From bb9345ca8a44473366adce1d3b172518ab962752 Mon Sep 17 00:00:00 2001 From: Aaron Teo Date: Sat, 21 Jun 2025 16:50:05 +0800 Subject: [PATCH] ggml-cpu: activate nnpa for ggml_cpu_fp32_to_fp16 Signed-off-by: Aaron Teo --- ggml/src/ggml-cpu/ggml-cpu.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/ggml/src/ggml-cpu/ggml-cpu.c b/ggml/src/ggml-cpu/ggml-cpu.c index 1d9cb7574..a0920f09e 100644 --- a/ggml/src/ggml-cpu/ggml-cpu.c +++ b/ggml/src/ggml-cpu/ggml-cpu.c @@ -3142,6 +3142,16 @@ void ggml_cpu_fp32_to_fp16(const float * x, ggml_fp16_t * y, int64_t n) { _mm_storel_epi64((__m128i *)(y + i), y_vec); } #endif + +#if defined(__NNPA__) + for (; i + 7 < n; i += 8) { + float32x4_t v_xh = vec_xl(0, (const float *)(x + i + 0)); + float32x4_t v_xl = vec_xl(0, (const float *)(x + i + 4)); + uint16x8_t v_xd = vec_round_from_fp32(v_xh, v_xl, 0); + vec_xst(v_xd, 0, (ggml_fp16_t *)(y + i)); + } +#endif + for (; i < n; ++i) { y[i] = GGML_FP32_TO_FP16(x[i]); }