Went through all NumPy benchmarks got them running and documented them.

Of the 36 benchmarks 14 has implementation issues which includes segfaulting, cannot run with Bohrium due to lack of Bohrium features and other things.

blacks_scholes-python_numpy: Supports dogma "High-Performance NumPy without changing a single line of code.".

convolve-python_numpy: Dogmatized and documented Bohrium related issues.

convolve_2d-python_numpy: Dogmatized and documented Bohrium related issues.

convolve_3d-python_numpy: Dogmatized and documented Bohrium related issues.

convolve_separate-python_numpy: Dogmatized and documented Bohrium related issues.

gameoflife-python_numpy: Dogmatized.

gauss-python_numpy: Dogmatized.

Forgot this...

heat_equation-python_numpy: Dogmatized.

idl_init_bh-python-numpy: Fixed encoding error and noted issues.

idl_init_fast-python-numpy: Documented issues.

idl_init_orig-python-numpy: Documented issues.

jacobi: Added a brief description and dogmatized NumPy version.

jacobi-fixed: Added a brief description, dogmatized NumPy version and documented use of Bohrium features.

jacobi_solve: Added a brief description and dogmatized NumPy version.

k_nearest_neighbor: Documented issues.

knn_naive: Added description and dogmatized Numpy implementation.

knn_naive1: Added description and dogmatized Numpy implementation.

knn_naive: Removed since it is a duplicate of knn_naive1.

Leave knn_naive1 since it has history, knn_naive does not.

lattice_boltzmann_D2Q9: Documented issues and use.

lbm_3d: Documented use and dogmatized NumPy implementation.

LMM_swaption_vec: Documented use and issues.

lu: Brief docs and dogmatization.

Moved the LMM swaption benchmark.

benchmark-autodoc: Added display of issues and bohrium info.

Added notes in benchmarks matrix about issues and bohrium specifics.

jacobi_stencil: Same story...

mc: Same story...

mxmul: Same story.

nbody: Same story.

nbody: Same story.

same story

nbody_nice: Same story.

ndstencil: Same story.

point27: Same story.

ndstencil: Added dogma.

autodoc-benchmarks: Added the dogma-note to benchmark-matrix.

Pricing and Shallow water.

snakes_and_ladders...

...

sor ...

more...

Went through all NumPy benchmarks got them running and documented them.

Of the 36 benchmarks 14 has implementation issues which includes segfaulting, cannot run with Bohrium due to lack of Bohrium features and other things.

benchmarks/black_scholes/python_numpy/black_scholes.py

@@ -1,9 +1,6 @@
 from __future__ import print_function
-import util
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
 
 def model(N, B):
     """Construct pseudo-data representing price samples?"""
@@ -62,17 +59,17 @@ def main():
     if B.inputfn:
         S = B.load_array()
     else:
-        S = model(N, B)              # Construct pseudo-data
+        S = model(N, B)                     # Construct pseudo-data
 
     if B.dumpinput:
         B.dump_arrays("black_scholes", {'input': S})
 
     B.start()
-    R = price(S, I, visualize=B.visualize)   # Run the model
+    R = price(S, I, visualize=B.visualize)  # Run the model
     B.stop()
     B.pprint()
-    # Convert to one array
-    for i in range(len(R)):
+
+    for i in range(len(R)):                 # Convert to one array
         if hasattr(R[i], "copy2numpy"):
             R[i] = R[i].copy2numpy()[()]
     if B.outputfn:

benchmarks/black_scholes/readme.rst

@@ -0,0 +1,4 @@
+A finance thing::
+
+  --size=1000000*10
+

benchmarks/convolve/python_numpy/bohrium.rst

@@ -0,0 +1,3 @@
+The view generator ``bohrium.stdviews`` module is used.
+
+Specifically the ``cartesian`` generator.

benchmarks/convolve/python_numpy/convolve.py

@@ -1,10 +1,7 @@
 from __future__ import print_function
 from PIL import Image
-import util
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
 from bohrium.stdviews import cartesian
 
 def apply_filter(size, weight):

benchmarks/convolve/python_numpy/issues.rst

@@ -0,0 +1,9 @@
+Fails when running with Bohrium::
+
+  python -m bohrium convolve.py --size=25 --bohrium=True
+
+Produces the following error::
+
+  ~/.local/lib/python2.7/site-packages/bohrium/__main__.py:20: RuntimeWarning: Encounter ing an operation not supported by Bohrium. It will be handled by the original NumPy. execfile(sys.argv[0])
+  Segmentation fault (core dumped)
+

benchmarks/convolve_2d/python_numpy/bohrium.rst

@@ -0,0 +1,3 @@
+The view generator ``bohrium.stdviews`` module is used.
+
+Specifically the ``cartesian`` generator.

