#!/usr/bin/env python
# coding: utf-8

# # BLAS Level 2 vs Level 3
# 
# Copyright (C) 2024 Andreas Kloeckner
# 
# <details>
# <summary>MIT License</summary>
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
# 
# The above copyright notice and this permission notice shall be included in
# all copies or substantial portions of the Software.
# 
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
# THE SOFTWARE.
# </details>

# In[1]:


import os
os.environ["OPENBLAS_NUM_THREADS"] = "1"

import numpy as np


# In[2]:


n = 2048
A = np.random.randn(n,n)
B = np.random.randn(n,n)


# Compare the timings for the following two operations, which perform the same work:

# In[3]:


get_ipython().run_cell_magic('timeit', '', 'A@B\n')


# In[4]:


get_ipython().run_cell_magic('timeit', '', 'for i in range(n):\n    A @ B[:,i]\n')


# - **Question:** Why this striking difference?
# - **Question:** Is the Python overhead (the `n`-long loop) likely to be a significant contributor to the timing difference?
# - **Question:** What BLAS functions underlie these nice, high-level numpy functions? (Using `perf top -z` in a Terminal may help.)

# In[5]:


get_ipython().run_cell_magic('timeit', '', 'for i in range(n):\n    A\n    B[:,i]\n')


# In[ ]: