Tom Deakin
|
d463c036b1
|
Move version string to top of common header
|
2015-07-29 16:02:26 +01:00 |
|
Tom Deakin
|
0bd4bbe82d
|
Make init arrays 1.0f, etc if float
|
2015-07-29 14:47:37 +01:00 |
|
Tom Deakin
|
e3de82d63c
|
Make 3.0 floating point in kernel if float
|
2015-07-29 14:31:26 +01:00 |
|
Tom Deakin
|
dcf734387e
|
Free host arrays at end of program
|
2015-07-28 13:59:31 +01:00 |
|
Tom Deakin
|
a9a087622d
|
Use large try/catch in main function for clarity
|
2015-07-28 13:37:15 +01:00 |
|
Tom Deakin
|
3c248195ea
|
Add try/catch around kernel executions
|
2015-07-28 13:18:02 +01:00 |
|
Tom Deakin
|
807af7d679
|
Remove cuda check in main line Makefile - only needs to be in CUDA part
|
2015-07-28 12:09:23 +01:00 |
|
Tom Deakin
|
dea95fe68e
|
Move common exceptions to common.h
|
2015-07-28 12:06:21 +01:00 |
|
Tom Deakin
|
201bc4638f
|
Bug fix: library linker error in Makefile on OS X
|
2015-07-28 12:04:25 +01:00 |
|
Tom Deakin
|
2b00245e63
|
Remove large try/catch block in OpenCL - wrap around each OCL call
|
2015-07-28 12:03:44 +01:00 |
|
michael boulton
|
e605b056a6
|
Fixed overflow in total array size
|
2015-07-28 11:42:40 +01:00 |
|
Michael Boulton
|
0a89282d57
|
new files
|
2015-07-28 11:37:50 +01:00 |
|
Michael Boulton
|
b43eb9cf16
|
Merge changes
|
2015-07-28 11:37:20 +01:00 |
|
Tom Deakin
|
bb0dcce28b
|
Check for nvcc before building CUDA version
|
2015-07-27 17:40:24 +01:00 |
|
Tom Deakin
|
ad93754c3b
|
Print out memory sizes
|
2015-07-27 17:32:47 +01:00 |
|
Tom Deakin
|
7ba0b8fc18
|
Display precision in the header output
|
2015-07-27 17:19:19 +01:00 |
|
Tom Deakin
|
80955ff2dc
|
Add CUDA error checking
|
2015-07-27 17:07:07 +01:00 |
|
Tom Deakin
|
ec9d73b8d5
|
CUDA bug: ARRAY_SIZE on CLI not being resized
The ARRAY_SIZE was being resized before the CLI args were read
|
2015-07-27 16:49:36 +01:00 |
|
Tom Deakin
|
a74e7cee4c
|
Bug fix: sizes array wasn't updating with ARRAY_SIZE argument
sizes array was fixed at compile time not run time.
Moved the declaration to the correct scope.
|
2015-07-27 13:44:07 +01:00 |
|
Tom Deakin
|
c3ad5edcb3
|
Port float code to CUDA version
|
2015-07-23 12:49:25 +01:00 |
|
Tom Deakin
|
e7fc832771
|
Add --float option to cli
|
2015-07-23 12:15:58 +01:00 |
|
Tom Deakin
|
ae7a352809
|
Add a float option to the kernel
|
2015-07-23 12:12:02 +01:00 |
|
Tom Deakin
|
99d2fe8526
|
Add S9150 result
|
2015-07-22 12:44:07 +01:00 |
|
Tom Deakin
|
cd786dbe25
|
CUDA fix geting device name with the device index
|
2015-07-20 17:09:26 +01:00 |
|
Tom Deakin
|
91c804edcd
|
Add NVIDIA GTX 980 Ti and 780 Ti results
|
2015-07-16 18:47:12 +01:00 |
|
Tom Deakin
|
7e0821278a
|
Add NVIDIA GTX 980 and 580 results
|
2015-07-16 18:43:41 +01:00 |
|
Tom Deakin
|
5fc14bbba1
|
Change binary name in cuda help
|
2015-07-16 18:32:23 +01:00 |
|
Tom Deakin
|
30305d83e0
|
CUDA: select number to times from CLI
|
2015-07-16 18:31:54 +01:00 |
|
Tom Deakin
|
0edd0646a6
|
OCL: select number to times from CLI
|
2015-07-16 18:29:42 +01:00 |
|
Tom Deakin
|
c52c7f290c
|
Update README
|
2015-07-16 18:23:28 +01:00 |
|
Tom Deakin
|
0aaceef49d
|
Update Makefile with all and cuda stuff
|
2015-07-16 18:16:27 +01:00 |
|
Tom Deakin
|
39a63b7d32
|
CUDA fix bug with listing device names
|
2015-07-16 18:14:11 +01:00 |
|
Tom Deakin
|
153b6f2105
|
Add results directory with some sample results
|
2015-07-16 18:12:26 +01:00 |
|
Tom Deakin
|
f2536f8764
|
Enqueue CUDA kernels
|
2015-07-16 18:06:35 +01:00 |
|
Tom Deakin
|
2c2dbf2c3f
|
You have to specify block size in CUDA, so make sure ARRAY_SIZE divides 1024
|
2015-07-16 18:06:07 +01:00 |
|
Tom Deakin
|
9da2cd6d09
|
CUDA: make device buffers arrays not vectors
|
2015-07-16 17:53:09 +01:00 |
|
Tom Deakin
|
e76cc9fb62
|
Uncomment kernels in CUDA - have to remove restrict keyword
|
2015-07-16 17:46:50 +01:00 |
|
Tom Deakin
|
fa865dca4f
|
Enumerate and select CUDA devices
|
2015-07-16 17:46:03 +01:00 |
|
Tom Deakin
|
99a63245be
|
Add .gitignore file
|
2015-07-16 17:35:39 +01:00 |
|
Tom Deakin
|
cf6809ba37
|
Change output to say CUDA if CUDA version
|
2015-07-16 17:34:59 +01:00 |
|
Tom Deakin
|
3ff57ba9d6
|
Add parseInt to cuda version for device index
|
2015-07-16 17:33:58 +01:00 |
|
Tom Deakin
|
8c7224ab42
|
Remove cl_uint reference in CUDA version
|
2015-07-16 17:31:59 +01:00 |
|
Tom Deakin
|
06c24b2867
|
Remove OpenCL header and use CUDA header in CUDA version
|
2015-07-16 17:30:54 +01:00 |
|
Tom Deakin
|
435cb3b971
|
Remove catching cl::Error in CUDA version
|
2015-07-16 17:30:13 +01:00 |
|
Tom Deakin
|
cfe2f862ae
|
Add a half broken CUDA port
|
2015-07-16 17:28:28 +01:00 |
|
Tom Deakin
|
c31644d001
|
Revert "Revert "Remove option for float - only use doubles for now""
This reverts commit 903a66b2e9.
|
2015-07-16 17:01:48 +01:00 |
|
Tom Deakin
|
903a66b2e9
|
Revert "Remove option for float - only use doubles for now"
This reverts commit 4225ba0a2b.
|
2015-07-16 16:59:43 +01:00 |
|
Tom Deakin
|
4225ba0a2b
|
Remove option for float - only use doubles for now
|
2015-07-16 16:58:12 +01:00 |
|
Tom Deakin
|
e5bd4fcb19
|
Be consistent with indenting
|
2015-07-16 16:36:30 +01:00 |
|
Tom Deakin
|
97c179feb3
|
Add parseArguments etc to get the device from the command line
|
2015-07-16 16:32:49 +01:00 |
|