Mentions légales du service

Skip to content
Snippets Groups Projects
Commit 2c6c4e52 authored by Berenger Bramas's avatar Berenger Bramas
Browse files

Update tutorial

parent 4cb2371d
Branches
Tags
No related merge requests found
Hide whitespace changes
Inline Side-by-side
Addons/BenchEfficiency/execAllHomogeneous.sh 0 → 100644
+ 26
− 0
View file @ 2c6c4e52
#!/bin/bash
echo "Perform the computation for :"
echo "$SCALFMM_NB particles"
echo "$SCALFMM_H tree height"
echo "Up to $SCALFMM_MAX_NB_CPU CPUs"
echo ""
echo "Using granularities:"
echo "$SCALFMM_BS_CPU_SEQ and $SCALFMM_BS_CPU_PAR"
for (( cpu=1 ; cpu<=$SCALFMM_MAX_NB_CPU ; cpu++)) ; do
echo ">> CPU = $cpu"
STARPU_NCPUS=$cpu
STARPU_NCUDA=0
./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs $SCALFMM_BS_CPU_SEQ
rec_name="trace-nb_$SCALFMM_NB-h_$SCALFMM_H-bs_$SCALFMM_CPU_SEQ-CPU_$cpu.rec"
mv trace.rec output/$rec_name
python $SCALFMM_STARPU_DIR/bin/starpu_trace_state_stats.py -t output/$rec_name
./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs $SCALFMM_BS_CPU_PAR
rec_name="trace-nb_$SCALFMM_NB-h_$SCALFMM_H-bs_$SCALFMM_CPU_PAR-CPU_$cpu.rec"
mv trace.rec output/$rec_name
python $SCALFMM_STARPU_DIR/bin/starpu_trace_state_stats.py -t output/$rec_name
done
Addons/BenchEfficiency/scalfmm.html
+ 69
− 6
View file @ 2c6c4e52
...@@ -115,10 +115,12 @@ fi</code></pre> ...@@ -115,10 +115,12 @@ fi</code></pre>
if [[ ! -d scalfmm-public ]] ; then if [[ ! -d scalfmm-public ]] ; then
git clone --depth=1 https://scm.gforge.inria.fr/anonscm/git/scalfmm-public/scalfmm-public.git git clone --depth=1 https://scm.gforge.inria.fr/anonscm/git/scalfmm-public/scalfmm-public.git
fi fi
cd scalfmm-public/Build/ cd scalfmm-public/
export SCALFMM_SOURCE_DIR=`pwd`
Build/
export SCALFMM_BUILD_DIR=`pwd`</code></pre></li> export SCALFMM_BUILD_DIR=`pwd`</code></pre></li>
</ul> </ul>
<p><em>Output variables:</em> <code>SCALFMM_BUILD_DIR</code></p> <p><em>Output variables:</em> <code>SCALFMM_BUILD_DIR</code> <code>SCALFMM_SOURCE_DIR</code></p>
<ul> <ul>
<li><p>Configure (No MKL):</p> <li><p>Configure (No MKL):</p>
<pre class="bash"><code>cmake .. -DSCALFMM_BUILD_DEBUG=OFF -DSCALFMM_USE_MPI=OFF -DSCALFMM_BUILD_TESTS=ON -DSCALFMM_BUILD_UTESTS=OFF -DSCALFMM_USE_BLAS=ON -DSCALFMM_USE_MKL_AS_BLAS=OFF -DSCALFMM_USE_LOG=ON -DSCALFMM_USE_STARPU=ON -DSCALFMM_USE_CUDA=ON -DSCALFMM_USE_OPENCL=OFF -DHWLOC_DIR=$SCALFMM_HWLOC_DIR -DSTARPU_DIR=$SCALFMM_STARPU_DIR -DSCALFMM_USE_FFT=ON -DFFT_DIR=$SCALFMM_FFT_DIR</code></pre></li> <pre class="bash"><code>cmake .. -DSCALFMM_BUILD_DEBUG=OFF -DSCALFMM_USE_MPI=OFF -DSCALFMM_BUILD_TESTS=ON -DSCALFMM_BUILD_UTESTS=OFF -DSCALFMM_USE_BLAS=ON -DSCALFMM_USE_MKL_AS_BLAS=OFF -DSCALFMM_USE_LOG=ON -DSCALFMM_USE_STARPU=ON -DSCALFMM_USE_CUDA=ON -DSCALFMM_USE_OPENCL=OFF -DHWLOC_DIR=$SCALFMM_HWLOC_DIR -DSTARPU_DIR=$SCALFMM_STARPU_DIR -DSCALFMM_USE_FFT=ON -DFFT_DIR=$SCALFMM_FFT_DIR</code></pre></li>
...@@ -180,6 +182,43 @@ make testBlockedUnifCudaBench</code></pre> ...@@ -180,6 +182,43 @@ make testBlockedUnifCudaBench</code></pre>
&quot;P2M&quot;,1,&quot;Task&quot;,2.543303 &quot;P2M&quot;,1,&quot;Task&quot;,2.543303
&quot;L2P&quot;,1,&quot;Task&quot;,5.649106 &quot;L2P&quot;,1,&quot;Task&quot;,5.649106
&quot;M2L-level-2&quot;,1,&quot;Task&quot;,2.167273</code></pre> &quot;M2L-level-2&quot;,1,&quot;Task&quot;,2.167273</code></pre>
<p>Most of the script are in the addon directories</p>
<pre><code>export SCALFMM_AB=$SCALFMM_SOURCE_DIR/Addons/BenchEfficiency/</code></pre>
<h2 id="homogeneous-efficiencies">Homogeneous Efficiencies</h2>
<p>Here we compute the efficiencies for a given test case on CPU only.</p>
<p>Go in the build dir and create output dir</p>
<pre><code>cd $SCALFMM_BUILD_DIR
mkdir homogeneous</code></pre>
<p>Set up the configuration variables:</p>
<pre class="bash"><code>SCALFMM_NB=10000000
SCALFMM_H=7
SCALFMM_MIN_BS=100
SCALFMM_MAX_BS=3000
SCALFMM_MAX_NB_CPU=24</code></pre>
<p>Find best granularity in sequential and in parallel:</p>
<pre class="bash"><code>STARPU_NCPUS=1
STARPU_NCUDA=0
SCALFMM_BS_CPU_SEQ=`$SCALFMM_AB/scalfmmFindBs.sh &quot;./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs&quot; $SCALFMM_MIN_BS $SCALFMM_MAX_BS | $SCALFMM_AB/scalfmm_extract_key &quot;@BEST BS&quot; `
if [[ `which gnuplot | wc -l` == &quot;1&quot; ]] ; then
gnuplot -e &quot;filename=&#39;seq-bs-search&#39;&quot; $SCALFMM_AB/scalfmmFindBs.gplot
fi
STARPU_NCPUS=$SCALFMM_MAX_NB_CPU
STARPU_NCUDA=0
=`$SCALFMM_AB/scalfmmFindBs.sh &quot;./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs&quot; $SCALFMM_MIN_BS $SCALFMM_MAX_BS | $SCALFMM_AB/scalfmm_extract_key &quot;@BEST BS&quot; `
if [[ `which gnuplot | wc -l` == &quot;1&quot; ]] ; then
gnuplot -e &quot;filename=&#39;par-bs-search&#39;&quot; $SCALFMM_AB/scalfmmFindBs.gplot
fi</code></pre>
<p>Then we compute the efficiency using both granulirities and keep the .rec files.</p>
<pre class="bash"><code>source $SCALFMM_AB/execAllHomogeneous.sh</code></pre>
<p>We should end with all the rec files and their corresponding time files</p>
<pre class="bash"><code></code></pre>
<p>We compute the efficiencies</p>
<pre class="bash"><code></code></pre>
<p>We end with efficiency for the application and for the operators.</p>
<pre class="bash"><code></code></pre>
<p>We can plot each of them</p>
<pre class="bash"><code></code></pre>
<h2 id="generating-execution-results">Generating Execution Results</h2> <h2 id="generating-execution-results">Generating Execution Results</h2>
<p>For test case <code>-nb 10000000</code> (10 million) and <code>-h 6</code> (height of the tree equal to 6), we first want to know the best granularity <code>-bs</code>.</p> <p>For test case <code>-nb 10000000</code> (10 million) and <code>-h 6</code> (height of the tree equal to 6), we first want to know the best granularity <code>-bs</code>.</p>
