compiler/nir: Add new flags to lower pack/unpack split instructions

And add new rules to do this lowering in nir_opt_algebraic.py.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6309>

diff --git a/src/compiler/nir/nir.h b/src/compiler/nir/nir.h
index 79a6108be73e7f93777f1139baaff8235b93ad07..083bdf207f42b073bbc9bf13192c43fe3292fed2 100644 (file)
--- a/src/compiler/nir/nir.h
+++ b/src/compiler/nir/nir.h
@@ -3064,11 +3064,15 @@ typedef struct nir_shader_compiler_options {
    bool lower_pack_snorm_2x16;
    bool lower_pack_unorm_4x8;
    bool lower_pack_snorm_4x8;
+   bool lower_pack_64_2x32_split;
+   bool lower_pack_32_2x16_split;
    bool lower_unpack_half_2x16;
    bool lower_unpack_unorm_2x16;
    bool lower_unpack_snorm_2x16;
    bool lower_unpack_unorm_4x8;
    bool lower_unpack_snorm_4x8;
+   bool lower_unpack_64_2x32_split;
+   bool lower_unpack_32_2x16_split;
 
    bool lower_pack_split;
 

diff --git a/src/compiler/nir/nir_opt_algebraic.py b/src/compiler/nir/nir_opt_algebraic.py
index 1762591bd12761311259b97c776f235747e409a0..14e7f30d9760d6105e9981345f3163ca3d5ae4e3 100644 (file)
--- a/src/compiler/nir/nir_opt_algebraic.py
+++ b/src/compiler/nir/nir_opt_algebraic.py
@@ -1045,6 +1045,14 @@ optimizations.extend([
    (('ibfe', a,  0, 16), ('extract_i16', a, 0), '!options->lower_extract_word'),
    (('ibfe', a, 16, 16), ('extract_i16', a, 1), '!options->lower_extract_word'),
 
+   # Lower pack/unpack
+   (('pack_64_2x32_split', a, b), ('ior', ('u2u64', a), ('ishl', ('u2u64', b), 32)), 'options->lower_pack_64_2x32_split'),
+   (('pack_32_2x16_split', a, b), ('ior', ('u2u32', a), ('ishl', ('u2u32', b), 16)), 'options->lower_pack_32_2x16_split'),
+   (('unpack_64_2x32_split_x', a), ('u2u32', a), 'options->lower_unpack_64_2x32_split'),
+   (('unpack_64_2x32_split_y', a), ('u2u32', ('ushr', a, 32)), 'options->lower_unpack_64_2x32_split'),
+   (('unpack_32_2x16_split_x', a), ('u2u16', a), 'options->lower_unpack_32_2x16_split'),
+   (('unpack_32_2x16_split_y', a), ('u2u16', ('ushr', a, 16)), 'options->lower_unpack_32_2x16_split'),
+
    # Useless masking before unpacking
    (('unpack_half_2x16_split_x', ('iand', a, 0xffff)), ('unpack_half_2x16_split_x', a)),
    (('unpack_32_2x16_split_x', ('iand', a, 0xffff)), ('unpack_32_2x16_split_x', a)),