2 commits

Author	SHA1	Message	Date
Konstantin Seurer	de32f9275f	treewide: add & use parent instr helpers ... We add a bunch of new helpers to avoid the need to touch >parent_instr, including the full set of: * nir_def_is_* * nir_def_as_*_or_null * nir_def_as_* [assumes the right instr type] * nir_src_is_* * nir_src_as_* * nir_scalar_is_* * nir_scalar_as_* Plus nir_def_instr() where there's no more suitable helper. Also an existing helper is renamed to unify all the names, while we're churning the tree: * nir_src_as_alu_instr -> nir_src_as_alu ..and then we port the tree to use the helpers as much as possible, using nir_def_instr() where that does not work. Acked-by: Marek Olšák <maraeo@gmail.com> --- To eliminate nir_def::parent_instr we need to churn the tree anyway, so I'm taking this opportunity to clean up a lot of NIR patterns. Co-authored-by: Konstantin Seurer <konstantin.seurer@gmail.com> Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38313>	2025-11-12 21:22:13 +00:00
Job Noorman	3908a228bd	nir: add opt_uub pass ... Add a pass that uses nir_unsigned_upper_bound to simplify some ALU operations: - iand src, mask: if mask is constant with N least significant bits set and uub(src) < 2^N, the iand does nothing and can be removed. - ult src, const: if uub(src) < cmp -> true - uge src, const: if uub(src) < cmp -> false - ilt src, const: if uub(src) >= 0 && cmp < 0 -> false - if uub(src) >= 0 && cmp >= 0 -> ult src, const - ige src, const: if uub(src) >= 0 && cmp < 0 -> true - if uub(src) >= 0 && cmp >= 0 -> uge src, const - umin src, const: if uub(src) <= const -> src - umax src, const: if uub(src) <= const -> const - imin src, const: if uub(src) >= 0 && const < 0 -> const - if uub(src) >= 0 && const >= 0 -> umin src, const - imax src, const: if uub(src) >= 0 && const < 0 -> src - if uub(src) >= 0 && const >= 0 -> umax src, const - imul src0, src1: if uub(srci) < UINT16_MAX -> umul_16x16 src0, src1 - imul src0, src1: if uub(srci) < UINT24_MAX -> umul24 src0, src1 - imul src0, src1: if uub(srci) < UINT23_MAX -> imul24 src0, src1 The imul optimization needs to be explicitly enabled using a pass option. This is useful since 1) most backends don't support umul_16x16, and 2) some passes (e.g., nir_opt_load_store_vectorize) need to analyze imuls so lowering them before running such a pass makes their job more difficult. Signed-off-by: Job Noorman <jnoorman@igalia.com> Reviewed-by: Timur Kristóf <timur.kristof@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37869>	2025-11-07 10:23:29 +00:00