cory-master/BabelStream
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
603 Commits 1 Branch 0 Tags 2.5 MiB
683b8fcf88
Commit Graph

603 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Tom Deakin
683b8fcf88 update README 2021-02-18 13:51:24 +00:00
Tom Deakin
13dfff45a6
Merge pull request #90 from UoB-HPC/generalise-run 
Generalise run function
 2021-02-18 13:18:38 +00:00
Tom Deakin
ba47571ab2 update changelog 2021-02-18 12:37:58 +00:00
Tom Deakin
3deb9f8eff Generalise the run functions to share construction of models and check vectors 
An enum is added and set by the command line options to choose the running mode:
all 5 kernels or triad only.

There is now only one run fuction which call the constructor of each
model implementation, calling another routine to run the kernels.
The output is then determined by the enum value.
 2021-02-18 12:35:12 +00:00
Tom Deakin
018d8a4510 [OpenCL] Remove dot kernel object in deconstructor 2021-02-02 15:45:54 +00:00
Tom Deakin
30231575cb Tidy CUDA memory mode Makefile 2021-02-02 12:33:18 +00:00
Tom Deakin
5d697fdfe9 Add missing OpenMP flag to Intel CPU builds 2021-02-02 11:49:16 +00:00
Tom Deakin
f99f8d35d9 Revert "Add nstream kernel from PRK" 
This reverts commit 1e94a41f3c.
 2021-02-02 11:25:27 +00:00
Tom Deakin
877f820282 Revert "Update README with nstream citations" 
This reverts commit cb0c345ad5.
 2021-02-02 11:25:14 +00:00
Tom Deakin
cb0c345ad5 Update README with nstream citations 2021-02-02 11:24:41 +00:00
Tom Deakin
1e94a41f3c Add nstream kernel from PRK 
PRK has a nstream kernel, which is Triad with a += update.
This means there are 3 reads and a write, which is a higher
read/write ratio. In addition, non-temporal stores for the
write on CPUs will not be beneficial, and so compilers should
take care to emit these for the other kernels, but not these.
 2021-02-01 17:41:30 +00:00
Tom Deakin
435a104f6e Check input array size is positive 2021-01-12 15:30:41 +00:00
Tom Deakin
903eb40d19 Add parseInt function for parsing CLI arguments for array size 2021-01-12 10:28:01 +00:00
Tom Deakin
15001000c5 use signed int for array size in RAJA 2021-01-12 10:25:45 +00:00
Tom Deakin
87ab797490 use signed ints for HC 2021-01-12 10:25:16 +00:00
Tom Deakin
66aaec281f use signed ints for HIP 2021-01-12 10:24:27 +00:00
Tom Deakin
9a69d3d5d9 use signed array size for SYCL 2021-01-12 10:24:00 +00:00
Tom Deakin
20c3284629 Update CHANGELOG with signed int change 2021-01-12 10:23:21 +00:00
Tom Deakin
d01b46a87a use signed ints for STD C++20 2021-01-12 10:22:53 +00:00
Tom Deakin
ecc47f5320 use signed ints for STD C++17 2021-01-12 10:22:29 +00:00
Tom Deakin
94c7c3dbd8 use signed array size for OpenCL 2021-01-12 10:21:48 +00:00
Tom Deakin
693a7e7478 use signed array size for CUDA 2021-01-12 10:20:44 +00:00
Tom Deakin
850c63d69b use signed ints for ACC array size 2021-01-12 10:14:44 +00:00
Tom Deakin
e6c200a2d3 use signed int for Kokkos array size 2021-01-12 10:13:53 +00:00
Tom Deakin
00de932454 Save array size argument as signed integer 2021-01-12 10:09:55 +00:00
Tom Deakin
a9fd663471 Make OpenMP array size signed 2021-01-12 10:04:51 +00:00
Tom Deakin
4abb080a0e Fix GCC AMD build for OpenMP offload 2020-12-30 14:40:21 +00:00
Tom Deakin
cf42335e7a Merge branch 'cuda-memory' into main 2020-12-07 15:15:37 +00:00
Tom Deakin
9c211bca96 Update changelog for CUDA memory mode 2020-12-07 15:13:06 +00:00
Tom Deakin
e8fb3a6be4 Add C++20 version using for_each_n and range factories 
Closes #85
 2020-12-07 14:55:54 +00:00
Tom Deakin
ffa221fd35 Fix OpenMP Clang NVIDIA Target flags (missing sm architecture) with new NVARCH option 
Example usage:
make -f OpenMP.make COMPILER=CLANG TARGET=NVIDIA NVARCH=sm_61