<p>This parameter will certainly not be the same for sequential/parallel/heterogenous configurations.</p> <p>This parameter will certainly not be the same for sequential/parallel/heterogenous configurations.</p>
...@@ -189,11 +228,26 @@ SCALFMM_MIN_BS=100 ...@@ -189,11 +228,26 @@ SCALFMM_MIN_BS=100
SCALFMM_MAX_BS=3000 SCALFMM_MAX_BS=3000
SCALFMM_MAX_NB_CPU=24 SCALFMM_MAX_NB_CPU=24
SCALFMM_MAX_NB_GPU=4</code></pre> SCALFMM_MAX_NB_GPU=4</code></pre>
<pre class="bash"><code>SCALFMM_BS_CPU_SEQ=`scalfmm_bench_get_best_bs -nb $SCALFMM_NB -h $SCALFMM_H -start $SCALFMM_MIN_BS -end $SCALFMM_MAX_BS | scalfmm_extract_key &quot;@BEST BS&quot; ` <pre class="bash"><code>STARPU_NCPUS=1
STARPU_NCUDA=0
SCALFMM_BS_CPU_SEQ=`$SCALFMM_AB/scalfmmFindBs.sh -nb $SCALFMM_NB -h $SCALFMM_H $SCALFMM_MIN_BS $SCALFMM_MAX_BS | $SCALFMM_AB/scalfmm_extract_key &quot;@BEST BS&quot; `
if [[ `which gnuplot | wc -l` == &quot;1&quot; ]] ; then
gnuplot -e &quot;filename=&#39;seq-bs-search&#39;&quot; $SCALFMM_AB/scalfmmFindBs.gplot
fi
SCALFMM_BS_CPU_PAR=`scalfmm_bench_get_best_bs -nb $SCALFMM_NB -h $SCALFMM_H -start $SCALFMM_MIN_BS -end $SCALFMM_MAX_BS | scalfmm_extract_key &quot;@BEST BS&quot; ` STARPU_NCPUS=$SCALFMM_MAX_NB_CPU
STARPU_NCUDA=0
SCALFMM_BS_CPU_PAR=`$SCALFMM_AB/scalfmmFindBs.sh -nb $SCALFMM_NB -h $SCALFMM_H $SCALFMM_MIN_BS $SCALFMM_MAX_BS | $SCALFMM_AB/scalfmm_extract_key &quot;@BEST BS&quot; `
if [[ `which gnuplot | wc -l` == &quot;1&quot; ]] ; then
gnuplot -e &quot;filename=&#39;par-bs-search&#39;&quot; $SCALFMM_AB/scalfmmFindBs.gplot
fi
SCALFMM_BS_CPU_GPU=`scalfmm_bench_get_best_bs -nb $SCALFMM_NB -h $SCALFMM_H -start $SCALFMM_MIN_BS -end $SCALFMM_MAX_BS | scalfmm_extract_key &quot;@BEST BS&quot; `</code></pre> STARPU_NCPUS=$SCALFMM_MAX_NB_CPU
STARPU_NCUDA=$SCALFMM_MAX_NB_GPU
SCALFMM_BS_CPU_GPU=`$SCALFMM_AB/scalfmmFindBs.sh -nb $SCALFMM_NB -h $SCALFMM_H $SCALFMM_MIN_BS $SCALFMM_MAX_BS | $SCALFMM_AB/scalfmm_extract_key &quot;@BEST BS&quot; `
if [[ `which gnuplot | wc -l` == &quot;1&quot; ]] ; then
gnuplot -e &quot;filename=&#39;cpugpu-bs-search&#39;&quot; $SCALFMM_AB/scalfmmFindBs.gplot
fi</code></pre>
<p>Then, we can execute three best configurations, and keep .rec for each of them:</p> <p>Then, we can execute three best configurations, and keep .rec for each of them:</p>
<pre class="bash"><code>STARPU_NCPUS=1 <pre class="bash"><code>STARPU_NCPUS=1
STARPU_NCUDA=0 STARPU_NCUDA=0
...@@ -224,13 +278,22 @@ STARPU_NCUDA=0 ...@@ -224,13 +278,22 @@ STARPU_NCUDA=0
./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs $SCALFMM_BS_CPU_PAR ./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs $SCALFMM_BS_CPU_PAR
SCALFMM_SEQ_CPU_BS_REC=&quot;trace-nb_$SCALFMM_NB-h_$SCALFMM_H-bs_$SCALFMM_CPU_SEQ-CPU_$STARPU_NCPUS-GPU_$STARPU_NCUDA.rec&quot; SCALFMM_SEQ_CPU_BS_REC=&quot;trace-nb_$SCALFMM_NB-h_$SCALFMM_H-bs_$SCALFMM_CPU_SEQ-CPU_$STARPU_NCPUS-GPU_$STARPU_NCUDA.rec&quot;
mv trace.rec SCALFMM_SEQ_CPU_BS_REC mv trace.rec $SCALFMM_SEQ_CPU_BS_REC
