Open GoogleCodeExporter opened 8 years ago
Built with crosstool12-gcc-4.3.1 on my desktop (harpertown core2), pybench
reports:
Test minimum run-time average run-time
this other diff this other diff
Totals: 5103ms 5204ms -1.9% 5246ms 5323ms -1.4%
(this=trunk/build32_ct12/pybench.out,
other=only_why_reports_exceptions/build32_ct12/pybench.out)
Test minimum run-time average run-time
this other diff this other diff
Totals: 4249ms 4308ms -1.4% 4329ms 4387ms -1.3%
(this=trunk/build64_ct12/pybench.out,
other=only_why_reports_exceptions/build64_ct12/pybench.out)
(i.e. the patch is slower)
For 2to3, I get:
trunk64:
real 0m24.847s
user 0m24.030s
sys 0m0.104s
trunk32:
real 0m26.955s
user 0m26.346s
sys 0m0.128s
why64:
real 0m24.143s
user 0m23.681s
sys 0m0.080s
why32:
real 0m27.587s
user 0m26.942s
sys 0m0.064s
(i.e. 32-bit is ~2.2% slower; 64-bit is ~1.5% faster) 2to3 run times vary by
about
.3s of user time = 1%, so the differences are significant.
Original comment by jyass...@gmail.com
on 11 Dec 2008 at 12:40
LGTM. We can live with the slow-dow since this makes things so much simpler
conceptually.
Original comment by collinw
on 6 Jan 2009 at 11:00
Original issue reported on code.google.com by
jyass...@gmail.com
on 8 Dec 2008 at 12:17