diff --git a/libs/vkd3d-shader/hlsl_codegen.c b/libs/vkd3d-shader/hlsl_codegen.c
index ef0a8a802..0461af509 100644
--- a/libs/vkd3d-shader/hlsl_codegen.c
+++ b/libs/vkd3d-shader/hlsl_codegen.c
@@ -3916,6 +3916,233 @@ static bool remove_trivial_conditional_branches(struct hlsl_ctx *ctx, struct hls
     return true;
 }
 
+static bool is_conditional_block_simple(const struct hlsl_block *cond_block)
+{
+    static const unsigned int max_cost = 10;
+    struct hlsl_ir_node *instr;
+    unsigned int cost = 0;
+
+    LIST_FOR_EACH_ENTRY(instr, &cond_block->instrs, struct hlsl_ir_node, entry)
+    {
+        switch (instr->type)
+        {
+            case HLSL_IR_CONSTANT:
+            case HLSL_IR_STRING_CONSTANT:
+            case HLSL_IR_SWIZZLE:
+                break;
+
+            case HLSL_IR_EXPR:
+                ++cost;
+                break;
+
+            case HLSL_IR_JUMP:
+                return false;
+
+            case HLSL_IR_STORE:
+                if (hlsl_ir_store(instr)->lhs.var->is_tgsm)
+                    return false;
+                ++cost;
+                break;
+
+            case HLSL_IR_LOAD:
+                if (hlsl_ir_load(instr)->src.var->is_tgsm)
+                    return false;
+                break;
+
+            default:
+                return false;
+        }
+
+        if (cost > max_cost)
+            return false;
+    }
+
+    return true;
+}
+
+static bool can_flatten_conditional_block(struct hlsl_ctx *ctx, const struct hlsl_block *cond_block)
+{
+    struct hlsl_ir_node *instr;
+
+    LIST_FOR_EACH_ENTRY(instr, &cond_block->instrs, struct hlsl_ir_node, entry)
+    {
+        switch (instr->type)
+        {
+            case HLSL_IR_CALL:
+            case HLSL_IR_RESOURCE_STORE:
+            case HLSL_IR_INTERLOCKED:
+            case HLSL_IR_SYNC:
+                goto fail;
+
+            case HLSL_IR_JUMP:
+            {
+                struct hlsl_ir_jump *jump = hlsl_ir_jump(instr);
+
+                if (jump->type != HLSL_IR_JUMP_DISCARD_NZ && jump->type != HLSL_IR_JUMP_DISCARD_NEG)
+                {
+                    hlsl_fixme(ctx, &instr->loc, "Flattening conditional blocks with non-discard jump instructions.");
+                    return false;
+                }
+
+                hlsl_fixme(ctx, &instr->loc, "Flattening conditional blocks with discard instructions.");
+                return false;
+            }
+
+            case HLSL_IR_STORE:
+                if (hlsl_ir_store(instr)->lhs.var->is_tgsm)
+                    goto fail;
+                break;
+
+            case HLSL_IR_IF:
+            {
+                struct hlsl_ir_if *iff = hlsl_ir_if(instr);
+
+                if (!can_flatten_conditional_block(ctx, &iff->then_block)
+                        || !can_flatten_conditional_block(ctx, &iff->else_block))
+                    return false;
+                break;
+            }
+
+            case HLSL_IR_LOOP:
+            {
+                struct hlsl_ir_loop *loop = hlsl_ir_loop(instr);
+
+                if (!can_flatten_conditional_block(ctx, &loop->iter)
+                        || !can_flatten_conditional_block(ctx, &loop->body))
+                    return false;
+                break;
+            }
+
+            case HLSL_IR_SWITCH:
+            {
+                struct hlsl_ir_switch *s = hlsl_ir_switch(instr);
+                struct hlsl_ir_switch_case *c;
+
+                LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry)
+                {
+                    if (!can_flatten_conditional_block(ctx, &c->body))
+                        return false;
+                }
+                break;
+            }
+
+            case HLSL_IR_CONSTANT:
+            case HLSL_IR_EXPR:
+            case HLSL_IR_INDEX:
+            case HLSL_IR_LOAD:
+            case HLSL_IR_RESOURCE_LOAD:
+            case HLSL_IR_STRING_CONSTANT:
+            case HLSL_IR_SWIZZLE:
+            case HLSL_IR_COMPILE:
+            case HLSL_IR_SAMPLER_STATE:
+            case HLSL_IR_STATEBLOCK_CONSTANT:
+                break;
+        }
+    }
+
+    return true;
+
+fail:
+    hlsl_error(ctx, &instr->loc, VKD3D_SHADER_ERROR_HLSL_CANNOT_FLATTEN,
+            "Conditional branches with side effects cannot be flattened.");
+    return false;
+}
+
+static bool lower_conditional_block_stores(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr,
+        struct hlsl_ir_node *cond, bool is_then)
+{
+    struct hlsl_ir_node *load, *new_val;
+    struct hlsl_ir_store *store;
+    struct hlsl_type *rhs_type;
+    struct hlsl_block block;
+
+    if (instr->type != HLSL_IR_STORE)
+        return false;
+    store = hlsl_ir_store(instr);
+    rhs_type = store->rhs.node->data_type;
+
+    VKD3D_ASSERT(rhs_type->class <= HLSL_CLASS_VECTOR);
+    VKD3D_ASSERT(cond->data_type->e.numeric.dimx == 1);
+
+    hlsl_block_init(&block);
+
+    load = hlsl_block_add_load_index(ctx, &block, &store->lhs, NULL, &store->node.loc);
+
+    if (store->writemask && !hlsl_types_are_equal(rhs_type, load->data_type))
+        load = hlsl_block_add_swizzle(ctx, &block, hlsl_swizzle_from_writemask(store->writemask),
+                rhs_type->e.numeric.dimx, load, &store->node.loc);
+
+    if (rhs_type->e.numeric.dimx != 1)
+        cond = hlsl_block_add_swizzle(ctx, &block, HLSL_SWIZZLE(X, X, X, X),
+                rhs_type->e.numeric.dimx, cond, &store->node.loc);
+
+    if (is_then)
+        new_val = hlsl_add_conditional(ctx, &block, cond, store->rhs.node, load);
+    else
+        new_val = hlsl_add_conditional(ctx, &block, cond, load, store->rhs.node);
+
+    list_move_before(&store->node.entry, &block.instrs);
+    hlsl_src_remove(&store->rhs);
+    hlsl_src_from_node(&store->rhs, new_val);
+    return true;
+}
+
+struct flatten_conditional_block_ctx
+{
+    struct hlsl_ir_node *cond;
+    bool is_then;
+};
+
+static bool lower_conditional_block_instrs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context)
+{
+    struct flatten_conditional_block_ctx *flatten_ctx = context;
+
+    return lower_conditional_block_stores(ctx, instr, flatten_ctx->cond, flatten_ctx->is_then);
+}
+
+static bool flatten_conditional_branches(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context)
+{
+    struct flatten_conditional_block_ctx flatten_ctx;
+    struct hlsl_ir_if *iff;
+    bool force_flatten;
+
+    if (instr->type != HLSL_IR_IF)
+        return false;
+    iff = hlsl_ir_if(instr);
+
+    if (iff->flatten_type == HLSL_IF_FORCE_BRANCH)
+        return false;
+
+    force_flatten = iff->flatten_type == HLSL_IF_FORCE_FLATTEN
+            || hlsl_version_lt(ctx, 2, 1); /* Always flatten branches for SM < 2.1. */
+
+    if (force_flatten)
+    {
+        if (!can_flatten_conditional_block(ctx, &iff->then_block)
+                || !can_flatten_conditional_block(ctx, &iff->else_block))
+            return false;
+    }
+    else if (!is_conditional_block_simple(&iff->then_block) || !is_conditional_block_simple(&iff->else_block))
+    {
+        /* Only flatten simple blocks by default. */
+        return false;
+    }
+
+    flatten_ctx.cond = iff->condition.node;
+
+    flatten_ctx.is_then = true;
+    hlsl_transform_ir(ctx, lower_conditional_block_instrs, &iff->then_block, &flatten_ctx);
+
+    flatten_ctx.is_then = false;
+    hlsl_transform_ir(ctx, lower_conditional_block_instrs, &iff->else_block, &flatten_ctx);
+
+    list_move_before(&instr->entry, &iff->then_block.instrs);
+    list_move_before(&instr->entry, &iff->else_block.instrs);
+    list_remove(&instr->entry);
+    hlsl_free_instr(instr);
+    return true;
+}
+
 static bool normalize_switch_cases(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context)
 {
     struct hlsl_ir_switch_case *c, *def = NULL;
@@ -8494,6 +8721,7 @@ static void hlsl_run_folding_passes(struct hlsl_ctx *ctx, struct hlsl_block *bod
         progress |= replace_ir(ctx, fold_swizzle_chains, body);
         progress |= replace_ir(ctx, fold_trivial_swizzles, body);
         progress |= hlsl_transform_ir(ctx, remove_trivial_conditional_branches, body, NULL);
+        progress |= hlsl_transform_ir(ctx, flatten_conditional_branches, body, NULL);
     } while (progress);
     replace_ir(ctx, fold_redundant_casts, body);
 }
@@ -10047,11 +10275,9 @@ static void sm1_generate_vsir_instr_if(struct hlsl_ctx *ctx, struct vsir_program
     struct hlsl_ir_node *instr = &iff->node;
     struct vkd3d_shader_instruction *ins;
 
