unknown
1970-01-01 00:00:00 UTC
No.
I don't have much of an understanding of these alignment issues. I do know
that Salvador Fandino had memory alignment issues with his Math::Int128
module and Windows. There's a thread about this at
http://www.perlmonks.org/?node_id=1052707
with the particular solution that salva used at
http://www.perlmonks.org/?node_id=1052819
At the end of that thread you'll see that I also struck similar issues with
64-bit MinGW builds of Math::Float128, for which I found that a similar
approach worked. Specifically, I did:
#ifdef __MINGW64__
typedef __float128 float128 __attribute__ ((aligned(8)));
#else
typedef __float128 float128;
#endif
and then wrote the functions in terms of the "float128" type instead of the
"__float128" C data type.
I don't know whether any of that applies in some way to the situation at
hand here.
If you think it doesn't, let me know and I'll make some enquiries - for I've
heard of nothing else (apart from the _aligned_malloc/_aligned_free option).
Cheers,
Rob
I don't have much of an understanding of these alignment issues. I do know
that Salvador Fandino had memory alignment issues with his Math::Int128
module and Windows. There's a thread about this at
http://www.perlmonks.org/?node_id=1052707
with the particular solution that salva used at
http://www.perlmonks.org/?node_id=1052819
At the end of that thread you'll see that I also struck similar issues with
64-bit MinGW builds of Math::Float128, for which I found that a similar
approach worked. Specifically, I did:
#ifdef __MINGW64__
typedef __float128 float128 __attribute__ ((aligned(8)));
#else
typedef __float128 float128;
#endif
and then wrote the functions in terms of the "float128" type instead of the
"__float128" C data type.
I don't know whether any of that applies in some way to the situation at
hand here.
If you think it doesn't, let me know and I'll make some enquiries - for I've
heard of nothing else (apart from the _aligned_malloc/_aligned_free option).
Cheers,
Rob