On Tue, Jul 28, 2015 at 12:25:06PM +0100, Alan Lawrence wrote:
> gcc/ChangeLog:
>
> * config/aarch64/aarch64-builtins.c (aarch64_fp16_type_node): New.
> (aarch64_init_builtins): Make aarch64_fp16_type_node, use for __fp16.
>
> * config/aarch64/aarch64-modes.def: Add HFmode.
>
> * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define
> __ARM_FP16_FORMAT_IEEE and __ARM_FP16_ARGS. Set bit 1 of __ARM_FP.
>
> * config/aarch64/aarch64.c (aarch64_init_libfuncs,
> aarch64_promoted_type): New.
>
> (aarch64_float_const_representable_p): Disable HFmode.
> (aarch64_mangle_type): Mangle half-precision floats to "Dh".
> (TARGET_PROMOTED_TYPE): Define to aarch64_promoted_type.
> (TARGET_INIT_LIBFUNCS): Define to aarch64_init_libfuncs.
>
> * config/aarch64/aarch64.md (mov<mode>): Include HFmode using GPF_F16.
> (movhf_aarch64, extendhfsf2, extendhfdf2, truncsfhf2, truncdfhf2): New.
>
> * config/aarch64/iterators.md (GPF_F16): New.
>
>
> gcc/testsuite/ChangeLog:
>
> * gcc.target/aarch64/f16_movs_1.c: New test.
OK with some minor tweaks to some comments.
> diff --git a/gcc/config/aarch64/aarch64-modes.def
> b/gcc/config/aarch64/aarch64-modes.def
> index b17b90d..c30059b 100644
> --- a/gcc/config/aarch64/aarch64-modes.def
> +++ b/gcc/config/aarch64/aarch64-modes.def
> @@ -36,6 +36,10 @@ CC_MODE (CC_DLTU);
> CC_MODE (CC_DGEU);
> CC_MODE (CC_DGTU);
>
> +/* Half-precision floating point for arm_neon.h float16_t. */
> +FLOAT_MODE (HF, 2, 0);
> +ADJUST_FLOAT_FORMAT (HF, &ieee_half_format);
> +
Minor technicality. This is for the __fp16 type, which arm_neon.h aliases
to float16_t. As __fp16 is the name given in the AAPCS64 for the type,
I'd prefer if we used that in the comments documenting the feature.
So, s/arm_neon.h float16_t/__fp16/
> +/* Implement TARGET_PROMOTED_TYPE to promote float16 to 32 bits. */
Reword as something like:
To promote "__fp16" to "float"
> #undef TARGET_ADDRESS_COST
> #define TARGET_ADDRESS_COST aarch64_address_cost
>
> diff --git a/gcc/config/aarch64/aarch64.h b/gcc/config/aarch64/aarch64.h
> index a22c6e4..44fe4f9 100644
> --- a/gcc/config/aarch64/aarch64.h
> +++ b/gcc/config/aarch64/aarch64.h
> @@ -57,7 +57,9 @@
> if (TARGET_FLOAT) \
> { \
> builtin_define ("__ARM_FEATURE_FMA"); \
> - builtin_define_with_int_value ("__ARM_FP", 0x0C); \
> + builtin_define_with_int_value ("__ARM_FP", 0x0E); \
> + builtin_define ("__ARM_FP16_FORMAT_IEEE"); \
> + builtin_define ("__ARM_FP16_ARGS"); \
You'll be in competition with Kyrill's changes for target attributes here,
he moves these all to a new file.
> diff --git a/gcc/config/aarch64/iterators.md b/gcc/config/aarch64/iterators.md
> index 498358a..a6b351b 100644
> --- a/gcc/config/aarch64/iterators.md
> +++ b/gcc/config/aarch64/iterators.md
> @@ -38,6 +38,9 @@
> ;; Iterator for General Purpose Floating-point registers (32- and 64-bit
> modes)
> (define_mode_iterator GPF [SF DF])
>
> +;; Iterator for General Purpose Float regs, inc float16_t.
> +(define_mode_iterator GPF_F16 [HF SF DF])
> +
s/float16_t/__fp16/
Thanks,
James