Fixes #61
 2020-12-07 12:23:11 +00:00
Tom Deakin
5a93022fc1 Update OpenACC for Issue #80 2020-12-07 11:50:20 +00:00
Tom Deakin
b00120d346 Update STD C++17 for Issue #80 2020-12-07 11:32:22 +00:00
Tom Deakin
74f705cac9 Update OpenMP for Issue #80 2020-12-07 10:41:48 +00:00
Tom Deakin
829aa15da0 Allocate driver solution check vectors *after* the main computation 
Each Stream implementation owns its own data, so the driver code
shouldn't allocate a large array just before. On processors with
strong NUMA effects and smaller memory capacities per NUMA domain,
these checking vectors can result in the main arrays being
allocated in the wrong NUMA domain.

The fix is to simply move the driver allocation until after the
computation has finished and we want to check the answers.

This commit only changes the driver; each model will be updated
in subsequent commits.

Fixes #80.
 2020-12-07 10:39:37 +00:00
Tom Deakin
f373927ce8 Rename branch name 2020-12-07 10:23:27 +00:00
Tom Deakin
f271d5563d
Merge pull request #84 from gonzalobg/cxx_parallel_stl 
Add NVIDIA HPC SDK C++ parallel STL implementation
 2020-12-03 14:15:45 +00:00
Gonzalo Brito Gadeschi
0855805ce2 Add NVIDIA HPC SDK C++ parallel STL implementation 
This commits adds an implementation using the C++ parallel STL.
The Makefile uses the NVIDIA HPC SDK `nvc++` compiler with the `-stdpar` flag.
Tested using the NVIDIA HPC SDK 20.9.
 2020-11-23 03:08:44 -08:00
Tom Deakin
5182342403
Update CHANGELOG.md 2020-10-26 09:58:57 +00:00
Tom Deakin
8ae8c70188
Merge pull request #81 from Kerilk/master 
Ensure OpenCL destructors are called in the "correct" order.
 2020-10-26 09:58:05 +00:00
Brice Videau
e92d034f64 Ensure OpenCL destructors are called in the correct order. 2020-10-16 18:05:23 -05:00
Tom Deakin
6f46267e6c Add AOMP build options 2020-08-13 17:46:45 +01:00
Tom Deakin
66d915fa2e
[OpenMP] Fix ARMCLANG Makefile bug where it didn't set the flags 2020-08-12 15:39:13 +01:00
Tom Deakin
f31181dedb
Add -O3 flat to HIP.make to fix segmentation fault 2020-08-12 14:09:22 +01:00
Tom Deakin
da3946a7d5 Add missing O3 flag for OpenMP ARMCLANG 2020-08-07 17:09:46 +01:00
Tom Deakin
0ff841bbf5
Update CHANGELOG.md 2020-08-07 12:29:28 +01:00
Tom Deakin
17f057c38a
Merge pull request #79 from tom91136/master 
Update build flags for SYCL, Kokkos, and OpenMP, tracking newest versions of each compiler
 2020-08-07 12:27:43 +01:00
Tom Lin
cdaf6cb88e Fixed a bug where ComputeCpp's flags is omitted 
Renamed INTEL_GT -> INTEL_GPU
Only use NVCC with Kokkos if not using HIPCC
 2020-08-07 11:00:56 +01:00
Tom Lin
59274d6a91 Add NVIDIA as target for dpcpp 2020-08-05 08:54:40 +01:00
Tom Lin
603dc7d136 Add HIP compilers for Kokkos 2020-08-05 08:49:18 +01:00