbr_
3.62 2656.71 191.41 15730 0.00 0.00 hlradia_
2.85 2807.19 150.48 __ieee754_logf
2.64 2946.87 139.68 15730 0.00 0.00 phtask_
2.34 3070.80 123.93 1452 0.00 0.00 invlo4_
2.25 3189.55 118.75 __ieee754_expf
2.23 3307.65 118.10 fesetround
2.21 3424.45 116.80 fesetenv
2.08 3534.31 109.86 15730 0.00 0.00 hlprevap_
After:
% cumulative self self total
time seconds seconds calls Ks/call Ks/call name
14.22 575.66 575.66 93120 0.00 0.00 verint_
9.83 973.59 397.93 __ieee754_powf
5.84 1210.14 236.55 93120 0.00 0.00 bixint_
5.14 1418.12 207.98 97 0.00 0.03 sl2tim_
5.05 1622.62 204.50 12610 0.00 0.00 hlcondcv_
4.28 1795.91 173.29 12610 0.00 0.00 hlvcbr_
3.64 1943.48 147.56 12610 0.00 0.00 hlradia_
3.12 2069.66 126.19 __ieee754_logf
2.77 2181.96 112.30 12610 0.00 0.00 phtask_
2.39 2278.56 96.60 1164 0.00 0.00 invlo4_
2.34 2373.12 94.56 fesetround
2.26 2464.71 91.59 fesetenv
2.13 2551.07 86.36 12610 0.00 0.00 hlprevap_
2.06 2634.31 83.24 12610 0.00 0.00 phys_
2 Ghz Athlon 64, AMD64 Debian testing, gfortran 4.3.0.
--
Toon Moene - e-mail: [EMAIL PROTECTED] - phone: +31 346 214290
Saturnushof 14, 3738 XG Maartensdijk, The Netherlands
A maintainer of GNU Fortran: http://gcc.gnu.org/fortran/
Who's working on GNU Fortran:
http://gcc.gnu.org/ml/gcc/2006-01/msg00000.html
