On Sat, Jun 15, 2013 at 2:26 PM, Charles R Harris <charlesr.har...@gmail.com > wrote:
> > > On Sat, Jun 15, 2013 at 2:23 PM, Warren Weckesser < > warren.weckes...@gmail.com> wrote: > >> >> On Sat, Jun 15, 2013 at 4:03 PM, Julian Taylor < >> jtaylor.deb...@googlemail.com> wrote: >> >>> On 15.06.2013 21:57, Warren Weckesser wrote: >>> > >>> > On Sat, Jun 15, 2013 at 3:15 PM, Julian Taylor >>> >>> > @warren, can you please bisect the commit causing this? >>> > >>> > >>> > >>> > Here's the culprit: >>> > >>> > aef286debfd11a62f1c337dea55624cee7fd4d9e is the first bad commit >>> > commit aef286debfd11a62f1c337dea55624cee7fd4d9e >>> > Author: Julian Taylor <jtaylor.deb...@googlemail.com >>> > <mailto:jtaylor.deb...@googlemail.com>> >>> > Date: Mon Jun 10 19:38:58 2013 +0200 >>> > >>> > ENH: enable unaligned loads on x86 >>> > >>> > x86 can handle unaligned load and there is no hand vectorized code >>> in >>> > this file. It would be a serious compiler bug if it adds >>> vectorization >>> > without checking for alignment. >>> > Enables fast complex128 copies which are unaligned on 32 bit gcc >>> unless >>> > compiled with -malign-double. >>> > >>> > :040000 040000 d0948d1e1d942d41d50ce9e57bdc430db9a16f9e >>> > 45a48f383353857b8d0dd24e542c7ab6f137448c M numpy >>> > >>> > Link: >>> > >>> https://github.com/numpy/numpy/commit/aef286debfd11a62f1c337dea55624cee7fd4d9e >>> > >>> >>> Interesting, possibly there is some inconsistencies in the macros in >>> this file. >>> But I don't understand why I can't reproduce it. >>> Does it happen with python 3.2 too? >>> >> >> >> Yes, it happens with 3.3.2, 3.3.1, and 3.2.5. >> > > Also on 2.7. The -O3 flag seems to cause the problem. > > But I see it on test_3d_shaped_dtype. Chuck
_______________________________________________ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion