Re: [PATCH 1/3] aarch64: Add support for fp8 convert and scale

2024-11-11 Thread Richard Sandiford
writes: > The AArch64 FEAT_FP8 extension introduces instructions for conversion > and scaling. > > This patch introduces the following intrinsics: > 1. vcvt{1|2}_{bf16|high_bf16|low_bf16}_mf8_fpm. > 2. vcvt{q}_mf8_f16_fpm. > 3. vcvt_{high}_mf8_f32_fpm. > 4. vscale{q}_{f16|f32|f64}. > > We introduc

Re: [PATCH 1/3] aarch64: Add support for fp8 convert and scale

2024-11-07 Thread Saurabh Jha
On 11/7/2024 9:03 AM, Kyrylo Tkachov wrote: Hi Saurabh, On 6 Nov 2024, at 11:03, saurabh@arm.com wrote: The AArch64 FEAT_FP8 extension introduces instructions for conversion and scaling. This patch introduces the following intrinsics: 1. vcvt{1|2}_{bf16|high_bf16|low_bf16}_mf8_fpm. 2.

Re: [PATCH 1/3] aarch64: Add support for fp8 convert and scale

2024-11-07 Thread Kyrylo Tkachov
Hi Saurabh, > On 6 Nov 2024, at 11:03, saurabh@arm.com wrote: > > > The AArch64 FEAT_FP8 extension introduces instructions for conversion > and scaling. > > This patch introduces the following intrinsics: > 1. vcvt{1|2}_{bf16|high_bf16|low_bf16}_mf8_fpm. > 2. vcvt{q}_mf8_f16_fpm. > 3. vcvt_

[PATCH 1/3] aarch64: Add support for fp8 convert and scale

2024-11-06 Thread saurabh.jha
The AArch64 FEAT_FP8 extension introduces instructions for conversion and scaling. This patch introduces the following intrinsics: 1. vcvt{1|2}_{bf16|high_bf16|low_bf16}_mf8_fpm. 2. vcvt{q}_mf8_f16_fpm. 3. vcvt_{high}_mf8_f32_fpm. 4. vscale{q}_{f16|f32|f64}. We introduced three new aarch64_built