Table 7-1. optimization table for callout blocks, Summary, Summary -19 – National Instruments AutoCode NI MATRIX User Manual

Page 210

Chapter 7

Code Optimization

7-19

Summary

All of the optimizations discussed so far can be used at the same time
without limitations. All of these optimizations can work together and
potentially bring about a cumulative effect in reducing the code size or
improving the execution speed of an application. In general, the restart
optimization, varblock read, and constant propagation can help reduce the
code size, while merging of INITs, constant propagation, and varblock read
optimizations improve execution speed. The reuse of temporaries helps
reduce stack size, and could prevent stack overflow problems. Prudent use
of the Constant Block can reduce code size, and its use is preferred over the
Algebraic Expression Block for defining constants. By using these
optimizations, you can generate optimal C or Ada source code. The
executable (or the object) file size depends on the compiler used for
compiling the source code. Execution speed of an application ultimately
depends on the target processor.

Table 7-1. Optimization Table for Callout Blocks

Block

Output (X)

Input A

Input B

MatrixInverse

Copy only if output is
not a single array

Mandatory copy-in

Not applicable

MatrixRightDivide
or MatrixLeftDivide
(if B is z

× n or n × 1)

Copy only if output is
not a single array

Mandatory copy-in

MatrixLeftDivide
(if B is m

× n, m > 1,

or n > 1)

Copy only if output is
not a single array

Mandatory copy-in

Copy only if input
connectivity prohibits
using source array
directly