diff options
| author | Jose Maria Casanova Crespo <[email protected]> | 2017-07-01 07:58:26 +0200 |
|---|---|---|
| committer | Jose Maria Casanova Crespo <[email protected]> | 2017-12-06 08:57:18 +0100 |
| commit | 1f440d00d2b6ae6f74fb850ea5acec1f1b5efa58 (patch) | |
| tree | 46460eeeeeeefe82036e98926071b06ba92e441e /src/compiler/glsl | |
| parent | 2af63683bc61e1efb8f634697770d314ef07c882 (diff) | |
nir: Handle fp16 rounding modes at nir_type_conversion_op
nir_type_conversion enables new operations to handle rounding modes to
convert to fp16 values. Two new opcodes are enabled nir_op_f2f16_rtne
and nir_op_f2f16_rtz.
The undefined behaviour doesn't has any effect and uses the original
nir_op_f2f16 operation.
v2: Indentation fixed (Jason Ekstrand)
v3: Use explicit case for undefined rounding and assert if
rounding mode is used for non 16-bit float conversions
(Jason Ekstrand)
Reviewed-by: Jason Ekstrand <[email protected]>
Diffstat (limited to 'src/compiler/glsl')
| -rw-r--r-- | src/compiler/glsl/glsl_to_nir.cpp | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/src/compiler/glsl/glsl_to_nir.cpp b/src/compiler/glsl/glsl_to_nir.cpp index 67f15977ee5..0493410aebe 100644 --- a/src/compiler/glsl/glsl_to_nir.cpp +++ b/src/compiler/glsl/glsl_to_nir.cpp @@ -1575,7 +1575,8 @@ nir_visitor::visit(ir_expression *ir) case ir_unop_u642i64: { nir_alu_type src_type = nir_get_nir_type_for_glsl_base_type(types[0]); nir_alu_type dst_type = nir_get_nir_type_for_glsl_base_type(out_type); - result = nir_build_alu(&b, nir_type_conversion_op(src_type, dst_type), + result = nir_build_alu(&b, nir_type_conversion_op(src_type, dst_type, + nir_rounding_mode_undef), srcs[0], NULL, NULL, NULL); /* b2i and b2f don't have fixed bit-size versions so the builder will * just assume 32 and we have to fix it up here. |