print more comparaison info and small refactoring.

theano/misc/check_blas.py

@@ -4,42 +4,85 @@
 #A,B,C matrix
 #a,b scalar
 
+s="""
+result for shapes=(2000,2000) and iters=100
+GTX 470 7.22s
+GTX 285, 6.84s
+GTX 480 5.83s
+"""
+
 import theano,numpy,time
 import theano.tensor as T
-
 shapes=(2000,2000)
 iters = 10
 
 
-a=theano.shared(numpy.ones(shapes, dtype=theano.config.floatX))
-b=theano.shared(numpy.ones(shapes, dtype=theano.config.floatX))
-c=theano.shared(numpy.ones(shapes, dtype=theano.config.floatX))
+def execute(verbose=True):
 
-f=theano.function([],updates={c:0.4*c+.8*T.dot(a,b)})
-print 'blas.ldflags=',theano.config.blas.ldflags
-print 'compiledir=',theano.config.compiledir
-if any( [x.op.__class__.__name__=='Gemm' for x in f.maker.env.toposort()]):
-    print 'Used the cpu'
-elif any( [x.op.__class__.__name__=='GpuGemm' for x in f.maker.env.toposort()]):
-    print 'Used the gpu'
+    a=theano.shared(numpy.ones(shapes, dtype=theano.config.floatX))
+    b=theano.shared(numpy.ones(shapes, dtype=theano.config.floatX))
+    c=theano.shared(numpy.ones(shapes, dtype=theano.config.floatX))
 
-print f.maker.env.toposort()
+    f=theano.function([],updates={c:0.4*c+.8*T.dot(a,b)})
+    if verbose:
+        print 'Some theano flags:'
+        print '    blas.ldflags=',theano.config.blas.ldflags
+        print '    compiledir=',theano.config.compiledir
+        print '    floatX=',theano.config.floatX
+        print 
+        print 'Numpy config:(used when the theano flags "blas.ldflags" is empty)'
+        numpy.show_config(); 
+        print 'Numpy dot module:',numpy.dot.__module__; 
+        print 'Numpy file location that was loaded:',numpy.__file__;
+        print 'Numpy version:',numpy.__version__
+        print 
+        if any( [x.op.__class__.__name__=='Gemm' for x in f.maker.env.toposort()]):
+            print 'Used the cpu'
+        elif any( [x.op.__class__.__name__=='GpuGemm' for x in f.maker.env.toposort()]):
+            print 'Used the gpu'
+        else:
+            print 'ERROR, not able to tell if theano used the cpu or the gpu'
+            print f.maker.env.toposort()
 
-t0=time.time()
-for i in range(iters):
-    f()
-print
-print 'times=%.2fs'%(time.time()-t0)
+    t0=time.time()
+    for i in range(iters):
+        f()
+    t1=time.time()
+    if verbose:
+        print
+        print 'this execution time took %.2fs'%(t1-t0)
+    return t1-t0
 
-print """
-Some result that you can compare again on the cpu:
 
-Lib: goto
-computer: Intel(R) Xeon(R) CPU E5430  @ 2.66GHz
+def jobman_job(state, channel):
+    execute()
+    return channel.COMPLETE
 
-OMP_NUM_THREADS     time
-1                   16.05s
-2                   8.22s
-4                   4.13s
-8                   2.32s
-"""
+def test():
+    execute()
+    
+
+if __name__ == "__main__":
+    execute()
+    print """
+ 	Some result that you can compare again. They where 10 executions of gemm in float64 with matrix of shape 2000x2000 on FC9.
+ 	
+        We tested 3 cpus: Xeon E5345, Xeon E5430 and Xeon E5450
+
+        Lib tested:
+            * numpy with ATLAS from distribution(FC9) package (1 thread)
+            * manually compiled numpy and ATLAS with 2 threads
+            * goto with 1, 2, 4 and 8 threads.
+
+ 	lib/nb threads    E5345(s)  E5430(s)  E5450(s)
+
+        numpy_FC9_atlas/1 39.2s     35.0s     30.7s
+ 	goto/1            18.7s     16.1s     14.2s
+        numpy_MAN_atlas/2 12.0s     11.6s     10.2s
+ 	goto/2            9.5s      8.1s      7.1s
+ 	goto/4            4.9s      4.4s      3.7s
+ 	goto/8            2.7s      2.4s      2.0s
+ 	"""
+
+    print 
+    print "We timed",iters,"executions of gemm with matrix of shapes",shapes