1
2
3
APPEN-
DIX
6
7
8
Appendix 3 Me
thod of R
epla
cing Basic Mod
el QCPU
or H
igh
Perfor
mance
Model QCPU with U
niversal
Model QCPU
Appendix 3.4 Fun
ctions
APPENDICES
App
-
44
(b) Replacing a part of floating-point operation instructions with double-precision floating-point
operation instructions
Only operations that require high accuracy are replaced with double-precision floating-point opera-
tion instructions.
Using the ECON and EDCON instructions, convert floating-point data mutually between single pre-
cision and double-precision. The flow of a replacement program is as follows:
• Data required for operations are converted from single precision to double precision using the ECON
instruction.
• Operations are performed in double precision using the double-precision floating-point operation instruc-
tions.
• Operation results are converted from double precision to single precision using the EDCON instruction.
A program example that floating-point data are converted mutually between single precision and
double precision before and after operations is shown below.
1)
Device assignment
2)
Program before replacement
Example
Replacing the floating-point operation [A
B+C] (Using the ECON and EDCON
instructions)
TableApp.37 Device assignment
Before replacement
After replacement
Application
Device
Data type
Application
Device
Data type
Data A
D0 to D1
Floating-point data
(single precision)
Data A
D0 to D1
Floating-point data
(single precision)
Data B
D2 to D3
Data B
D2 to D3
Data C
D4 to D5
Data C
D4 to D5
Result
D6 to D7
Result
D6 to D7
Data A
D10 to D13
Floating-point data
(double precision)
Data B
D14 to D17
Data C
D18 to D21
Result
D22 to D25
Figure App.22 Sample program