-    if (hlsl_version_lt(ctx, 2, 1))
-    {
-        hlsl_fixme(ctx, &instr->loc, "Flatten \"if\" conditionals branches.");
-        return;
-    }
+    /* Conditional branches should have already been flattened for SM < 2.1. */
+    VKD3D_ASSERT(hlsl_version_ge(ctx, 2, 1));
+
     VKD3D_ASSERT(condition->data_type->e.numeric.dimx == 1 && condition->data_type->e.numeric.dimy == 1);
 
     if (!(ins = generate_vsir_add_program_instruction(ctx, program, &instr->loc, VSIR_OP_IFC, 0, 2)))
@@ -14880,6 +15106,14 @@ static void process_entry_function(struct hlsl_ctx *ctx, struct list *semantic_v
 
     hlsl_run_folding_passes(ctx, body);
 
+    if (profile->major_version < 4)
+    {
+        /* Ternary operations can be potentially introduced by hlsl_run_folding_passes(). */
+        replace_ir(ctx, lower_ternary, body);
+        if (ctx->profile->type != VKD3D_SHADER_TYPE_PIXEL)
+            replace_ir(ctx, lower_cmp, body);
+    }
+
     do
         compute_liveness(ctx, body);
     while (hlsl_transform_ir(ctx, dce, body, NULL));
diff --git a/libs/vkd3d-shader/vkd3d_shader_private.h b/libs/vkd3d-shader/vkd3d_shader_private.h
index 976c50451..6802f4ed2 100644
--- a/libs/vkd3d-shader/vkd3d_shader_private.h
+++ b/libs/vkd3d-shader/vkd3d_shader_private.h
@@ -175,6 +175,7 @@ enum vkd3d_shader_error
     VKD3D_SHADER_ERROR_HLSL_MISSING_PRIMITIVE_TYPE      = 5043,
     VKD3D_SHADER_ERROR_HLSL_MISPLACED_STREAM_OUTPUT     = 5044,
     VKD3D_SHADER_ERROR_HLSL_MISSING_INPUT_PATCH         = 5045,
+    VKD3D_SHADER_ERROR_HLSL_CANNOT_FLATTEN              = 5046,
 
     VKD3D_SHADER_WARNING_HLSL_IMPLICIT_TRUNCATION       = 5300,
     VKD3D_SHADER_WARNING_HLSL_DIVISION_BY_ZERO          = 5301,
diff --git a/tests/hlsl/cf-cond-types.shader_test b/tests/hlsl/cf-cond-types.shader_test
index ce49600a2..449c1581c 100644
--- a/tests/hlsl/cf-cond-types.shader_test
+++ b/tests/hlsl/cf-cond-types.shader_test
@@ -113,7 +113,7 @@ float4 main() : sv_target
     return 0;
 }
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 uniform float1 f;
 float4 main() : sv_target
 {
@@ -124,8 +124,8 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float4 -2.0 0.0 0.0 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (1.0, 1.0, 1.0, 1.0)
+draw quad
+probe (0, 0) f32(1.0, 1.0, 1.0, 1.0)
 uniform 0 float4 -0.0 0.0 0.0 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.0, 0.0, 0.0, 0.0)
+draw quad
+probe (0, 0) f32(0.0, 0.0, 0.0, 0.0)
diff --git a/tests/hlsl/conditional.shader_test b/tests/hlsl/conditional.shader_test
index 1f9d5b5db..3e99b91f3 100644
--- a/tests/hlsl/conditional.shader_test
+++ b/tests/hlsl/conditional.shader_test
@@ -1,4 +1,4 @@
-[pixel shader todo(sm<4)]
+[pixel shader]
 uniform float4 u;
 
 float4 main() : sv_target
@@ -11,13 +11,13 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float4 0.0 0.0 0.0 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.9, 0.8, 0.7, 0.6)
+draw quad
+probe (0, 0) f32(0.9, 0.8, 0.7, 0.6)
 uniform 0 float4 0.1 0.0 0.0 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.1, 0.2, 0.3, 0.4)
+draw quad
+probe (0, 0) f32(0.1, 0.2, 0.3, 0.4)
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 uniform float4 u;
 
 float4 main() : sv_target
@@ -29,7 +29,7 @@ float4 main() : sv_target
         return float4(0.9, 0.8, 0.7, 0.6);
 }
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 uniform float4 u;
 
 float4 main() : sv_target
@@ -43,8 +43,8 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float4 0.0 0.0 0.0 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.9, 0.8, 0.7, 0.6)
+draw quad
+probe (0, 0) f32(0.9, 0.8, 0.7, 0.6)
 
 [pixel shader fail(sm<6)]
 float4 u;
@@ -370,7 +370,7 @@ todo(msl & sm>=6) draw quad
 probe (0, 0) rgba (1.0, 1.0, 0.0, 4.0)
 
 % Branches with resource stores can't be flattened.
-[pixel shader fail(sm<6) todo]
+[pixel shader fail(sm<6)]
 uniform float a;
 RWBuffer<float> u;
 
diff --git a/tests/hlsl/discard.shader_test b/tests/hlsl/discard.shader_test
index 8ed9ceeae..7398b229a 100644
--- a/tests/hlsl/discard.shader_test
+++ b/tests/hlsl/discard.shader_test
@@ -16,7 +16,7 @@ uniform 0 float4 9 8 7 6
 todo(sm<4 | glsl | msl & sm>=6) draw quad
 probe (0, 0) f32(1, 2, 3, 4)
 
-[pixel shader todo(sm<4)]
+[pixel shader todo]
 uniform float4 x;
 
 float4 main() : sv_target
@@ -29,10 +29,10 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float4 1 2 3 4
-todo(sm<4 | glsl | msl & sm>=6) draw quad
+todo(sm<6 | msl) draw quad
 probe (0, 0) f32(1, 2, 3, 4)
 uniform 0 float4 9 8 7 6
-todo(sm<4 | glsl | msl & sm>=6) draw quad
+todo(sm<6 | msl) draw quad
 probe (0, 0) f32(1, 2, 3, 4)
 
