x86_64 extended the sse2 movnti instruction to support 64-bit integer registers
as well. i don't see any builtin for generating this... nor is there an
intrinsic listed in the intel manuals or apparently in icc 10.0.023 header
files either. the natural name would be _mm_stream_si64 -- which fortunately
does not conflict with _mm_stream_pi, the mmx 64-bit store version.
on a whim i tried this:
void foo(unsigned long *d, unsigned long v)
{
__builtin_ia32_movntq((unsigned long long *)d, v);
}
results in this code:
0000000000000000 <foo>:
0: 48 89 74 24 f8 mov %rsi,0xfffffffffffffff8(%rsp)
5: 0f 6f 44 24 f8 movq 0xfffffffffffffff8(%rsp),%mm0
a: 0f e7 07 movntq %mm0,(%rdi)
d: c3 retq
perhaps this builtin could be overloaded to generate the "movnti %rsi,(%rdi)"
directly instead of shuffling through the mmx reg file?
(note if i throw in -mtune=core2 it eliminates the trip through the stack)
-dean
/home/odo/gcc/bin/gcc -v
Using built-in specs.
Target: x86_64-unknown-linux-gnu
Configured with: ../gcc/configure --prefix=/home/odo/gcc --disable-multilib
--disable-biarch x86_64-unknown-linux-gnu --enable-languages=c
Thread model: posix
gcc version 4.3.0 20071029 (experimental) (GCC)
--
Summary: streaming 64-bit integer stores
Product: gcc
Version: 4.3.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: dean at arctic dot org
GCC build triplet: x86_64-unknown-linux-gnu
GCC host triplet: x86_64-unknown-linux-gnu
GCC target triplet: x86_64-unknown-linux-gnu
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33944
