pv

$ python runtests.py --bench -t scipy/linalg/benchmarks/bench_levinson_durbin.py 
Building, see build.log...
Build OK
Running benchmarks for scipy
NumPy version 1.7.1
NumPy is installed in /usr/lib/python2.7/dist-packages/numpy
SciPy version 0.14.0.dev-649d452
SciPy is installed in /home/pauli/prj/scipy/scipy/build/testenv/lib/python2.7/site-packages/scipy
Python version 2.7.5+ (default, Sep 19 2013, 13:48:49) [GCC 4.8.1]
nose version 1.3.0

pv

from scipy.sparse import lil_matrix
from numpy import where, zeros, prod, vstack, ones
from time import time

n, m = 1000, 20000
M = vstack((zeros((n, m)), ones((25, m)))) # this changes time for lil very sufficiently
I, J = where(M)

# 1: time for sparse
M = lil_matrix(M)

pv

Function "fpinst" not defined.
Breakpoint 1 (fpinst) pending.
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".

Breakpoint 1, fpinst (iopt=0, t=..., n=104, c=..., k=3, x=0,5, l=81, tt=..., nn=0, cc=..., nest=106) at scipy/interpolate/fitpack/fpinst.f:1
1	      subroutine fpinst(iopt,t,n,c,k,x,l,tt,nn,cc,nest)
#0  fpinst (iopt=0, t=..., n=104, c=..., k=3, x=0,5, l=81, tt=..., nn=0, cc=..., nest=106) at scipy/interpolate/fitpack/fpinst.f:1
#1  0x00007fffeec867cc in insert (iopt=0, t=..., n=104, c=..., k=3, x=0,5, tt=..., nn=0, cc=..., nest=106, ier=0) at scipy/interpolate/fitpack/insert.f:100
#2  0x00007fffeec78742 in fitpack_insert (dummy=0x0, args=0x15afef0) at scipy/interpolate/src/__fitpack.h:827

pv

$ python runtests.py --bench \						     								
  -t scipy/sparse/benchmarks/bench_sparse.py:BenchmarkSparse.bench_setitem \
  -t scipy/sparse/benchmarks/bench_sparse.py:BenchmarkSparse.bench_getitem  								     


                 BEFORE (32cd96d98b040)                                                       AFTER (c5e3ae442639b98)			
									     								
           Sparse Matrix fancy __setitem__		   	   		                Sparse Matrix fancy __setitem__		   
========================================================== 	   	      ========================================================== 
      N | s.patt. |     csr |     csc |     lil |     dok |	   	            N | s.patt. |     csr |     csc |     lil |     dok |

pv

#!/usr/bin/env python
"""
counter.py

Count the number of elementwise and matrix multiplication call sites,
by running code and intercepting calls.
"""
import os
import sys
import argparse

pv

#include <locale.h>  
#include <stdio.h>  

int badcall_();  
int main() 
{  
        FILE *f;  
        f = fopen("bug.dat", "w");  
        fprintf(f, "1.2345\n");  
        fclose(f);  

pv

30
5 1 10 0 2 1
5 1 4 0 2 1
5 3 12 0 3 1
5 3 4 0 3 1
5 4 7 0 3 4
5 4 5 0 2 4
6 4 13 1 2 4 3
5 5 14 0 4 2
5 7 16 3 4 0

pv

import numpy as np
from scipy import sparse
import time
import pysparse

def main():
    print "%-6s %15s %15s " % ("% N", "scipy.sparse", "pysparse")

    for N in xrange(5, 16):
        N = 2**N

pv

               Levinson-Durbin vs. generic solver
          T x = y; T.shape == (n, n); y.shape == (n, m)
==============================================================
   n   |   m   |        dtype       | L-D time  | generic time
       |       |                    | (seconds) |             
--------------------------------------------------------------
   100 |     1 |            float64 |   0.00122 |   0.00189 
   100 |    10 |            float64 |   0.00727 |    0.0019 
   100 |   100 |            float64 |    0.0666 |   0.00459 
   100 |  1000 |            float64 |     0.614 |    0.0171 

pv

 Levinson-Durbin [Cython version from pr-4302] vs. generic solver
          T x = y; T.shape == (n, n); y.shape == (n, m)
==============================================================
   n   |   m   |        dtype       | L-D time  | generic time
       |       |                    | (seconds) |             
--------------------------------------------------------------
   100 |     1 |            float64 |  0.000103 |  0.000162 
   100 |    10 |            float64 |  0.000561 |  0.000167 
   300 |     1 |            float64 |  0.000258 |   0.00136