Add fastdivmod support for SYCL

WeiqunZhang · Feb 2, 2024 · 17ebf48 · 17ebf48
1 parent 1229ef4
commit 17ebf48
Showing 1 changed file with 15 additions and 2 deletions.
diff --git a/Src/Base/AMReX_Math.H b/Src/Base/AMReX_Math.H
@@ -209,6 +209,19 @@ constexpr T powi (T x) noexcept
     }
 }
 
+#if defined(AMREX_INT128_SUPPORTED)
+AMREX_GPU_DEVICE AMREX_FORCE_INLINE
+std::uint64_t umulhi (std::uint64_t a, std::uint64_t b)
+{
+#if defined(AMREX_USE_CUDA) || defined(AMREX_USE_HIP)
+    return __umul64hi(a, b);
+#else
+    auto tmp = amrex::UInt128_t(a) * amrex::UInt128_t(b);
+    return std::uint64_t(tmp >> 64);
+#endif
+}
+#endif
+
 /***************************************************************************************************
  * Copyright (c) 2017 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
  * SPDX-License-Identifier: BSD-3-Clause
@@ -312,11 +325,11 @@ struct FastDivmodU64
     AMREX_GPU_HOST_DEVICE
     std::uint64_t divide (std::uint64_t dividend) const
     {
-#if defined(AMREX_INT128_SUPPORTED) && (defined(AMREX_USE_CUDA) || defined(AMREX_USE_HIP))
+#if defined(AMREX_INT128_SUPPORTED)
         AMREX_IF_ON_DEVICE((
             auto x = dividend;
             if (multiplier) {
-                x = __umul64hi(dividend + round_up, multiplier);
+                x = amrex::Math::umulhi(dividend + round_up, multiplier);
             }
             return (x >> shift_right);
         ))