diff options
author | Samuel Iglesias Gonsálvez <[email protected]> | 2017-03-08 09:27:49 +0100 |
---|---|---|
committer | Francisco Jerez <[email protected]> | 2017-04-14 14:56:08 -0700 |
commit | 50a5217637636f066feabefd7fe46d0ff7778a64 (patch) | |
tree | 1cb8682f240a83ae14f4c4ecb8621f327689362f /src | |
parent | cfaf14a12607a8e9fd3d86a0c0219c428401f68f (diff) |
i965/vec4: split d2x conversion and data gathering from one opcode to two explicit ones
When doing a 64-bit to a smaller data type size conversion, the destination should
be aligned to 64-bits. Because of that, we need to gather the data after the
actual conversion.
Until now, these two operations were done by VEC4_OPCODE_FROM_DOUBLE but
now we split them explicitely in two different instructions:
VEC4_OPCODE_FROM_DOUBLE just do the conversion and
VEC4_OPCODE_PICK_LOW_32BIT will gather the data.
Signed-off-by: Samuel Iglesias Gonsálvez <[email protected]>
Reviewed-by: Francisco Jerez <[email protected]>
Diffstat (limited to 'src')
-rw-r--r-- | src/intel/compiler/brw_vec4_generator.cpp | 8 | ||||
-rw-r--r-- | src/intel/compiler/brw_vec4_nir.cpp | 1 |
2 files changed, 1 insertions, 8 deletions
diff --git a/src/intel/compiler/brw_vec4_generator.cpp b/src/intel/compiler/brw_vec4_generator.cpp index e1a12ba5ffe..65f3a9a9f00 100644 --- a/src/intel/compiler/brw_vec4_generator.cpp +++ b/src/intel/compiler/brw_vec4_generator.cpp @@ -1961,14 +1961,6 @@ generate_code(struct brw_codegen *p, src[0].width = BRW_WIDTH_4; brw_MOV(p, spread_dst, src[0]); - /* As we have set horizontal stride 1 instead of 2 in IVB/BYT, we - * need to fix it here to have the expected value. - */ - if (devinfo->gen == 7 && !devinfo->is_haswell) - spread_dst = stride(dst, 8, 4, 2); - - brw_MOV(p, dst, spread_dst); - brw_set_default_access_mode(p, BRW_ALIGN_16); break; } diff --git a/src/intel/compiler/brw_vec4_nir.cpp b/src/intel/compiler/brw_vec4_nir.cpp index 6dd5789225d..64371a16de5 100644 --- a/src/intel/compiler/brw_vec4_nir.cpp +++ b/src/intel/compiler/brw_vec4_nir.cpp @@ -1191,6 +1191,7 @@ vec4_visitor::emit_conversion_from_double(dst_reg dst, src_reg src, emit(VEC4_OPCODE_FROM_DOUBLE, temp2, src_reg(temp)) ->size_written = 2 * REG_SIZE; + emit(VEC4_OPCODE_PICK_LOW_32BIT, temp2, src_reg(retype(temp2, BRW_REGISTER_TYPE_DF))); vec4_instruction *inst = emit(MOV(dst, src_reg(temp2))); inst->saturate = saturate; } |