Merge pull request #1714 from NVIDIA/u128_div

fix uint128
This commit is contained in:
Dustyn Blasig
2024-08-16 07:14:59 -05:00
committed by GitHub

View File

@@ -194,7 +194,7 @@ struct alignas(16) uint128_t
uint64_t remainder{0};
#if defined(CUTLASS_UINT128_NATIVE)
remainder = uint64_t(native % divisor);
#elif defined(CUTLASS_INT128_ARITHMETIC_DIV)
#elif defined(CUTLASS_INT128_ARITHMETIC_DIV) && ! defined (__CUDA_ARCH__)
// implemented using MSVC's arithmetic intrinsics
(void)_udiv128(hilo_.hi, hilo_.lo, divisor, &remainder);
#else