diff options
| author | Kenneth Graunke <[email protected]> | 2019-06-03 11:43:38 -0700 |
|---|---|---|
| committer | Kenneth Graunke <[email protected]> | 2019-06-05 16:45:12 -0700 |
| commit | d13059f4d50d370eec76495dd2b51fdf5c6e1aee (patch) | |
| tree | 04845e57328a1c81631ca19bc1153886ff2bb639 /src/gallium/drivers | |
| parent | fa56a3795f9f4b15013f2163193c54611fd2b8e0 (diff) | |
nouveau: Use nir_lower_doubles instead of lower_fmod64 on nvc0.
We currently have two duplicate mechanisms for lowering fmod@64.
One is a nir_opt_algebraic rule keyed off of options->lower_fmod64,
and the other is nir_lower_doubles, which offers a full gamut of
fp64 lowering. The latter works slightly better in some corner cases,
so I'm trying to eliminate lower_fmod64 and drop the redundancy.
Reviewed-by: Marek Olšák <[email protected]>
Diffstat (limited to 'src/gallium/drivers')
| -rw-r--r-- | src/gallium/drivers/nouveau/nvc0/nvc0_screen.c | 3 |
1 files changed, 1 insertions, 2 deletions
diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c b/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c index 998116fc251..09f8c4135e5 100644 --- a/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c +++ b/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c @@ -906,7 +906,6 @@ static const nir_shader_compiler_options nir_options = { .lower_fsat = false, .lower_fsqrt = false, // TODO: only before gm200 .lower_fmod32 = true, - .lower_fmod64 = true, .lower_bitfield_extract = false, .lower_bitfield_extract_to_shifts = false, .lower_bitfield_insert = false, @@ -955,7 +954,7 @@ static const nir_shader_compiler_options nir_options = { .lower_mul_2x32_64 = true, // TODO .max_unroll_iterations = 32, .lower_int64_options = nir_lower_divmod64, // TODO - .lower_doubles_options = 0, // TODO + .lower_doubles_options = nir_lower_dmod, // TODO }; static const void * |