diff options
-rw-r--r-- | src/intel/compiler/brw_compiler.c | 42 |
1 files changed, 11 insertions, 31 deletions
diff --git a/src/intel/compiler/brw_compiler.c b/src/intel/compiler/brw_compiler.c index 5d88a66f1fa..f1acf838887 100644 --- a/src/intel/compiler/brw_compiler.c +++ b/src/intel/compiler/brw_compiler.c @@ -69,12 +69,6 @@ static const struct nir_shader_compiler_options scalar_nir_options = { COMMON_SCALAR_OPTIONS, }; -static const struct nir_shader_compiler_options scalar_nir_options_gen11 = { - COMMON_OPTIONS, - COMMON_SCALAR_OPTIONS, - .lower_flrp32 = true, -}; - static const struct nir_shader_compiler_options vector_nir_options = { COMMON_OPTIONS, @@ -84,27 +78,6 @@ static const struct nir_shader_compiler_options vector_nir_options = { */ .fdot_replicates = true, - /* Prior to Gen6, there are no three source operations for SIMD4x2. */ - .lower_flrp32 = true, - - .lower_pack_snorm_2x16 = true, - .lower_pack_unorm_2x16 = true, - .lower_unpack_snorm_2x16 = true, - .lower_unpack_unorm_2x16 = true, - .lower_extract_byte = true, - .lower_extract_word = true, - .max_unroll_iterations = 32, -}; - -static const struct nir_shader_compiler_options vector_nir_options_gen6 = { - COMMON_OPTIONS, - - /* In the vec4 backend, our dpN instruction replicates its result to all the - * components of a vec4. We would like NIR to give us replicated fdot - * instructions because it can optimize better for us. - */ - .fdot_replicates = true, - .lower_pack_snorm_2x16 = true, .lower_pack_unorm_2x16 = true, .lower_unpack_snorm_2x16 = true, @@ -197,11 +170,18 @@ brw_compiler_create(void *mem_ctx, const struct gen_device_info *devinfo) struct nir_shader_compiler_options *nir_options = rzalloc(compiler, struct nir_shader_compiler_options); if (is_scalar) { - *nir_options = - devinfo->gen < 11 ? scalar_nir_options : scalar_nir_options_gen11; + *nir_options = scalar_nir_options; + + if (devinfo->gen >= 11) { + nir_options->lower_flrp32 = true; + } } else { - *nir_options = - devinfo->gen < 6 ? vector_nir_options : vector_nir_options_gen6; + *nir_options = vector_nir_options; + + if (devinfo->gen < 6) { + /* Prior to Gen6, there are no three source operations. */ + nir_options->lower_flrp32 = true; + } } nir_options->lower_int64_options = int64_options; nir_options->lower_doubles_options = fp64_options; |