profiling power method

Author: lorenzotomada

scripts/plot_scalability.py

@@ -0,0 +1,51 @@
+import matplotlib.pyplot as plt
+import numpy as np
+import os
+
+
+csv_file = "tmp_data/timings.csv"
+
+if not os.path.exists(csv_file):
+    print(
+        f"Error: {csv_file} not found!\nMake sure you have executed the script 'submit.sbatch' in the 'shell' folder."
+    )
+    exit(1)
+
+data = np.genfromtxt(csv_file, delimiter=",", dtype=str, skip_header=1)
+
+
+dims = data[:, 0].astype(int)
+func_names = data[:, 1]
+times = data[:, 2].astype(float)
+
+
+unique_funcs = np.unique(func_names)
+
+plt.figure(figsize=(8, 6))
+
+
+for func in unique_funcs:
+    mask = func_names == func
+    plt.plot(dims[mask], times[mask], marker="o", linestyle="-", label=func)
+
+
+plt.xlabel(r"$\log(d)$")
+plt.ylabel("$t$")
+plt.xscale("log")
+plt.yscale("log")
+plt.title("Scalability")
+
+plt.legend()
+plt.grid(True)
+plt.tight_layout()
+
+
+output_dir = "logs"
+os.makedirs(output_dir, exist_ok=True)
+output_path = os.path.join(output_dir, "scalability.png")
+plt.savefig(output_path)
+print(f"Plot saved to {output_path}")
+
+plt.show()
+plt.close()
+plt.clf()

scripts/run.py

@@ -1,8 +1,12 @@
+import cupyx.scipy.sparse as cpsp
+import cupy as cp
 from pyclassify import (
     eigenvalues_np,
     eigenvalues_sp,
+    eigenvalues_cp,
     power_method,
     power_method_numba,
+    power_method_cp,
 )
 from pyclassify.utils import make_symmetric, read_config
 import numpy as np
@@ -27,20 +31,29 @@
 kwargs = read_config(filename)
 dim = kwargs["dim"]
 density = kwargs["density"]
-# also tol and max iter
+tol = kwargs["tol"]
+max_iter = kwargs["max_iter"]
 
 
 matrix = sp.random(dim, dim, density=density, format="csr")
 matrix = make_symmetric(matrix)
+cp_symm_matrix = cpsp.csr_matrix(matrix)
+
 
 eigs_np = eigenvalues_np(matrix.toarray(), symmetric=True)
 eigs_sp = eigenvalues_sp(matrix, symmetric=True)
+eigs_cp = eigenvalues_cp(cp_symm_matrix)
+
 
 index_np = np.argmax(np.abs(eigs_np))
 index_sp = np.argmax(np.abs(eigs_sp))
+index_cp = np.argmax(np.abs(eigs_cp))
+
 
 biggest_eigenvalue_np = eigs_np[index_np]
 biggest_eigenvalue_sp = eigs_sp[index_sp]
+biggest_eigenvalue_cp = eigs_cp[index_cp]
 
 biggest_eigenvalue_pm = power_method(matrix)
 biggest_eigenvalue_pm_numba = power_method_numba(matrix.toarray())
+biggest_eigenvalue_cp = power_method_cp(cp_symm_matrix)

shell/submit.sbatch

@@ -6,8 +6,8 @@
 #SBATCH --gpus=1
 #SBATCH --ntasks-per-node=1
 #SBATCH --gpus-per-task=1
-#SBATCH --mem=2000
-#SBATCH --time=01:00:00
+#SBATCH --mem=10000
+#SBATCH --time=12:00:00
 #SBATCH --mail-user=ltomada@sissa.it
 #SBATCH --output=%x.o%j.%N
 #SBATCH --error=%x.e%j.%N
@@ -29,6 +29,40 @@ export SLURM_NTASKS_PER_NODE=1  # due to Ulysses's bug
 module load cuda/12.1
 conda init
 source ~/.bashrc
-conda activate /scratch/ltomada/miniconda3/envs/devtools_scicomp
-python -m pytest -v
-#python scripts/run.py fit --config=experiments/config.yaml
+conda activate /scratch/ltomada/miniconda3/envs/dtsc
+
+mkdir -p tmp_data experiments logs
+touch tmp_data/timings.csv
+
+export LINE_PROFILE=0
+
+density=0.1
+tol=0.0001
+max_iter=1000
+
+
+for dim in 100 500 1000; do
+    echo "d: $dim" # for readability
+    echo "dim: $dim" > experiments/config.yaml    
+    echo "density: $density" >> experiments/config.yaml
+    echo "tol: $tol" >> experiments/config.yaml
+    echo "max_iter: $max_iter" >> experiments/config.yaml
+
+    python -m kernprof -l -o tmp_data/profile.dat scripts/run.py --config=experiments/config
+
+    if [ ! -f tmp_data/timings.csv ]; then
+        echo "dim,function_name,time" > tmp_data/timings.csv
+    fi
+
+    python -m line_profiler -rmt tmp_data/profile.dat | awk -v dim="$dim" '
+        /Function: / {func_name=$2}
+        /Total time:/ {if (func_name != "") print dim, func_name, $3}' OFS=',' >> tmp_data/timings.csv
+
+    echo "dim: $dim - times saved in tmp_data/timings.csv" 
+    echo '--------------------------------------------------'
+done
+
+
+python scripts/plot_scalability.py
+echo "Experiment completed!"
+#rm -rf tmp_data

shell/test.sh

@@ -0,0 +1,40 @@
+#!/bin/bash
+
+#SBATCH --partition=gpu2
+#SBATCH --job-name=dtsc
+#SBATCH --nodes=1
+#SBATCH --gpus=1
+#SBATCH --ntasks-per-node=1
+#SBATCH --gpus-per-task=1
+#SBATCH --mem=10000
+#SBATCH --time=02:00:00
+#SBATCH --mail-user=ltomada@sissa.it
+#SBATCH --output=%x.o%j.%N
+#SBATCH --error=%x.e%j.%N
+
+# Print job details
+NOW=`date +%H:%M-%a-%d/%b/%Y`
+echo '------------------------------------------------------'
+echo 'This job is allocated on '$SLURM_JOB_CPUS_PER_NODE' cpu(s)'
+echo 'Job is running on node(s): '
+echo  $SLURM_JOB_NODELIST
+echo '------------------------------------------------------'
+#
+# ==== End of Info part (say things) ===== #
+#
+
+cd $SLURM_SUBMIT_DIR           
+export SLURM_NTASKS_PER_NODE=1  # due to Ulysses's bug
+
+module load cuda/12.1
+conda init
+source ~/.bashrc
+conda activate /scratch/ltomada/miniconda3/envs/dtsc
+
+mkdir -p tmp_data experiments logs
+
+python -m pytest -v > logs/test.log
+
+touch tmp_data/timings.csv
+
+export LINE_PROFILE=0

src/pyclassify/utils.py

@@ -33,7 +33,6 @@ def check_square_matrix(A):
         raise ValueError("Matrix must be square!")
 
 
-@profile
 def make_symmetric(A):
     """
     Ensures the input matrix is symmetric by averaging it with its transpose.