Changes

Page history
Update Documentation authored by Camille Coti's avatar Camille Coti
Hide whitespace changes
Inline Side-by-side
Documentation.md
View page @ 148f910d
......@@ -2,7 +2,7 @@
```
module use $HOME/x86/modulefiles
module load openblas/git cuda gcc hwloc cmake automake autoconf
module load openblas/git cuda gcc hwloc cmake automake autoconf cmake
```
## StarPU
......@@ -61,3 +61,75 @@ Then... voila
```
~/x86/apex/install/bin/apex_exec --apex:cuda ./axpy
```
You should see an output with several parts:
- the welcome message: this might be slightly different depending on the branch you are using or something like that, this is Kevin's ascii art.
```
( )
( )\ ) ( /(
)\ (()/( ( )\())
((((_)( /(_)))\ ((_)\
)\ _ )\ (_)) ((_) __((_)
(_)_\(_)| _ \| __|\ \/ /
/ _ \ | _/| _| > <
/_/ \_\ |_| |___|/_/\_\
APEX Version: v2.5.1-86808d4-develop
Built on: 17:21:31 Sep 27 2022 (Release)
C++ Language Standard version : 201703
GCC Compiler version : 10.2.0
```
- Global counters: this is where you can see the scheduler's internal callbacks.
```
Counter : #samples | minimum | mean | maximum | stddev
------------------------------------------------------------------------------------------------
1 Minute Load average : 5 2.000 5.906 21.530 7.812
CPU Guest % : 4 0.000 0.000 0.000 0.000
CPU I/O Wait % : 4 0.000 0.000 0.000 0.000
CPU IRQ % : 4 0.000 0.000 0.000 0.000
CPU Idle % : 4 23.671 26.125 31.452 3.111
CPU Nice % : 4 0.000 0.000 0.000 0.000
CPU Steal % : 4 0.000 0.000 0.000 0.000
CPU System % : 4 2.205 6.233 11.003 3.494
CPU User % : 4 60.357 67.509 73.984 5.377
CPU soft IRQ % : 4 0.125 0.133 0.142 0.008
DRAM Energy : 4 9.000 10.000 11.000 0.707
Package-0 Energy : 4 123.000 150.500 169.000 19.216
Peak number of ready tasks (g_peak_ready) : 16 0.000 2.188 5.000 2.214
Peak number of tasks submitted (g_peak_s… : 16 0.000 0.625 1.000 0.484
Total submitted tasks (g_total_submitted) : 16 0.000 3.250 8.000 3.192
Worker 0 w_cumul_execution_time : 2 1.08e+06 1.08e+06 1.08e+06 44.941
Worker 0 w_total_executed : 2 1.000 1.500 2.000 0.500
Worker 1 w_cumul_execution_time : 3 2.04e+06 2.04e+06 2.04e+06 96.856
[...]
```
- GPU and CPU timers
```
------------------------------------------------------------------------------------------------
GPU Timers : #calls | mean | total | % total
------------------------------------------------------------------------------------------------
GPU: StarPU exec :1 wrapper_func:31 : 2 1.020 2.040 45.543
GPU: StarPU exec :1 init_cublas_func:47 : 1 2.040 2.040 45.542
GPU: StarPU exec :2 wrapper_func:31 : 2 0.735 1.470 32.811
GPU: StarPU exec :2 init_cublas_func:47 : 1 1.470 1.470 32.810
GPU: StarPU exec :0 wrapper_func:31 : 2 0.542 1.084 24.189
GPU: StarPU exec :0 init_cublas_func:47 : 1 1.084 1.084 24.188
GPU: StarPU exec :1 axpy_gpu:68 : 1 0.000 0.000 0.002
GPU: StarPU exec :2 axpy_gpu:68 : 1 0.000 0.000 0.001
GPU: StarPU exec :1 set_cublas_stream_func:63 : 1 0.000 0.000 0.000
GPU: StarPU exec :0 set_cublas_stream_func:63 : 1 0.000 0.000 0.000
GPU: StarPU exec :2 set_cublas_stream_func:63 : 1 0.000 0.000 0.000
------------------------------------------------------------------------------------------------
CPU Timers : #calls | #yields | mean | total | % total
-----------------------------------------------------------------------------------------------------------
APEX MAIN : 1 0 4.480 4.480 100.000
StarPU init : 2 0 2.943 5.886 --n/a--
StarPU driver :11 : 1 0 2.655 2.655 59.255
StarPU driver :6 : 1 0 2.652 2.652 59.192
StarPU driver :7 : 1 0 2.652 2.652 59.192
StarPU driver :8 : 1 0 2.652 2.652 59.186
[...]
```
\ No newline at end of file