D04.07. FMS Printing Parameter IPRMV

FMS Printing Parameter IPRMV

Data Type

Integer

Default Value

DESCRIPTION:

This parameter controls the amount of printed output produced by calls to the matrix-vector multiply routines RSDMVM, RNDMVM, CHDMVM, CSDMVM, CNDMVM, RSDSVM, CHDSVM, CSDSVM, and RNDVMM, CNDVMM. The print code summary table lists the values of IPRMV and the output produced.

The following is a detail description of each option. The value of IPRMV you specify is obtained by adding together the numeric value of each option you select. Because these numeric values are a power of 2, they have the effect of setting bits in IPRMV.

0
No output from matrix-vector multiply is printed.

+1
Start and stop timing is printed. The following is an example:

 START OF FMS MATRIX-VECTORS MULTIPLY
 END   OF FMS MATRIX-VECTORS MULTIPLY.
 TIME(SEC): CPU   =       4.941,        WALL=       4.985

+2
Timing information for each vector group is printed. The following is an example:

 F M S     M A T R I X   -   V E C T O R S     M U L T I P L Y
 <----FUNCTION--->   <--VECTORS-->     <-CPU  TIME(SEC)->     <-WALL TIME(SEC)->
                     FIRST    LAST     DELTA        TOTAL     DELTA        TOTAL
 M*V   MULTIPLY          1      24     1.159        1.159     1.204        1.204
 M*V   MULTIPLY         25      48     1.143        2.301     1.160        2.364
 M*V   MULTIPLY         49      72     1.166        3.468     1.184        3.548
 M*V   MULTIPLY         73      96     1.177        4.645     1.199        4.747
 M*V   MULTIPLY         97     100     0.288        4.933     0.303        5.050

+4
The {X} vectors are printed before matrix multiplication.
+8
The [F] matrix is printed before matrix multiplication.
(RNDVMM, CNDVMM only).
+16
The {X} vectors are printed after diagonal multiplication, [D]{X}.
+32
The output {Y} = [A]{X} vectors are printed after matrix multiplication.
+64
The diagonals [D] are printed.
+256
The lower triangle matrix [AL] and upper triangle matrix [AU] (nonsymmetric) are printed.
+1024
The actual performance achieved in Megaflops is printed. This computation is based on an actual count of the number of floating point operations (adds and multiplies) required and the actual time spent.
NOTE: The operation count may take some time. Therefore this option is not recommended as a default for production applications. It is provided for application tuning.

CAUTION: For large problems this parameter can produce a lot of output. You can use the FMS Parameters NEQPR1 and NEQPR2 to create a window on the matrix and vector terms printed.