./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs $SCALFMM_BS_CPU_GPU ./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs $SCALFMM_BS_CPU_GPU
SCALFMM_SEQ_GPU_BS_REC=&quot;trace-nb_$SCALFMM_NB-h_$SCALFMM_H-bs_$SCALFMM_CPU_SEQ-CPU_$STARPU_NCPUS-GPU_$STARPU_NCUDA.rec&quot; SCALFMM_SEQ_GPU_BS_REC=&quot;trace-nb_$SCALFMM_NB-h_$SCALFMM_H-bs_$SCALFMM_CPU_SEQ-CPU_$STARPU_NCPUS-GPU_$STARPU_NCUDA.rec&quot;
mv trace.rec $SCALFMM_SEQ_GPU_BS_REC</code></pre> mv trace.rec $SCALFMM_SEQ_GPU_BS_REC</code></pre>
<p>From these files, we are able to get the different efficencies.</p> <p>From these files, we are able to get the different efficencies.</p>
<h2 id="post-processing-and-plot">Post-processing and Plot</h2> <h2 id="post-processing-and-plot">Post-processing and Plot</h2>
<p>From the file:</p>
<ul>
<li><code>$SCALFMM_SEQ_REC</code> : the resulting file from the sequential execution with best sequential granularity</li>
<li><code>$SCALFMM_PAR_REC</code> : the resulting file from a parallel execution (no GPU) with best parallel granularity</li>
<li><code>$SCALFMM_PAR_CPU_GPU_REC</code> : the resulting file from a parallel execution (hybrid) with best parallel-hybrid granularity</li>
<li><code>$SCALFMM_PAR_GPU_REC</code> : the resulting file with all possible tasks on GPU with best parallel-hybrid granularity</li>
<li><code>$SCALFMM_SEQ_CPU_BS_REC</code> : the resulting file from sequential execution with best parallel granularity</li>
<li><code>$SCALFMM_SEQ_GPU_BS_REC</code> : the resulting file from sequential execution with best parallel-hybrid granularity</li>
</ul>
<p>Getting all the efficency Solving the linear programming problem</p> <p>Getting all the efficency Solving the linear programming problem</p>
<p>Plotting the results</p> <p>Plotting the results</p>
<h2 id="automatization">Automatization</h2> <h2 id="automatization">Automatization</h2>
......
Addons/BenchEfficiency/scalfmm.md
+ 101
− 6
View file @ 2c6c4e52
...@@ -165,11 +165,13 @@ cd $SCALFMM_TEST_DIR ...@@ -165,11 +165,13 @@ cd $SCALFMM_TEST_DIR
if [[ ! -d scalfmm-public ]] ; then if [[ ! -d scalfmm-public ]] ; then
git clone --depth=1 https://scm.gforge.inria.fr/anonscm/git/scalfmm-public/scalfmm-public.git git clone --depth=1 https://scm.gforge.inria.fr/anonscm/git/scalfmm-public/scalfmm-public.git
fi fi
cd scalfmm-public/Build/ cd scalfmm-public/
export SCALFMM_SOURCE_DIR=`pwd`
Build/
export SCALFMM_BUILD_DIR=`pwd` export SCALFMM_BUILD_DIR=`pwd`
``` ```
*Output variables:* `SCALFMM_BUILD_DIR` *Output variables:* `SCALFMM_BUILD_DIR` `SCALFMM_SOURCE_DIR`
+ Configure (No MKL): + Configure (No MKL):
```bash ```bash
...@@ -261,6 +263,75 @@ Should give something like: ...@@ -261,6 +263,75 @@ Should give something like:
"M2L-level-2",1,"Task",2.167273 "M2L-level-2",1,"Task",2.167273
``` ```
Most of the script are in the addon directories
```
export SCALFMM_AB=$SCALFMM_SOURCE_DIR/Addons/BenchEfficiency/
```
## Homogeneous Efficiencies
Here we compute the efficiencies for a given test case on CPU only.
