nir: split lower_ffma into lower_ffma16/32/64

AMD wants different behavior for each bit size

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6756>

6 files changed, 21 insertions, 7 deletions

diff --git a/src/gallium/drivers/lima/lima_program.c b/src/gallium/drivers/lima/lima_program.c
index 87029d3140a..30a3f527181 100644
--- a/src/gallium/drivers/lima/lima_program.c
+++ b/src/gallium/drivers/lima/lima_program.c
@@ -42,7 +42,9 @@
 #include "ir/lima_ir.h"
 
 static const nir_shader_compiler_options vs_nir_options = {
-   .lower_ffma = true,
+   .lower_ffma16 = true,
+   .lower_ffma32 = true,
+   .lower_ffma64 = true,
    .lower_fpow = true,
    .lower_ffract = true,
    .lower_fdiv = true,
@@ -60,7 +62,9 @@ static const nir_shader_compiler_options vs_nir_options = {
 };
 
 static const nir_shader_compiler_options fs_nir_options = {
-   .lower_ffma = true,
+   .lower_ffma16 = true,
+   .lower_ffma32 = true,
+   .lower_ffma64 = true,
    .lower_fpow = true,
    .lower_fdiv = true,
    .lower_fmod = true,
diff --git a/src/gallium/drivers/llvmpipe/lp_screen.c b/src/gallium/drivers/llvmpipe/lp_screen.c
index 5f05a468e3f..218d80e5c48 100644
--- a/src/gallium/drivers/llvmpipe/lp_screen.c
+++ b/src/gallium/drivers/llvmpipe/lp_screen.c
@@ -549,7 +549,9 @@ static const struct nir_shader_compiler_options gallivm_nir_options = {
    .lower_bitfield_insert_to_shifts = true,
    .lower_bitfield_extract_to_shifts = true,
    .lower_sub = true,
-   .lower_ffma = true,
+   .lower_ffma16 = true,
+   .lower_ffma32 = true,
+   .lower_ffma64 = true,
    .lower_fmod = true,
    .lower_hadd = true,
    .lower_add_sat = true,
diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_from_nir.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_from_nir.cpp
index c5e54779ad7..64453edf19b 100644
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_from_nir.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_from_nir.cpp
@@ -3206,7 +3206,9 @@ nvir_nir_shader_compiler_options(int chipset)
 {
    nir_shader_compiler_options op = {};
    op.lower_fdiv = (chipset >= NVISA_GV100_CHIPSET);
-   op.lower_ffma = false;
+   op.lower_ffma16 = false;
+   op.lower_ffma32 = false;
+   op.lower_ffma64 = false;
    op.fuse_ffma16 = false; /* nir doesn't track mad vs fma */
    op.fuse_ffma32 = false; /* nir doesn't track mad vs fma */
    op.fuse_ffma64 = false; /* nir doesn't track mad vs fma */
diff --git a/src/gallium/drivers/radeonsi/si_get.c b/src/gallium/drivers/radeonsi/si_get.c
index 1f19fdd8817..51a839056e7 100644
--- a/src/gallium/drivers/radeonsi/si_get.c
+++ b/src/gallium/drivers/radeonsi/si_get.c
@@ -945,7 +945,9 @@ void si_init_screen_get_functions(struct si_screen *sscreen)
        * Keep FMA enabled on gfx10 to test it, which helps us validate correctness
        * for gfx10.3 on gfx10.
        */
-      .lower_ffma = sscreen->info.chip_class <= GFX9,
+      .lower_ffma16 = sscreen->info.chip_class <= GFX9,
+      .lower_ffma32 = sscreen->info.chip_class <= GFX9,
+      .lower_ffma64 = sscreen->info.chip_class <= GFX9,
       .fuse_ffma16 = sscreen->info.chip_class >= GFX10,
       .fuse_ffma32 = sscreen->info.chip_class >= GFX10,
       .fuse_ffma64 = sscreen->info.chip_class >= GFX10,
diff --git a/src/gallium/drivers/vc4/vc4_program.c b/src/gallium/drivers/vc4/vc4_program.c
index 4f2956bb735..8bffb1d3840 100644
--- a/src/gallium/drivers/vc4/vc4_program.c
+++ b/src/gallium/drivers/vc4/vc4_program.c
@@ -2179,7 +2179,9 @@ static const nir_shader_compiler_options nir_options = {
         .lower_extract_byte = true,
         .lower_extract_word = true,
         .lower_fdiv = true,
-        .lower_ffma = true,
+        .lower_ffma16 = true,
+        .lower_ffma32 = true,
+        .lower_ffma64 = true,
         .lower_flrp32 = true,
         .lower_fmod = true,
         .lower_fpow = true,
diff --git a/src/gallium/drivers/zink/zink_compiler.c b/src/gallium/drivers/zink/zink_compiler.c
index b9d1b666b14..8f0c16cb13b 100644
--- a/src/gallium/drivers/zink/zink_compiler.c
+++ b/src/gallium/drivers/zink/zink_compiler.c
@@ -126,7 +126,9 @@ lower_discard_if(nir_shader *shader)
 
 static const struct nir_shader_compiler_options nir_options = {
    .lower_all_io_to_temps = true,
-   .lower_ffma = true,
+   .lower_ffma16 = true,
+   .lower_ffma32 = true,
+   .lower_ffma64 = true,
    .lower_fdph = true,
    .lower_flrp32 = true,
    .lower_fpow = true,