Skip to content

Commit 893b2f3

Browse files
committed
Split Accelerator.hpp into multiple tcc files, one for each accelerator API
1 parent 6c7c37e commit 893b2f3

8 files changed

Lines changed: 522 additions & 506 deletions

File tree

configure.ac

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -104,6 +104,14 @@ if test "x$enable_openmp" = "xyes"; then
104104
AC_MSG_FAILURE([OpenMP support was requested, but your compiler does not appear to support it])])
105105
fi
106106

107+
108+
if test "x$enable_cuda" = "xno" && test "x$enable_hip" = "xno" && test "x$enable_sycl" = "xno" && test "x$enable_openmp" = "xno" ; then
109+
AC_MSG_FAILURE([No accelerator API was provided (CUDA,HIP,SYCL,OpenMP)])
110+
fi
111+
112+
113+
114+
107115
# Check for the --enable-cublas option
108116
AC_ARG_ENABLE([cublas],
109117
[AS_HELP_STRING([--enable-cublas],

doc/installation.md

Lines changed: 11 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -1,11 +1,17 @@
11
## Dependencies
22

33
* MPI
4-
* OpenMP
5-
* CUDA, HIP, SYCL (Optional)
6-
* cuBLAS (CUDA), rocBLAS (HIP), oneMKL (SYCL) (Optional)
7-
8-
The non-GPU implementation is primarily for testing and development, and is not likely to achieve optimal performance
4+
* Accelerator API, one of:
5+
* CUDA
6+
* HIP
7+
* SYCL
8+
* OpenMP (CPU-threading only, for testing/development -- not performant!)
9+
* Host-side thread API (optional)
10+
* OpenMP
11+
* BLAS library for high-performance linear algebra (optional, recommended)
12+
* cuBLAS (CUDA)
13+
* rocBLAS (HIP)
14+
* oneMKL (SYCL)
915

1016
## Installation
1117

0 commit comments

Comments
 (0)