Go in the build dir and create output dir
```
cd $SCALFMM_BUILD_DIR
mkdir homogeneous
```
Set up the configuration variables:
```bash
SCALFMM_NB=10000000
SCALFMM_H=7
SCALFMM_MIN_BS=100
SCALFMM_MAX_BS=3000
SCALFMM_MAX_NB_CPU=24
```
Find best granularity in sequential and in parallel:
```bash
STARPU_NCPUS=1
STARPU_NCUDA=0
SCALFMM_BS_CPU_SEQ=`$SCALFMM_AB/scalfmmFindBs.sh "./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs" $SCALFMM_MIN_BS $SCALFMM_MAX_BS | $SCALFMM_AB/scalfmm_extract_key "@BEST BS" `
if [[ `which gnuplot | wc -l` == "1" ]] ; then
gnuplot -e "filename='seq-bs-search'" $SCALFMM_AB/scalfmmFindBs.gplot
fi
STARPU_NCPUS=$SCALFMM_MAX_NB_CPU
STARPU_NCUDA=0
=`$SCALFMM_AB/scalfmmFindBs.sh "./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs" $SCALFMM_MIN_BS $SCALFMM_MAX_BS | $SCALFMM_AB/scalfmm_extract_key "@BEST BS" `
if [[ `which gnuplot | wc -l` == "1" ]] ; then
gnuplot -e "filename='par-bs-search'" $SCALFMM_AB/scalfmmFindBs.gplot
fi
```
Then we compute the efficiency using both granulirities and keep the .rec files.
```bash
source $SCALFMM_AB/execAllHomogeneous.sh
```
We should end with all the rec files and their corresponding time files
```bash
```
We compute the efficiencies
```bash
```
We end with efficiency for the application and for the operators.
```bash
```
We can plot each of them
```bash
```
## Generating Execution Results ## Generating Execution Results
For test case `-nb 10000000` (10 million) and `-h 6` (height of the tree equal to 6), For test case `-nb 10000000` (10 million) and `-h 6` (height of the tree equal to 6),
...@@ -278,11 +349,26 @@ SCALFMM_MAX_NB_GPU=4 ...@@ -278,11 +349,26 @@ SCALFMM_MAX_NB_GPU=4
``` ```
```bash ```bash
SCALFMM_BS_CPU_SEQ=`scalfmm_bench_get_best_bs -nb $SCALFMM_NB -h $SCALFMM_H -start $SCALFMM_MIN_BS -end $SCALFMM_MAX_BS | scalfmm_extract_key "@BEST BS" ` STARPU_NCPUS=1
STARPU_NCUDA=0
SCALFMM_BS_CPU_SEQ=`$SCALFMM_AB/scalfmmFindBs.sh -nb $SCALFMM_NB -h $SCALFMM_H $SCALFMM_MIN_BS $SCALFMM_MAX_BS | $SCALFMM_AB/scalfmm_extract_key "@BEST BS" `
if [[ `which gnuplot | wc -l` == "1" ]] ; then
gnuplot -e "filename='seq-bs-search'" $SCALFMM_AB/scalfmmFindBs.gplot
fi
SCALFMM_BS_CPU_PAR=`scalfmm_bench_get_best_bs -nb $SCALFMM_NB -h $SCALFMM_H -start $SCALFMM_MIN_BS -end $SCALFMM_MAX_BS | scalfmm_extract_key "@BEST BS" ` STARPU_NCPUS=$SCALFMM_MAX_NB_CPU
STARPU_NCUDA=0
SCALFMM_BS_CPU_PAR=`$SCALFMM_AB/scalfmmFindBs.sh -nb $SCALFMM_NB -h $SCALFMM_H $SCALFMM_MIN_BS $SCALFMM_MAX_BS | $SCALFMM_AB/scalfmm_extract_key "@BEST BS" `
if [[ `which gnuplot | wc -l` == "1" ]] ; then
gnuplot -e "filename='par-bs-search'" $SCALFMM_AB/scalfmmFindBs.gplot
fi
SCALFMM_BS_CPU_GPU=`scalfmm_bench_get_best_bs -nb $SCALFMM_NB -h $SCALFMM_H -start $SCALFMM_MIN_BS -end $SCALFMM_MAX_BS | scalfmm_extract_key "@BEST BS" ` STARPU_NCPUS=$SCALFMM_MAX_NB_CPU
STARPU_NCUDA=$SCALFMM_MAX_NB_GPU
SCALFMM_BS_CPU_GPU=`$SCALFMM_AB/scalfmmFindBs.sh -nb $SCALFMM_NB -h $SCALFMM_H $SCALFMM_MIN_BS $SCALFMM_MAX_BS | $SCALFMM_AB/scalfmm_extract_key "@BEST BS" `
if [[ `which gnuplot | wc -l` == "1" ]] ; then
