This HTML automatically generated with rman for NEMO
Table of Contents

Name

ccdreorder - benchmark reordering (with optional openmp)

Synopsis

ccdreorder [parameter=value]

Description

Still in benchmark mode only.

Parameters

The following parameters are recognized in any order if the keyword is also given:
in=
Input file (ignored for now) []
out=
Output file (ignored for now) []
order=
New order of output cube (0=first dimension) [0,1,2]
dims=
Dimensions of the cube. The first listed dimension is the one fastest running in memory. In C we use row-major, so data[30][20][10] in this case. [10,20,30]
seed=
Random seed for [0,1] in the created cube [123]
iter=
Times to repeat the reorder. With iter=0 one can conveniently estimate the overhead. [1]

Examples

The following benchmarks were tried: 
time ccdreorder . . dims=100,100,100  iter=512 order=0,1,2  #  0.79
time ccdreorder . . dims=100,100,100  iter=512 order=2,1,0  #  2.32
time ccdreorder . . dims=200,200,200  iter=64  order=0,1,2  #  0.96
time ccdreorder . . dims=200,200,200  iter=64  order=2,1,0  #  3.44  
time ccdreorder . . dims=400,400,400  iter=8   order=0,1,2  #  1.71
time ccdreorder . . dims=400,400,400  iter=8   order=2,1,0  #  7.27
time ccdreorder . . dims=800,800,800  iter=1   order=0,1,2  #  7.46
time ccdreorder . . dims=800,800,800  iter=1   order=2,1,0  # 26.62
time ccdreorder  . . dims=11,10000,1000 order=0,2,1 help=cm iter=10  #  11.0
See also "make bench1" in NEMO/src/image/trans

Author

Peter Teuben

Update History


26-Dec-19    V0.1 Created    PJT

Table of Contents