I don't know if this compiler bug is known, but any align(x) where x > 4
does not work.

Robert Atkinson wrote:

> I don't know if this compiler bug is known,
> but any align(x) where x > 4 does not work.

And this is a pity, since for instance
SIMD needs 16-byte (128-bit) alignment.

--anders

Unfortunately this is how I discovered the bug, while portings some of my SSE code from C++.

Anders F Björklund wrote:

> Robert Atkinson wrote:
> 
>> I don't know if this compiler bug is known,
>> but any align(x) where x > 4 does not work.
> 
> And this is a pity, since for instance
> SIMD needs 16-byte (128-bit) alignment.
> 
> --anders

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Robert Atkinson schrieb am Wed, 04 May 2005 12:36:49 +0000:
> I don't know if this compiler bug is known, but any align(x) where x > 4
> does not work.

Added to DStress as
http://dstress.kuehne.cn/run/align_11.d
http://dstress.kuehne.cn/run/align_12.d
http://dstress.kuehne.cn/run/align_13.d
http://dstress.kuehne.cn/run/align_14.d
http://dstress.kuehne.cn/run/align_15.d
http://dstress.kuehne.cn/run/align_16.d
http://dstress.kuehne.cn/run/align_17.d

Note: I don't have MSWindows/DMC or BSD|Solaris/GCC, thus I don't know what the expected results for those systems are.

Thomas


-----BEGIN PGP SIGNATURE-----

iD8DBQFCe2Jo3w+/yD4P9tIRAjNNAKDRJz3Kfx1ywQrx4tB+xnJSsN/OxQCgswc1
eQoXB5pDKSQYze4FM2IDZG8=
=L7V0
-----END PGP SIGNATURE-----