diff options
| author | Francisco Jerez <[email protected]> | 2019-12-27 16:38:26 -0800 |
|---|---|---|
| committer | Francisco Jerez <[email protected]> | 2020-01-10 11:02:30 -0800 |
| commit | c20dc9b8363b5b497d4f6af5144590e8c2d940c0 (patch) | |
| tree | fc1729fa9ef1dea7758efeaa7c04ce3a6be1b455 /src/intel/compiler/brw_fs_scoreboard.cpp | |
| parent | 591f146fd2c7b265cd9e759c242e6d6437ea6578 (diff) | |
intel/fs: Make implied_mrf_writes() an fs_inst method.
This will be convenient in a later commit enabling SIMD32 fragment
shaders, and happens to fix the calculation for MATH instructions
which is currently inaccurate for SIMD-lowered instructions on Gen4-5
platforms (all of them on Gen4 in SIMD16 mode), since it was based on
the shader's dispatch width rather than on the actual execution size
of the instruction.
This causes some shader-db noise on Gen4 due to the more compact
register allocation interacting with the SEND dependency workarounds,
but otherwise no major changes.
Reviewed-by: Kenneth Graunke <[email protected]>
Diffstat (limited to 'src/intel/compiler/brw_fs_scoreboard.cpp')
| -rw-r--r-- | src/intel/compiler/brw_fs_scoreboard.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/src/intel/compiler/brw_fs_scoreboard.cpp b/src/intel/compiler/brw_fs_scoreboard.cpp index 67e07e7f26b..35586f2a107 100644 --- a/src/intel/compiler/brw_fs_scoreboard.cpp +++ b/src/intel/compiler/brw_fs_scoreboard.cpp @@ -896,7 +896,7 @@ namespace { } if (is_send(inst) && inst->base_mrf != -1) { - for (int j = 0; j < shader->implied_mrf_writes(inst); j++) + for (unsigned j = 0; j < inst->implied_mrf_writes(); j++) add_dependency(ids, deps[ip], dependency_for_write(inst, sb.get(brw_uvec_mrf(8, inst->base_mrf + j, 0)))); } |