1 files changed, 40 insertions, 0 deletions
diff --git a/profile_orders.py b/profile_orders.py
new file mode 100644
index 0000000..658b054
--- /dev/null
+++ b/profile_orders.py
@@ -0,0 +1,40 @@
+import numpy as np
+from numpy import random as npr
+from pyecsca.sca.stacked_traces import GPUTraceManager
+from pyecsca.sca.stacked_traces.stacked_traces import StackedTraces
+from test.sca.perf_stacked_combine import (
+    generate_dataset,
+    report,
+    timed,
+)
+
+NREPS = 100
+
+
+def main():
+    rm_times_storage = []
+    cm_times_storage = []
+
+    for _ in range(NREPS):
+        rm_samples = generate_dataset(npr.default_rng(), 2 ** 10, 2 ** 15)
+        rm_data = StackedTraces(rm_samples)
+        rm_tm = GPUTraceManager(
+            rm_data,
+        )
+        timed(rm_times_storage, log=True)(rm_tm.average)()
+
+        cm_samples = np.array(rm_samples, order="F")
+        cm_data = StackedTraces(cm_samples)
+        cm_tm = GPUTraceManager(
+            cm_data,
+        )
+        timed(cm_times_storage, log=True)(cm_tm.average)()
+
+    rm_times = np.array([duration for _, duration in rm_times_storage[1:]])
+    cm_times = np.array([duration for _, duration in cm_times_storage[1:]])
+    print(f"Row major:\n\t{np.mean(rm_times) / 1e6:.3f}ms ± {np.std(rm_times) / 1e6:.3f}ms")
+    print(f"Col major:\n\t{np.mean(cm_times) / 1e6:.3f}ms ± {np.std(cm_times) / 1e6:.3f}ms")
+
+
+if __name__ == "__main__":
+    main()