From 7695a5331230c585f5ce92291c4256973985ae5a Mon Sep 17 00:00:00 2001 From: Georgi Gerganov Date: Fri, 10 Jul 2026 11:06:42 +0300 Subject: [PATCH] ggml : use ggml_vqtbl1q_u8 for 32-bit compat (#0) --- ggml/src/ggml-cpu/arch/arm/quants.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/ggml/src/ggml-cpu/arch/arm/quants.c b/ggml/src/ggml-cpu/arch/arm/quants.c index 636d7be12..b988abf99 100644 --- a/ggml/src/ggml-cpu/arch/arm/quants.c +++ b/ggml/src/ggml-cpu/arch/arm/quants.c @@ -263,13 +263,13 @@ void ggml_vec_dot_q2_0_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const voi const uint8x16_t raw16 = vcombine_u8(raw, raw); // First 16 elements: replicate bytes 0-3, shift, mask, subtract 1 - uint8x16_t bytes0 = vqtbl1q_u8(raw16, idx_lo); + uint8x16_t bytes0 = ggml_vqtbl1q_u8(raw16, idx_lo); int8x16_t qv0 = vsubq_s8( vreinterpretq_s8_u8(vandq_u8(vshlq_u8(bytes0, shifts), mask2)), one); // Second 16 elements: replicate bytes 4-7, shift, mask, subtract 1 - uint8x16_t bytes1 = vqtbl1q_u8(raw16, idx_hi); + uint8x16_t bytes1 = ggml_vqtbl1q_u8(raw16, idx_hi); int8x16_t qv1 = vsubq_s8( vreinterpretq_s8_u8(vandq_u8(vshlq_u8(bytes1, shifts), mask2)), one);