Mesa (main): v3dv: don't lower uadd_carry and usub_borrow

[GitLab Mirror]

Module: Mesa
Branch: main
Commit: 152fc4fd289b6f4c5df955fed7cdec5139b8f316
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=152fc4fd289b6f4c5df955fed7cdec5139b8f316

Author: Iago Toral Quiroga <itoral at igalia.com>
Date:   Wed Jul  6 11:29:41 2022 +0200

v3dv: don't lower uadd_carry and usub_borrow

We can produce slightly better code for these in the backend, so
do that. For this we need to:

1. Fix our implementation of uadd_carry (which wasn't used) to return
   an integer instead of a boolean value.
2. Add an implementation of usub_borrow.

Notice these are only used in Vulkan. In GL these instructions are
always unconditionally lowered by the state tracker in GLSL IR so
we never get to see them in the backend.

Shader-db stats from a collection of Vulkan samples:

total instructions in shared programs: 122351 -> 122345 (<.01%)
instructions in affected programs: 196 -> 190 (-3.06%)
helped: 2
HURT: 0

total uniforms in shared programs: 18670 -> 18672 (0.01%)
uniforms in affected programs: 59 -> 61 (3.39%)
helped: 0
HURT: 2

total max-temps in shared programs: 13145 -> 13147 (0.02%)
max-temps in affected programs: 27 -> 29 (7.41%)
helped: 0
HURT: 2

total inst-and-stalls in shared programs: 123052 -> 123046 (<.01%)
inst-and-stalls in affected programs: 197 -> 191 (-3.05%)
helped: 2
HURT: 0

Reviewed-by: Alejandro Piñeiro <apinheiro at igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17372>

---

 src/broadcom/compiler/nir_to_vir.c  | 20 +++++++++++++++++++-
 src/broadcom/vulkan/v3dv_pipeline.c |  5 -----
 2 files changed, 19 insertions(+), 6 deletions(-)

diff --git a/src/broadcom/compiler/nir_to_vir.c b/src/broadcom/compiler/nir_to_vir.c
index 2bdbccad308..904f0a562f3 100644
--- a/src/broadcom/compiler/nir_to_vir.c
+++ b/src/broadcom/compiler/nir_to_vir.c
@@ -1263,6 +1263,18 @@ ntq_emit_cond_to_bool(struct v3d_compile *c, enum v3d_qpu_cond cond)
         return result;
 }
 
+static struct qreg
+ntq_emit_cond_to_int(struct v3d_compile *c, enum v3d_qpu_cond cond)
+{
+        struct qreg result =
+                vir_MOV(c, vir_SEL(c, cond,
+                                   vir_uniform_ui(c, 1),
+                                   vir_uniform_ui(c, 0)));
+        c->flags_temp = result.index;
+        c->flags_cond = cond;
+        return result;
+}
+
 static struct qreg
 f2f16_rtz(struct v3d_compile *c, struct qreg f32)
 {
@@ -1713,7 +1725,13 @@ ntq_emit_alu(struct v3d_compile *c, nir_alu_instr *instr)
         case nir_op_uadd_carry:
                 vir_set_pf(c, vir_ADD_dest(c, vir_nop_reg(), src[0], src[1]),
                            V3D_QPU_PF_PUSHC);
-                result = ntq_emit_cond_to_bool(c, V3D_QPU_COND_IFA);
+                result = ntq_emit_cond_to_int(c, V3D_QPU_COND_IFA);
+                break;
+
+        case nir_op_usub_borrow:
+                vir_set_pf(c, vir_SUB_dest(c, vir_nop_reg(), src[0], src[1]),
+                           V3D_QPU_PF_PUSHC);
+                result = ntq_emit_cond_to_int(c, V3D_QPU_COND_IFA);
                 break;
 
         case nir_op_pack_half_2x16_split:
diff --git a/src/broadcom/vulkan/v3dv_pipeline.c b/src/broadcom/vulkan/v3dv_pipeline.c
index a441c74ac07..82058c270e1 100644
--- a/src/broadcom/vulkan/v3dv_pipeline.c
+++ b/src/broadcom/vulkan/v3dv_pipeline.c
@@ -214,11 +214,6 @@ const nir_shader_compiler_options v3dv_nir_options = {
    .lower_pack_32_2x16 = true,
    .lower_pack_32_2x16_split = true,
    .lower_unpack_32_2x16_split = true,
-   /* FIXME: see if we can avoid the uadd_carry and usub_borrow lowering and
-    * get the tests to pass since it might produce slightly better code.
-    */
-   .lower_uadd_carry = true,
-   .lower_usub_borrow = true,
    /* FIXME: check if we can use multop + umul24 to implement mul2x32_64
     * without lowering.
     */