gnuplot -e "filename='cpugpu-bs-search'" $SCALFMM_AB/scalfmmFindBs.gplot
fi
``` ```
Then, we can execute three best configurations, and keep .rec for each of them: Then, we can execute three best configurations, and keep .rec for each of them:
...@@ -322,7 +408,7 @@ STARPU_NCUDA=0 ...@@ -322,7 +408,7 @@ STARPU_NCUDA=0
./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs $SCALFMM_BS_CPU_PAR ./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs $SCALFMM_BS_CPU_PAR
SCALFMM_SEQ_CPU_BS_REC="trace-nb_$SCALFMM_NB-h_$SCALFMM_H-bs_$SCALFMM_CPU_SEQ-CPU_$STARPU_NCPUS-GPU_$STARPU_NCUDA.rec" SCALFMM_SEQ_CPU_BS_REC="trace-nb_$SCALFMM_NB-h_$SCALFMM_H-bs_$SCALFMM_CPU_SEQ-CPU_$STARPU_NCPUS-GPU_$STARPU_NCUDA.rec"
mv trace.rec SCALFMM_SEQ_CPU_BS_REC mv trace.rec $SCALFMM_SEQ_CPU_BS_REC
./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs $SCALFMM_BS_CPU_GPU ./Tests/Release/testBlockedUnifCudaBench -nb $SCALFMM_NB -h $SCALFMM_H -bs $SCALFMM_BS_CPU_GPU
SCALFMM_SEQ_GPU_BS_REC="trace-nb_$SCALFMM_NB-h_$SCALFMM_H-bs_$SCALFMM_CPU_SEQ-CPU_$STARPU_NCPUS-GPU_$STARPU_NCUDA.rec" SCALFMM_SEQ_GPU_BS_REC="trace-nb_$SCALFMM_NB-h_$SCALFMM_H-bs_$SCALFMM_CPU_SEQ-CPU_$STARPU_NCPUS-GPU_$STARPU_NCUDA.rec"
...@@ -333,6 +419,15 @@ From these files, we are able to get the different efficencies. ...@@ -333,6 +419,15 @@ From these files, we are able to get the different efficencies.
## Post-processing and Plot ## Post-processing and Plot
From the file:
+ `$SCALFMM_SEQ_REC` : the resulting file from the sequential execution with best sequential granularity
+ `$SCALFMM_PAR_REC` : the resulting file from a parallel execution (no GPU) with best parallel granularity
+ `$SCALFMM_PAR_CPU_GPU_REC` : the resulting file from a parallel execution (hybrid) with best parallel-hybrid granularity
+ `$SCALFMM_PAR_GPU_REC` : the resulting file with all possible tasks on GPU with best parallel-hybrid granularity
+ `$SCALFMM_SEQ_CPU_BS_REC` : the resulting file from sequential execution with best parallel granularity
+ `$SCALFMM_SEQ_GPU_BS_REC` : the resulting file from sequential execution with best parallel-hybrid granularity
Getting all the efficency Getting all the efficency
Solving the linear programming problem Solving the linear programming problem
......
Addons/BenchEfficiency/scalfmmExtractKey.sh 0 → 100644
+ 10
− 0
View file @ 2c6c4e52
#!/bin/bash
if [[ $# -ne 1 ]] ; then
echo "You must pass a key as parameter"
return
fi
input=$(cat)
res=`echo "$input" | grep "$3" | cut -d'=' -f2 | cut -d's' -f1`
echo $res
Addons/BenchEfficiency/scalfmmFindBs.gplot
+ 2
− 2
View file @ 2c6c4e52
#!/usr/bin/gnuplot #!/usr/bin/gnuplot
reset reset
set terminal png set terminal png
set output "file.png" set output filename . '.png'
set xlabel "bs" set xlabel "bs"
...@@ -12,4 +12,4 @@ set grid ...@@ -12,4 +12,4 @@ set grid
set style data linespoints set style data linespoints
plot "<sort -g -k1 file.data" using 1:2 title "Example" plot "<sort -g -k1 benchBs.data" using 1:2 title "Example"
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment