[PATCH 0/2] MR69: vkd3d-shader/hlsl: Add sqrt() support for SM1.

[PATCH 1/2] vkd3d-shader/hlsl: Handle RSQ output for SM1.

[Nikolay Sivov]

From: Nikolay Sivov

---
 libs/vkd3d-shader/hlsl_sm1.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/libs/vkd3d-shader/hlsl_sm1.c b/libs/vkd3d-shader/hlsl_sm1.c
index ba22925e..71f7d5f6 100644
--- a/libs/vkd3d-shader/hlsl_sm1.c
+++ b/libs/vkd3d-shader/hlsl_sm1.c
@@ -632,13 +632,14 @@ static void write_sm1_expr(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *b
             break;

case HLSL_OP1_RCP:
+        case HLSL_OP1_RSQ:
             for (i = 0; i < instr->data_type->dimx; ++i)
             {
                 struct hlsl_reg src = arg1->reg, dst = instr->reg;

src.writemask = hlsl_combine_writemasks(src.writemask, 1u << i);
                 dst.writemask = hlsl_combine_writemasks(dst.writemask, 1u << i);
-                write_sm1_unary_op(ctx, buffer, D3DSIO_RCP, &dst, &src, 0);
+                write_sm1_unary_op(ctx, buffer, expr->op == HLSL_OP1_RCP ? D3DSIO_RCP : D3DSIO_RSQ, &dst, &src, 0);
             }
             break;

[PATCH 2/2] vkd3d-shader/hlsl: Implement sqrt() for SM1.

[Nikolay Sivov]

From: Nikolay Sivov

---
 libs/vkd3d-shader/hlsl_codegen.c | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/libs/vkd3d-shader/hlsl_codegen.c b/libs/vkd3d-shader/hlsl_codegen.c
index 9b644d1b..5163d913 100644
--- a/libs/vkd3d-shader/hlsl_codegen.c
+++ b/libs/vkd3d-shader/hlsl_codegen.c
@@ -1345,6 +1345,27 @@ static bool lower_division(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, voi
     return true;
 }

+/* Lower SQRT to RSQ + RCP. */
+static bool lower_sqrt(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context)
+{
+    struct hlsl_ir_expr *expr;
+    struct hlsl_ir_node *rsq;
+
+    if (instr->type != HLSL_IR_EXPR)
+        return false;
+    expr = hlsl_ir_expr(instr);
+    if (expr->op != HLSL_OP1_SQRT)
+        return false;
+
+    if (!(rsq = hlsl_new_unary_expr(ctx, HLSL_OP1_RSQ, expr->operands[0].node, instr->loc)))
+        return false;
+    list_add_before(&expr->node.entry, &rsq->entry);
+    expr->op = HLSL_OP1_RCP;
+    hlsl_src_remove(&expr->operands[0]);
+    hlsl_src_from_node(&expr->operands[0], rsq);
+    return true;
+}
+
 static bool lower_casts_to_bool(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context)
 {
     struct hlsl_type *type = instr->data_type, *arg_type;
@@ -2755,7 +2776,10 @@ int hlsl_emit_bytecode(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *entry
     while (progress);

if (profile->major_version < 4)
+    {
         transform_ir(ctx, lower_division, body, NULL);
+        transform_ir(ctx, lower_sqrt, body, NULL);
+    }

transform_ir(ctx, validate_static_object_references, body, NULL);

[PATCH 0/2] MR69: vkd3d-shader/hlsl: Add sqrt() support for SM1. - approved

[Zebediah Figura (＠zfigura)]

This merge request was approved by Zebediah Figura.

[PATCH 0/2] MR69: vkd3d-shader/hlsl: Add sqrt() support for SM1. - approved

[Henri Verbeet (＠hverbeet)]

This merge request was approved by Henri Verbeet.

[PATCH 0/2] MR69: vkd3d-shader/hlsl: Add sqrt() support for SM1. - approved

[Giovanni Mascellani (＠giomasce)]

This merge request was approved by Giovanni Mascellani.