On 11/11/16 15:42, James Greenhalgh wrote:
Hi,
Finally, having added support for single-step DFmode to HFmode conversions,
this patch adds support for _Float16 to the ARM back-end.
That means making sure that only __fp16 promotes and adding similar hooks to
those used in the AArch64 port giving the excess precision rules, and
marking HFmode as supported in libgcc.
Bootstrapped on an ARMv8-A machine, and crosstested with no issues.
OK?
This looks ok to me once the prerequisites are approved.
Thanks,
Kyrill
Thanks,
James
---
gcc/
2016-11-09 James Greenhalgh <james.greenha...@arm.com>
PR target/63250
* config/arm/arm-builtins.c (arm_simd_floatHF_type_node): Rename to...
(arm_fp16_type_node): ...This, make visibile.
(arm_simd_builtin_std_type): Rename arm_simd_floatHF_type_node to
arm_fp16_type_node.
(arm_init_simd_builtin_types): Likewise.
(arm_init_fp16_builtins): Likewise.
* config/arm/arm.c (arm_excess_precision): New.
(arm_floatn_mode): Likewise.
(TARGET_C_EXCESS_PRECISION): Likewise.
(TARGET_FLOATN_MODE): Likewise.
(arm_promoted_type): Only promote arm_fp16_type_node.
* config/arm/arm.h (arm_fp16_type_node): Declare.
gcc/testsuite/
2016-11-09 James Greenhalgh <james.greenha...@arm.com>
* lib/target-supports.exp (add_options_for_float16): Add
-mfp16-format=ieee when testign arm*-*-*.
