Given integral type A and non integral type B and depending on rounding
mode, optimization, compiler, and phase of the moon A(A)*B != A(A*B) so
split the two cases.
While at it, also make the template automagically work for complex types
instead of requiring manual casts, the general idea here is to allow
inlining and vectorization by treating all args as plain arrays, which is fine.
This works as expected with -tune=native, x64 implies sse2, and we do not
target any neon-less arm versions either.
Clang only array length hints can improve this even more.
Change-Id: I93f077f967daf2ed382d12cc20a54846b3688634
1)I did an experiment and compiled OpenBTS with clang yesterday, which
immediately highlighted two potential bugs in the Transceiver52 code.
I'm not sure they are indeed bugs and not the intended behavior, but
they look very much like that. The first one is below and the second
one is in the following mail.
GSM::Time() arguments are defined like #define USB_LATENCY_INTRVL
(10,0), which means that they are expanded into GSM::Time((10,0)).
This expression is a GSM::Time() with a single parameter where (10,0)
return value of the last argument, 0 in this case. I.e.
GSM::Time((10,0)) is equivalent to GSM::Time(0). I think this was not
the intention.
2) Printing \n after every complex number breaks output when you want to
print it in a single line, e.g. in many debug output.
I do not claim any copyright over this change, as it's very basic.
Looking forward to see it merged into mainline.
git-svn-id: http://wush.net/svn/range/software/public/openbts/trunk@4515 19bc5d8c-e614-43d4-8b26-e1612bc8e597