nir: Combine lower_fmod16/32 back into a single lower_fmod.

We originally had a single lower_fmod option.  In commit 2ab2d2e5, Sam
split 32 and 64-bit lowering into separate flags, with the rationale
that some drivers might want different options there.  This left 16-bit
unhandled, so Iago added a lower_fmod16 option in commit ca31df6f.

Now that lower_fmod64 is gone (in favor of nir_lower_doubles and
nir_lower_dmod), we re-combine lower_fmod16 and lower_fmod32 into a
single lower_fmod flag again.  I'm not aware of any hardware which
need lowering for one bitsize and not the other.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>

diff --git a/src/compiler/nir/nir.h b/src/compiler/nir/nir.h
index 794b18334369fcf5f04286d83fcaae1dad970dd3..e179481b8ea05c53d35fc5a76667390056cadee2 100644 (file)
--- a/src/compiler/nir/nir.h
+++ b/src/compiler/nir/nir.h
@@ -2248,8 +2248,7 @@ typedef struct nir_shader_compiler_options {
    bool lower_fpow;
    bool lower_fsat;
    bool lower_fsqrt;
-   bool lower_fmod16;
-   bool lower_fmod32;
+   bool lower_fmod;
    /** Lowers ibitfield_extract/ubitfield_extract to ibfe/ubfe. */
    bool lower_bitfield_extract;
    /** Lowers ibitfield_extract/ubitfield_extract to bfm, compares, shifts. */

diff --git a/src/compiler/nir/nir_opt_algebraic.py b/src/compiler/nir/nir_opt_algebraic.py
index c0894ce42c9eebe99e9358f9c183d9f4eb06f804..b1c9b071464e38c17129147e1786ecc15ef6d957 100644 (file)
--- a/src/compiler/nir/nir_opt_algebraic.py
+++ b/src/compiler/nir/nir_opt_algebraic.py
@@ -771,9 +771,9 @@ optimizations.extend([
    (('bcsel', ('ine', a, -1), ('ifind_msb', a), -1), ('ifind_msb', a)),
 
    # Misc. lowering
-   (('fmod@16', a, b), ('fsub', a, ('fmul', b, ('ffloor', ('fdiv', a, b)))), 'options->lower_fmod16'),
-   (('fmod@32', a, b), ('fsub', a, ('fmul', b, ('ffloor', ('fdiv', a, b)))), 'options->lower_fmod32'),
-   (('frem', a, b), ('fsub', a, ('fmul', b, ('ftrunc', ('fdiv', a, b)))), 'options->lower_fmod32'),
+   (('fmod@16', a, b), ('fsub', a, ('fmul', b, ('ffloor', ('fdiv', a, b)))), 'options->lower_fmod'),
+   (('fmod@32', a, b), ('fsub', a, ('fmul', b, ('ffloor', ('fdiv', a, b)))), 'options->lower_fmod'),
+   (('frem', a, b), ('fsub', a, ('fmul', b, ('ftrunc', ('fdiv', a, b)))), 'options->lower_fmod'),
    (('uadd_carry@32', a, b), ('b2i', ('ult', ('iadd', a, b), a)), 'options->lower_uadd_carry'),
    (('usub_borrow@32', a, b), ('b2i', ('ult', a, b)), 'options->lower_usub_borrow'),
 

diff --git a/src/freedreno/ir3/ir3_nir.c b/src/freedreno/ir3/ir3_nir.c
index 888ce5a70f6b080802f41f387b336fb176e74e34..23dabee1fb020068cee5724d241d15825d74f152 100644 (file)
--- a/src/freedreno/ir3/ir3_nir.c
+++ b/src/freedreno/ir3/ir3_nir.c
@@ -40,7 +40,7 @@ static const nir_shader_compiler_options options = {
                .lower_flrp32 = true,
                .lower_flrp64 = true,
                .lower_ffract = true,
-               .lower_fmod32 = true,
+               .lower_fmod = true,
                .lower_fdiv = true,
                .lower_isign = true,
                .lower_ldexp = true,
@@ -65,7 +65,7 @@ static const nir_shader_compiler_options options_a6xx = {
                .lower_flrp32 = true,
                .lower_flrp64 = true,
                .lower_ffract = true,
-               .lower_fmod32 = true,
+               .lower_fmod = true,
                .lower_fdiv = true,
                .lower_isign = true,
                .lower_ldexp = true,

diff --git a/src/gallium/drivers/freedreno/a2xx/ir2_nir.c b/src/gallium/drivers/freedreno/a2xx/ir2_nir.c
index 5cfb831b6e26ebb1bd4bb08892a8d073c07abb8f..36f3a679ff6df3596f270f9a34b593bf0eac518a 100644 (file)
--- a/src/gallium/drivers/freedreno/a2xx/ir2_nir.c
+++ b/src/gallium/drivers/freedreno/a2xx/ir2_nir.c
@@ -32,7 +32,7 @@
 static const nir_shader_compiler_options options = {
        .lower_fpow = true,
        .lower_flrp32 = true,
-       .lower_fmod32 = true,
+       .lower_fmod = true,
        .lower_fdiv = true,
        .lower_fceil = true,
        .fuse_ffma = true,

diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c b/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c
index 09f8c4135e55bf556a2c86b8d73fb6700e4e296d..527d5a8108a381476aa6081cfcf5110124c3738c 100644 (file)
--- a/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c
+++ b/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c
@@ -905,7 +905,7 @@ static const nir_shader_compiler_options nir_options = {
    .lower_fpow = false,
    .lower_fsat = false,
    .lower_fsqrt = false, // TODO: only before gm200
-   .lower_fmod32 = true,
+   .lower_fmod = true,
    .lower_bitfield_extract = false,
    .lower_bitfield_extract_to_shifts = false,
    .lower_bitfield_insert = false,

diff --git a/src/gallium/drivers/panfrost/midgard/midgard_compile.h b/src/gallium/drivers/panfrost/midgard/midgard_compile.h
index ba9e74adef394f5b1deb9482555d0bd4fff24200..2ee0dfbb1aee8c40501bb1bb52c9964209a31789 100644 (file)
--- a/src/gallium/drivers/panfrost/midgard/midgard_compile.h
+++ b/src/gallium/drivers/panfrost/midgard/midgard_compile.h
@@ -91,7 +91,7 @@ static const nir_shader_compiler_options midgard_nir_options = {
         .lower_flrp32 = true,
         .lower_flrp64 = true,
         .lower_ffract = true,
-        .lower_fmod32 = true,
+        .lower_fmod = true,
         .lower_fdiv = true,
         .lower_idiv = true,
         .lower_isign = true,

diff --git a/src/intel/compiler/brw_compiler.c b/src/intel/compiler/brw_compiler.c
index 9da549f74483681028b75cabad9968b86deb0fa3..550b2e7d7566ba9739d4a5bcd12854ce2aef5ca3 100644 (file)
--- a/src/intel/compiler/brw_compiler.c
+++ b/src/intel/compiler/brw_compiler.c
@@ -34,8 +34,7 @@
    .lower_fdiv = true,                                                        \
    .lower_scmp = true,                                                        \
    .lower_flrp16 = true,                                                      \
-   .lower_fmod16 = true,                                                      \
-   .lower_fmod32 = true,                                                      \
+   .lower_fmod = true,                                                        \
    .lower_bitfield_extract = true,                                            \
    .lower_bitfield_insert = true,                                             \
    .lower_uadd_carry = true,                                                  \