speed up a numpy code with huge array

[Vincent Davis]

It makes me think you are filling you available memory and using the disk as
cache. If this happens things will get real slow. You might take a look at
your system resources when this is running. I don't have much else to offer
and could be completely wrong.

Vincent

On Tue, May 25, 2010 at 1:05 PM, Alexzive <zasaconsulting at gmail.com> wrote:

> Hello Pythonguys!
>
> is there a way to improve the performance of the attached code ? it
> takes about 5 h on a dual-core (using only one core) when len(V)
> ~1MIL. V is an array which is supposed to store all the volumes of
> tetrahedral elements of a grid whose coord. are stored in NN (accessed
> trough the list of tetraelements --> EL)
>
>
> Thanks in advance!
> Alex
>
> ####
> print 'start ' + nameodb
> #path = '/windows/D/SIM-MM/3D/E_ortho/' + nameodb + '.odb'
> path = pt + nameodb + '.odb'
> odb = openOdb(path)
>
> N = odb.rootAssembly.instances['PART-1-1'].nodes
> if loadV==1:
>  pathV=pt+vtet
>  V=numpy.loadtxt(pathV)
>  VTOT = V[0]
>  L3 = V[1]
>  print 'using ' + vtet
> else:
>  NN=[]
>  B=[0,0,0,0]
>  for i in range(len(N)):
>        B[0] = N[i].label
>        B[1] = N[i].coordinates[0]
>        B[2] = N[i].coordinates[1]
>        B[3] = N[i].coordinates[2]
>        NN = append(NN,B)
>
>  NN=NN.reshape(-1,4)
>  EL = odb.rootAssembly.instances['PART-1-1'].elements
>
>  L1 = max(NN[:,1])-min(NN[:,1])
>  L2 = max(NN[:,2])-min(NN[:,2])
>  L3 = max(NN[:,3])-min(NN[:,3])
>  VTOT=L1*L2*L3
>  print 'VTOT: [mm³]' + str(VTOT)
>
>  V = array([])
>
>  print 'calculating new Vtet '
>  V = range(len(EL)+2)
>  V[0] = VTOT
>  V[1] = L3
>  for j in range(0,len(EL)):
>        Va = EL[j].connectivity[0]
>        Vb = EL[j].connectivity[1]
>        Vc = EL[j].connectivity[2]
>        Vd = EL[j].connectivity[3]
>        ix = where(NN[:,0] == Va)
>        Xa = NN[ix,1][0][0]
>        Ya = NN[ix,2][0][0]
>        Za = NN[ix,3][0][0]
>        ix = where(NN[:,0] == Vb)
>        Xb = NN[ix,1][0][0]
>        Yb = NN[ix,2][0][0]
>        Zb = NN[ix,3][0][0]
>        ix = where(NN[:,0] == Vc)
>        Xc = NN[ix,1][0][0]
>        Yc = NN[ix,2][0][0]
>        Zc = NN[ix,3][0][0]
>        ix = where(NN[:,0] == Vd)
>        Xd = NN[ix,1][0][0]
>        Yd = NN[ix,2][0][0]
>        Zd = NN[ix,3][0][0]
>        a =  [Xa,Ya,Za]
>        b =  [Xb,Yb,Zb]
>        c =  [Xc,Yc,Zc]
>        d =  [Xd,Yd,Zd]
>        aa = numpy.diff([b,a],axis=0)[0]
>        bb = numpy.diff([c,b],axis=0)[0]
>        cc = numpy.diff([d,c],axis=0)[0]
>        D=array([aa,bb,cc])
>        det=numpy.linalg.det(D)
>        V[j+2] = abs(det)/6
>  pathV = pt + vtet
>  savetxt(pathV, V, fmt='%.3e')
> ###
> --
> http://mail.python.org/mailman/listinfo/python-list
>

  *Vincent Davis
720-301-3003 *
vincent at vincentdavis.net
 my blog <http://vincentdavis.net> |
LinkedIn<http://www.linkedin.com/in/vincentdavis>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/python-list/attachments/20100525/3050d7fb/attachment-0001.html>