benchmarks/convolve_2d/python_numpy/convolve_2d.py

@@ -1,10 +1,7 @@
 from __future__ import print_function
 from PIL import Image
-import util
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
 from bohrium.stdviews import cartesian
 
 def gen_2d_filter(size, weight, datatype=np.float32):

benchmarks/convolve_2d/python_numpy/issues.rst

@@ -0,0 +1,9 @@
+Fails when running with Bohrium::
+
+  python -m bohrium convolve_2d.py --size=25 --bohrium=True
+
+Produces the following error::
+
+  ~/.local/lib/python2.7/site-packages/bohrium/__main__.py:20: RuntimeWarning: Encounter ing an operation not supported by Bohrium. It will be handled by the original NumPy. execfile(sys.argv[0])
+  Segmentation fault (core dumped)
+

benchmarks/convolve_3d/python_numpy/bohrium.rst

@@ -0,0 +1,3 @@
+The view generator ``bohrium.stdviews`` module is used.
+
+Specifically the ``cartesian`` generator.

benchmarks/convolve_3d/python_numpy/convolve_3d.py

@@ -1,10 +1,7 @@
 from __future__ import print_function
 from PIL import Image
-import util
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
 from bohrium.stdviews import cartesian
 
 def gen_3d_filter(size, weight, datatype=np.float32):

benchmarks/convolve_3d/python_numpy/issues.rst

@@ -0,0 +1,10 @@
+Fails when running with Bohrium::
+
+  python -m bohrium convolve_3d.py --size=25 --bohrium=True
+
+Produces the following error::
+
+  .local/lib/python2.7/site-packages/bohrium/__main__.py:21: RuntimeWarning: Encountering an operation not supported by Bohrium. It will be handled by the original NumPy.
+    else:
+  Segmentation fault (core dumped)
+

benchmarks/convolve_separate_std/python_numpy/convolve_separate_std.py

@@ -1,12 +1,8 @@
 from __future__ import print_function
 import sys
 from numpy.linalg import svd
-import util
-
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
 
 def generate_gauss_matrix(filter_shape, sigma, data_type=np.float32):
     """

benchmarks/convolve_separate_std/python_numpy/issues.rst

@@ -0,0 +1,8 @@
+Running::
+
+  python -m bohrium convolve_separate_std.py --size=500*15 --bohrium=True
+
+Produces the error::
+
+  Segmentation fault (core dumped)
+  

benchmarks/gameoflife/python_numpy/gameoflife.py

@@ -5,11 +5,8 @@
 
 So what does this code example illustrate?
 """
-import util
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
 
 SURVIVE_LOW     = 2
 SURVIVE_HIGH    = 3

benchmarks/gauss/python_numpy/gauss.py

@@ -1,9 +1,7 @@
 from __future__ import print_function
-import util
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
+
 import bohrium.linalg as la
 
 def main():

benchmarks/heat_equation/python_numpy/heat_equation.py

@@ -5,11 +5,8 @@
 
 So what does this code example illustrate?
 """
-import util
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
 
 def freezetrap(height, width, dtype=np.float32):
     """Construct the grid."""

benchmarks/idl_init_bh/python_numpy/idl_init_bh.py

@@ -1,3 +1,4 @@
+# This Python file uses the following encoding: utf-8
 from __future__ import print_function
 from time import time
 from sys import argv

benchmarks/idl_init_bh/python_numpy/issues.rst

@@ -0,0 +1,2 @@
+Several issues with this Benchmark, it might have worked a year ago, but it currently relies on Bohrium features that no longer exists and other problems.
+

benchmarks/idl_init_fast/python_numpy/idl_init_fast.py

@@ -1,3 +1,4 @@
+# This Python file uses the following encoding: utf-8
 from __future__ import print_function
 from time import time
 import numpy as np

benchmarks/idl_init_fast/python_numpy/issues.rst

@@ -0,0 +1,7 @@
+Several issues, among them:
+
+ * Does not use the benchpress util
+ * Fails horribly when executed with Bohrium
+ * ...
+
+Needs a lot of love.

