On 26/11/14 10:14, Christophe Lyon wrote:
Hi Kyrill,
Hi Christophe,
On 21 November 2014 at 16:52, Marcus Shawcroft
<marcus.shawcr...@gmail.com> wrote:
On 17 November 2014 17:35, Kyrill Tkachov <kyrylo.tkac...@arm.com> wrote:
2014-11-17 Kyrylo Tkachov <kyrylo.tkac...@arm.com>
* config/aarch64/arm_neon.h (vsqrt_f64): New intrinsic.
2014-11-17 Kyrylo Tkachov <kyrylo.tkac...@arm.com>
* gcc.target/aarch64/simd/vsqrt_f64_1.c
OK /Marcus
Your new test fails at the scan-assembly step because all the code is
optimized away (even at -O1).
Sorry about that, I could have sworn I saw it pass when I initially
wrote it...
In any case, I've committed this patch (r218117) as obvious to mark one
of the variables as volatile
to make sure it's not optimised away.
I've confirmed that the scan-assembler test fails without this patch and
passes with it.
Thanks,
Kyrill
Christophe.
diff --git a/gcc/testsuite/gcc.target/aarch64/simd/vsqrt_f64_1.c b/gcc/testsuite/gcc.target/aarch64/simd/vsqrt_f64_1.c
index 57fb6bb..7f99bd5 100644
--- a/gcc/testsuite/gcc.target/aarch64/simd/vsqrt_f64_1.c
+++ b/gcc/testsuite/gcc.target/aarch64/simd/vsqrt_f64_1.c
@@ -11,7 +11,7 @@ extern void abort (void);
int
main (void)
{
- float64x1_t in = vcreate_f64(0x3febd3e560634d7bULL);
+ volatile float64x1_t in = vcreate_f64(0x3febd3e560634d7bULL);
float64x1_t result = vsqrt_f64 (in);
float64_t expected = 0.9325321502142351;
