[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH 03/22] target/arm: Implement VFP fp16 for VFP_BINOP operation
|
From: |
Peter Maydell |
|
Subject: |
Re: [PATCH 03/22] target/arm: Implement VFP fp16 for VFP_BINOP operations |
|
Date: |
Thu, 27 Aug 2020 14:39:22 +0100 |
On Tue, 25 Aug 2020 at 19:14, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> On 8/24/20 7:29 AM, Peter Maydell wrote:
> > +DEF_HELPER_3(vfp_addh, f32, f32, f32, ptr)
>
> We do have an f16 entry in helper-head.h, for documentation purposes. It
> expands to the same uint32_t, so it doesn't make any practical difference.
>
>
> > +float32 VFP_HELPER(name, h)(float32 a, float32 b, void *fpstp) \
>
> And here it would be better to use uint32_t explicitly, because we're
> definitely not returning float32.
>
> I guess you could see if dh_ctype_f16 works here? But that looks a bit ugly
> to
> me. Perhaps we should have introduced a better typedef somewhere...
dh_ctype_f16 does work, yes. Unless you can think of a better name
I think we should go with using that. I'm not sure that "16 bit float
held in a 32-bit unsigned integer for the benefit of the TCG calling
convention" is ever going to have a particularly snappy name...
thanks
-- PMM
- [PATCH 00/22] target/arm: Implement fp16 for AArch32 VFP, Peter Maydell, 2020/08/24
- [PATCH 01/22] target/arm: Remove local definitions of float constants, Peter Maydell, 2020/08/24
- [PATCH 02/22] target/arm: Use correct ID register check for aa32_fp16_arith, Peter Maydell, 2020/08/24
- [PATCH 03/22] target/arm: Implement VFP fp16 for VFP_BINOP operations, Peter Maydell, 2020/08/24
- [PATCH 05/22] target/arm: Macroify trans functions for VFMA, VFMS, VFNMA, VFNMS, Peter Maydell, 2020/08/24
- [PATCH 04/22] target/arm: Implement VFP fp16 VMLA, VMLS, VNMLS, VNMLA, VNMUL, Peter Maydell, 2020/08/24
- [PATCH 06/22] target/arm: Implement VFP fp16 for fused-multiply-add, Peter Maydell, 2020/08/24
- [PATCH 07/22] target/arm: Macroify uses of do_vfp_2op_sp() and do_vfp_2op_dp(), Peter Maydell, 2020/08/24
- [PATCH 08/22] target/arm: Implement VFP fp16 for VABS, VNEG, VSQRT, Peter Maydell, 2020/08/24