benchmarks/idl_init_orig/python_numpy/issues.rst

@@ -0,0 +1,7 @@
+Several issues, among them:
+
+ * Does not use the benchpress util
+ * Fails horribly when executed with Bohrium
+ * ...
+
+Needs a lot of love.

benchmarks/jacobi/python_numpy/bohrium.rst

@@ -0,0 +1 @@
+Uses Bohrium features from ``bohrium.stdviews`` specifically ``no_border``, ``grid``, and ``diagonals``.

benchmarks/jacobi/python_numpy/jacobi.py

@@ -1,9 +1,6 @@
 from __future__ import print_function
-import util
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
 from bohrium.stdviews import no_border, grid, diagonals
 
 def jacobi_init(size):

benchmarks/jacobi/readme.rst

@@ -0,0 +1,3 @@
+This implementation runs jacobi on a quadratic grid until it converges.
+
+Example params: --size=1000

benchmarks/jacobi_fixed/python_numpy/bohrium.rst

@@ -0,0 +1 @@
+Uses Bohrium features from ``bohrium.stdviews`` specifically ``no_border``, ``grid``, and ``diagonals``.

benchmarks/jacobi_fixed/python_numpy/jacobi_fixed.py

@@ -1,9 +1,6 @@
 from __future__ import print_function
-import util
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
 from bohrium.stdviews import no_border, grid, diagonals
 
 def jacobi_fixed_init(size):

benchmarks/jacobi_fixed/readme.rst

@@ -0,0 +1,4 @@
+Run jacobi for a fixed number of iterations on a quadratic grid::
+
+  python jacobi_fixed.py --size=5000*10
+

benchmarks/jacobi_solve/python_numpy/jacobi_solve.py

@@ -1,9 +1,6 @@
 from __future__ import print_function
-import util
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
 
 def freezetrap(height, width, dtype=np.float32):
     grid        = np.zeros((height+2,width+2), dtype=dtype)

benchmarks/jacobi_solve/readme.rst

@@ -0,0 +1,4 @@
+Runs Jacobi on a rectangular grid for until it converges or until max iterations is reached::
+
+  --size=1000*1000*100
+

benchmarks/jacobi_stencil/python_numpy/jacobi_stencil.py

@@ -1,9 +1,6 @@
 from __future__ import print_function
-import util
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
 
 def freezetrap(height, width, dtype=np.float32):
     grid        = np.zeros((height+2,width+2), dtype=dtype)

benchmarks/jacobi_stencil/readme.rst

@@ -0,0 +1,4 @@
+Runs jacobi on a rectangular grid without convergence for the given amount of iterations::
+
+  --size=5000*5000*10
+

benchmarks/k_nearest_neighbor/python_numpy/issues.rst

@@ -0,0 +1 @@
+Does not actually call the implementation.

benchmarks/k_nearest_neighbor/python_numpy/k_nearest_neighbor.py

@@ -5,11 +5,8 @@
 
 So what does this code example illustrate?
 """
-import util
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
 
 def classify(sample, training, group=None, k=1):
     assert sample.ndim == 2

benchmarks/k_nearest_neighbor/readme.rst

@@ -0,0 +1 @@
+A think this was an attempt at writing a better KNN implementation.

benchmarks/knn_naive1/python_numpy/knn_naive1.py

@@ -1,9 +1,6 @@
 from __future__ import print_function
-import util
-if util.Benchmark().bohrium:
-    import bohrium as np
-else:
-    import numpy as np
+from benchpress import util
+import numpy as np
 
 def data_range(B, N, F):
     """Pseudo-data setup using Python-range."""

benchmarks/knn_naive1/readme.rst

@@ -0,0 +1,4 @@
+Naive implementation of KNN --size=N*Features*Neighbors::
+
+  --size=20000000*10*4
+

benchmarks/lattice_boltzmann_D2Q9/python_numpy/issues.rst

@@ -0,0 +1,4 @@
+Copies data back and forth between NumPy and Bohrium::
+
+  .local/lib/python2.7/site-packages/bohrium/__main__.py:55: RuntimeWarning: Encountering an operation not supported by Bohrium. It will be handled by the original NumPy.
+