-Xuse_fp16 [Professional Edition only] [V1.05.00 or later]
|
This option enables the half-precision floating-point type.
[Specification format]
- | Interpretation when omitted |
The half-precision floating-point type is disabled.
[Detailed description]
- | This option enables the half-precision floating-point type, which is a unique type extended from the C language standard. |
- | Specify "on" or "off" in value. If value is omitted, it is assumed that "on" has been specified. |
- | In the following cases, a warning will be output and this option will be ignored. |
- | This option is specified simultaneously with the -strict_std. |
- | This option is specified simultaneously with the -Xcpu=g3k option. |
- | This option is specified simultaneously with the -Xfloat=soft option. |
- | This option is specified simultaneously with the -Xround=zero option. |
[Example of use]
- | To enable the __fp16 type, describe as: |