Mesa (main): zink: move bo load offset adjustment to compiler passes

[GitLab Mirror]

Module: Mesa
Branch: main
Commit: ee22cd619fb164fb6b0c53d6915e6901b52b97fa
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=ee22cd619fb164fb6b0c53d6915e6901b52b97fa

Author: Mike Blumenkrantz <michael.blumenkrantz at gmail.com>
Date:   Tue Sep 21 17:11:05 2021 -0400

zink: move bo load offset adjustment to compiler passes

Reviewed-by: Dave Airlie <airlied at redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13484>

---

 .../drivers/zink/nir_to_spirv/nir_to_spirv.c       | 14 +++-------
 src/gallium/drivers/zink/zink_compiler.c           | 31 ++++++++++++++++++++++
 2 files changed, 35 insertions(+), 10 deletions(-)

diff --git a/src/gallium/drivers/zink/nir_to_spirv/nir_to_spirv.c b/src/gallium/drivers/zink/nir_to_spirv/nir_to_spirv.c
index 6a8c8eca156..e3b8a27a42b 100644
--- a/src/gallium/drivers/zink/nir_to_spirv/nir_to_spirv.c
+++ b/src/gallium/drivers/zink/nir_to_spirv/nir_to_spirv.c
@@ -1941,8 +1941,6 @@ emit_load_bo(struct ntv_context *ctx, nir_intrinsic_instr *intr)
 
    /* destination type for the load */
    SpvId type = get_dest_uvec_type(ctx, &intr->dest);
-   /* an id of an array member in bytes */
-   SpvId uint_size = emit_uint_const(ctx, 32, MIN2(bit_size, 32) / 8);
 
    /* we grab a single array member at a time, so it's a pointer to a uint */
    SpvId pointer_type = spirv_builder_type_pointer(&ctx->builder,
@@ -1955,18 +1953,13 @@ emit_load_bo(struct ntv_context *ctx, nir_intrinsic_instr *intr)
     *    uint base[array_size];
     * };
     *
-    * where 'array_size' is set as though every member of the ubo takes up a vec4,
-    * even if it's only a vec2 or a float.
-    *
     * first, access 'base'
     */
    SpvId member = emit_uint_const(ctx, 32, 0);
-   /* this is the offset (in bytes) that we're accessing:
+   /* this is the array member we're accessing:
     * it may be a const value or it may be dynamic in the shader
     */
    SpvId offset = get_src(ctx, &intr->src[1]);
-   /* calculate the byte offset in the array */
-   SpvId vec_offset = emit_binop(ctx, SpvOpUDiv, uint_type, offset, uint_size);
    /* OpAccessChain takes an array of indices that drill into a hierarchy based on the type:
     * index 0 is accessing 'base'
     * index 1 is accessing 'base[index 1]'
@@ -1976,7 +1969,7 @@ emit_load_bo(struct ntv_context *ctx, nir_intrinsic_instr *intr)
     * (composite|vector)_extract both take literals
     */
    for (unsigned i = 0; i < num_components; i++) {
-      SpvId indices[2] = { member, vec_offset };
+      SpvId indices[2] = { member, offset };
       SpvId ptr = spirv_builder_emit_access_chain(&ctx->builder, pointer_type,
                                                   bo, indices,
                                                   ARRAY_SIZE(indices));
@@ -1985,8 +1978,9 @@ emit_load_bo(struct ntv_context *ctx, nir_intrinsic_instr *intr)
          constituents[i] = emit_atomic(ctx, SpvOpAtomicLoad, uint_type, ptr, 0, 0);
       else
          constituents[i] = spirv_builder_emit_load(&ctx->builder, uint_type, ptr);
+
       /* increment to the next member index for the next load */
-      vec_offset = emit_binop(ctx, SpvOpIAdd, uint_type, vec_offset, one);
+      offset = emit_binop(ctx, SpvOpIAdd, uint_type, offset, one);
    }
 
    /* if we're loading a 64bit value, we have to reassemble all the u32 values we've loaded into u64 values
diff --git a/src/gallium/drivers/zink/zink_compiler.c b/src/gallium/drivers/zink/zink_compiler.c
index 5bdfdc9cdae..59c88aae5ed 100644
--- a/src/gallium/drivers/zink/zink_compiler.c
+++ b/src/gallium/drivers/zink/zink_compiler.c
@@ -637,6 +637,32 @@ decompose_attribs(nir_shader *nir, uint32_t decomposed_attrs, uint32_t decompose
    return true;
 }
 
+static bool
+rewrite_bo_access_instr(nir_builder *b, nir_instr *instr, void *data)
+{
+   if (instr->type != nir_instr_type_intrinsic)
+      return false;
+   nir_intrinsic_instr *intr = nir_instr_as_intrinsic(instr);
+   switch (intr->intrinsic) {
+   case nir_intrinsic_load_ssbo:
+   case nir_intrinsic_load_ubo:
+   case nir_intrinsic_load_ubo_vec4:
+      b->cursor = nir_before_instr(instr);
+      nir_instr_rewrite_src_ssa(instr, &intr->src[1], nir_udiv_imm(b, intr->src[1].ssa, MIN2(nir_dest_bit_size(intr->dest), 32) / 8));
+      return true;
+   case nir_intrinsic_store_ssbo:
+   default:
+      break;
+   }
+   return false;
+}
+
+static bool
+rewrite_bo_access(nir_shader *shader)
+{
+   return nir_shader_instructions_pass(shader, rewrite_bo_access_instr, nir_metadata_dominance, NULL);
+}
+
 static void
 assign_producer_var_io(gl_shader_stage stage, nir_variable *var, unsigned *reserved, unsigned char *slot_map)
 {
@@ -872,6 +898,8 @@ zink_shader_compile(struct zink_screen *screen, struct zink_shader *zs, nir_shad
       default: break;
       }
    }
+   if (screen->driconf.inline_uniforms)
+      NIR_PASS_V(nir, rewrite_bo_access);
    if (inlined_uniforms) {
       optimize_nir(nir);
 
@@ -1384,6 +1412,9 @@ zink_shader_create(struct zink_screen *screen, struct nir_shader *nir,
          nir->info.fs.color_is_dual_source ? 1 : 8);
    NIR_PASS_V(nir, lower_64bit_vertex_attribs);
    NIR_PASS_V(nir, unbreak_bos);
+   /* run in compile if there could be inlined uniforms */
+   if (!screen->driconf.inline_uniforms)
+      NIR_PASS_V(nir, rewrite_bo_access);
 
    if (zink_debug & ZINK_DEBUG_NIR) {
       fprintf(stderr, "NIR shader:\n---8<---\n");