Table 29 Fine-tuning optimization options (continued)

Level

Level

Function

 

 

collisions. The default is

 

 

+Onocache_pad_common.

+O[no]dataprefetch

+O2 or higher

Insert [do not insert]

 

 

instructions within innermost

 

 

loops to explicitly prefetch

 

 

data from memory into the

 

 

data cache. The default is

 

 

+Onodataprefetch.

+O[no]dynopt

All

Enable [disable] dynamic

 

 

optimization for the output

 

 

file, if the run-time

 

 

environment supports this

 

 

feature. Both forms of this

 

 

option change the default

 

 

setting, which allows the

 

 

run-time environment to

 

 

enable or disable dynamic

 

 

optimization according to a

 

 

system-wide default. This

 

 

option applies only to

 

 

executable files and shared

 

 

libraries.chatr(1) can be

 

 

used to change this setting,

 

 

including restoration of the

 

 

default setting, after the

 

 

output file has been created.

+O[no]entrysched

All

Perform [do not perform]

 

 

instruction scheduling on

 

 

entry and exit code. The

 

 

default is

 

 

+Onoentrysched.

 

 

NOTE: This option is valid

 

 

only on the PA-RISC systems.

+O[no]fastaccess

All

Enable [disable] fast access

 

 

to global data. The default

 

 

is +Onofastaccessat

 

 

levels 1, 2, and 3;

 

 

+Ofastaccessat level 4.

+Ofast

All

Select a combination of

 

 

optimization options for

 

 

optimum execution speed

 

 

and reasonable build times.

 

 

Currently chosen options

 

 

are:

 

 

+O2

 

 

+Ofltacc=relaxed

 

 

+Onolimit

 

 

+DSnative

 

 

+FPD

 

 

Wl,+pi,1M-

 

 

-Wl,+pd,1M

 

 

-Wl,+mergeseg

Using options to control optimization

93

Page 93
Image 93
HP UX Fortran Software Is +Onofastaccess at, +Ofastaccess at level, +O2, +Ofltacc=relaxed, +DSnative, Wl,+pi,1M, Wl,+pd,1M