From a9460da19c5b19aa54eff979cae02b2c77c33b0d Mon Sep 17 00:00:00 2001 From: dzaima Date: Fri, 9 Aug 2024 05:17:04 +0300 Subject: [PATCH] more missing NEON comments --- src/builtins/scan.c | 2 +- src/builtins/transpose.c | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/src/builtins/scan.c b/src/builtins/scan.c index 5e582904..565db8f7 100644 --- a/src/builtins/scan.c +++ b/src/builtins/scan.c @@ -10,7 +10,7 @@ // < SWAR // =≤≥>- in terms of ≠<∨∧+ with adjustments // Arithmetic operand, rank 1: -// ⌈⌊ Scalar, SSE, AVX in log(vector width) steps +// ⌈⌊ Scalar, SSE, AVX in log(vector width) steps (SHOULD add NEON) // + Overflow-checked scalar or AVX2 // Ad-hoc boolean-valued handling for ≠∨ // SHOULD extend rank 1 special cases to cell bound 1 diff --git a/src/builtins/transpose.c b/src/builtins/transpose.c index 3dd2c86f..80a4ac58 100644 --- a/src/builtins/transpose.c +++ b/src/builtins/transpose.c @@ -12,6 +12,7 @@ // i8: 16×16; i16: 16×8; i32: 8×8; f64: 4×4 // COULD use half-width or smaller kernels to improve odd sizes // Scalar transpose or loop used for overhang of 1 +// SHOULD add NEON // Reorder Axes // If 𝕨 indicates the identity permutation, return 𝕩