[OPENMP][NVPTX]Enable fast shuffles on 64bit values only if CUDA >= 9.
Summary:
Shuffle on 64bit data is allowed only for CUDA >= 9.0. Also, fixed the
constant for the mask, need one extra L in the end.
Reviewers: gtbercea, kkwli0
Subscribers: guansong, caomhin, openmp-commits
Differential Revision: https://reviews.llvm.org/D55440
git-svn-id: https://llvm.org/svn/llvm-project/openmp/trunk@348758 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/libomptarget/deviceRTLs/nvptx/src/reduction.cu b/libomptarget/deviceRTLs/nvptx/src/reduction.cu
index a05a6e0..2546302 100644
--- a/libomptarget/deviceRTLs/nvptx/src/reduction.cu
+++ b/libomptarget/deviceRTLs/nvptx/src/reduction.cu
@@ -76,7 +76,17 @@
}
EXTERN int64_t __kmpc_shuffle_int64(int64_t val, int16_t delta, int16_t size) {
- return __SHFL_DOWN_SYNC(0xFFFFFFFFFFFFFFFFL, val, delta, size);
+#if defined(CUDART_VERSION) && CUDART_VERSION >= 9000
+ return __SHFL_DOWN_SYNC(0xFFFFFFFFFFFFFFFFLL, (long long)val, (unsigned)delta,
+ (int)size);
+#else
+ int lo, hi;
+ asm volatile("mov.b64 {%0,%1}, %2;" : "=r"(lo), "=r"(hi) : "l"(val));
+ hi = __SHFL_DOWN_SYNC(0xFFFFFFFF, hi, delta, size);
+ lo = __SHFL_DOWN_SYNC(0xFFFFFFFF, lo, delta, size);
+ asm volatile("mov.b64 %0, {%1,%2};" : "=l"(val) : "r"(lo), "r"(hi));
+ return val;
+#endif
}
static INLINE void gpu_regular_warp_reduce(void *reduce_data,
