 Samuel Thibault
|
5c7a0254df
follow-up on r15935: also disable non-blocking streams with old versions of cuda for matrix_as_vector
|
9 年 前 |
 Nathalie Furmento
|
bb61f3808d
update cnrs copyright
|
10 年 前 |
|
Samuel Thibault
|
0d90ae3f88
drop including sys/time.h, now useless
|
10 年 前 |
|
Samuel Thibault
|
169bf34d54
port r13398 from 1.1: eradicate use of timeval
|
10 年 前 |
|
Nathalie Furmento
|
c803479334
tests/microbenchs/matrix_as_vector.c: fix typo, define matrix_codelet and not vector_codelet
|
10 年 前 |
|
Samuel Thibault
|
7d2bf0c630
Enable asynchronous CUDA kernels wherever possible
|
11 年 前 |
|
Nathalie Furmento
|
5fdde967b0
tests: partly revert #11510 and turn all cpu implementations codelets not static so that MIC code is able to successfully call dlsym on the function
|
11 年 前 |
|
Nathalie Furmento
|
0c6c658cb1
tests: turn function static when it makes sense to do so
|
11 年 前 |
|
Samuel Thibault
|
8b83bc72aa
port r11470 from 1.1: pin memory for matrix_as_vector
|
11 年 前 |
|
Nathalie Furmento
|
98d9eb5c64
Functions starpu_insert_task and starpu_mpi_insert_task are renamed
|
11 年 前 |
 Corentin Salingue
|
bc3fd88dfd
replace 0 with macro part 4
|
12 年 前 |
|
Samuel Thibault
|
b227dbcfdc
Use STARPU_ATTRIBUTE_* macros
|
12 年 前 |
|
Nathalie Furmento
|
850c49fabc
Rename function starpu_helper_cublas_init to starpu_cublas_init
|
12 年 前 |
|
Nathalie Furmento
|
53e5ad2db2
tests/microbenchs/matrix_as_vector.c: free matrix
|
12 年 前 |
|
Nathalie Furmento
|
7445ed6406
tests: minor fixes following icc warnings
|
13 年 前 |
|
Nathalie Furmento
|
6b5e787563
tests/microbenchs/matrix_as_vector.c: decrease number of loops when running on valgrind
|
13 年 前 |
|
Nathalie Furmento
|
b1e32c1cc5
tests/microbenchs/matrix_as_vector.c: calling twice starpu_helper_cublas_init fails
|
13 年 前 |
|
Nathalie Furmento
|
a4f3da6a64
tests: only include starpu.h as it includes all the other starpu_ .h files
|
13 年 前 |
|
Samuel Thibault
|
29b7f32f77
Provide good examples by always using streams for cublas
|
13 年 前 |
|
Samuel Thibault
|
57e59bf2d9
provide good examples by always using cudaMemcpyAsync, not cudaMemcpy
|
13 年 前 |
|
Nathalie Furmento
|
ff5a7ac636
tests/microbenchs/matrix_as_vector.c: no need to test size lower than 1K
|
13 年 前 |
|
Nathalie Furmento
|
2aa6c64b28
tests/microbenchs/matrix_as_vector.c: mark test as skipped on ENODEV
|
13 年 前 |
|
Nathalie Furmento
|
4a858bf87c
tests/microbenchs/matrix_as_vector.c: always declare cuda kernels, and give them a empty body when cuda is not enabled
|
13 年 前 |
|
Nathalie Furmento
|
ffd3f77a04
tests/microbenchs/matrix_as_vector.c: only enable code when CUDA is enabled
|
13 年 前 |
|
Nathalie Furmento
|
bdd3680ca1
tests/microbenchs: add benchmark matrix_as_vector.c to compare the access of a matrix object as a vector and a matrix
|
13 年 前 |
