1D One-dimensional (as in 1D data distribution)Bandwidth (or throughput) for the network

2D Two-dimensional (as in 2D data distribution)

Total number of floating-point operations

CSRC_ Entry in DESC_ indicating the process column over which

the first column of the array is distributed

CTXT_ Entry in DESC_ indicating the BLACS context associated with the

Total number of data items communicated

DESC_ Array descriptor for a global array

DESCAArray descriptor for global arrayA

DESCBArray descriptor for global arrayBDLEN_ Length of the descriptor DESC_

DTYPE_ First entry of the descriptor DESC_, identifying

E() Estimated parallel efficiencyTime per floating point operation in matrix-matrix multiply

Gflop/s Gigaflops ( floating point operations) per second

IAGlobal row index in the global arrayAindicating the

ICTXTBLACS context associated with a process gridINFO Output integer argument of driver and computational routines

indicating the success or failure of the routine

JAGlobal column index in the global arrayAindicating theLLD_ Entry in DESC_ indicating the local leading dimension of the

Local leading dimension of the local array

ALocal leading dimension of the local array

B(K_) Number of columns that a process receives if columns

of a matrix are distributed over

ccolumns of its process row.(K_) Number of rows that a process would receive if rows

of a matrix are distributed over

rrows of its process column.

MGlobal number of rows of the distributed submatrixM_ Entry of DESC_ indicating the number of rows in the global

MBGlobal row block size for partitioning the global matrixMB_ Entry in DESC_ indicating the block size used to distribute

MB/s Megabyte per second

Mflop/s Megaflops ( floating point operations) per second

MYCOLThe calling process's column coordinate in the process grid.

MYROWThe calling process's row coordinate in the process grid

NGlobal number of columns of the distributed submatrixN_ Entry of DESC_ indicating the number of columns in the

NBGlobal column block size for partitioning the global matrixNB_ Entry in DESC_ indicating the block size used to distribute

the columns of the global array

NBRHSGlobal column block size for the solution matrix

NPCOLNumber of process columns in the process grid (equivalent to )

NPROCSTotal number of processes in the process grid (equivalent toP)

NPROWNumber of process rows in the process grid (equivalent to )

NRHSGlobal number of columns in the global solution matrixBNUMROC TOOLS routine used to calculate the number of rows or columns

PTotal number of processes in the process grid, i.e.,Number of process columns in the process grid

Number of process rows in the process grid

RSRC_ Entry in DESC_ indicating the process row over which the

first row of the array is distributed

Distributed submatrix

A(IA:IA+M-1,JA:JA+N-1)

T() Estimated parallel execution time() Estimated serial execution time

Time per floating-point operation (typically )

Time required to solve a problem of size

NonPprocessorsTime required to solve a problem of size

N/2 onPprocessorsTime per data item communicated

Tue May 13 09:21:01 EDT 1997