 [pixel shader todo(sm<4)]
@@ -56,7 +56,7 @@ uniform 0 float4 9 8 7 6
 todo(sm<4 | glsl | msl & sm>=6) draw quad
 probe (0, 0) f32(1, 2, 3, 4)
 
-[pixel shader todo(sm<4)]
+[pixel shader todo]
 uniform float4 x;
 
 float4 main() : sv_target
@@ -72,10 +72,10 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float4 1 2 3 4
-todo(sm<4 | glsl | msl & sm>=6) draw quad
+todo(sm<6 | msl) draw quad
 probe (0, 0) f32(1, 2, 3, 4)
 uniform 0 float4 9 8 7 6
-todo(sm<4 | glsl | msl & sm>=6) draw quad
+todo(sm<6 | msl) draw quad
 probe (0, 0) f32(1, 2, 3, 4)
 
 [require]
diff --git a/tests/hlsl/for.shader_test b/tests/hlsl/for.shader_test
index e49cf09e3..de291b7b0 100644
--- a/tests/hlsl/for.shader_test
+++ b/tests/hlsl/for.shader_test
@@ -5,7 +5,7 @@ void main(float4 pos : position, out float tex : texcoord, out float4 out_pos :
     out_pos = pos;
 }
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 float4 main(float tex : texcoord) : sv_target
 {
     int i;
@@ -23,10 +23,10 @@ float4 main(float tex : texcoord) : sv_target
 }
 
 [test]
-todo(sm<4) draw quad
-probe (  0, 0, 159, 480) rgba (10.0, 35.0, 0.0, 0.0)
-probe (161, 0, 479, 480) rgba (10.0, 38.0, 0.0, 0.0)
-probe (481, 0, 640, 480) rgba ( 5.0, 10.0, 0.0, 0.0)
+draw quad
+probe (  0, 0, 159, 480) f32(10.0, 35.0, 0.0, 0.0)
+probe (161, 0, 479, 480) f32(10.0, 38.0, 0.0, 0.0)
+probe (481, 0, 640, 480) f32( 5.0, 10.0, 0.0, 0.0)
 
 [require]
 shader model >= 4.0
diff --git a/tests/hlsl/function-return.shader_test b/tests/hlsl/function-return.shader_test
index 208bfeebb..1e9b275d8 100644
--- a/tests/hlsl/function-return.shader_test
+++ b/tests/hlsl/function-return.shader_test
@@ -32,7 +32,7 @@ float4 main() : sv_target
 draw quad
 probe (0, 0) rgba (0.2, 0.1, 0.8, 0.5);
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 uniform float f;
 
 float func(out float o)
@@ -79,19 +79,19 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float 0.1
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.3, 0.2, 0.6, 0.3) 1
+draw quad
+probe (0, 0) f32(0.3, 0.2, 0.6, 0.3) 1
 uniform 0 float 0.4
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.6, 0.5, 0.6, 0.3) 1
+draw quad
+probe (0, 0) f32(0.6, 0.5, 0.6, 0.3) 1
 uniform 0 float 0.6
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.6, 0.5, 0.4, 0.5) 1
+draw quad
+probe (0, 0) f32(0.6, 0.5, 0.4, 0.5) 1
 uniform 0 float 0.8
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.8, 0.7, 0.4, 0.5) 1
+draw quad
+probe (0, 0) f32(0.8, 0.7, 0.4, 0.5) 1
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 uniform float f;
 
 float func(out float o)
@@ -134,14 +134,14 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float 0.1
-todo(sm<4 | msl & sm>=6) draw quad
-probe (0, 0) rgba (0.2, 0.1, 0.2, 0.1) 1
+todo(msl & sm>=6) draw quad
+probe (0, 0) f32(0.2, 0.1, 0.2, 0.1) 1
 uniform 0 float 0.5
-todo(sm<4 | msl & sm>=6) draw quad
-probe (0, 0) rgba (0.5, 0.4, 1.0, 0.9) 1
+todo(msl & sm>=6) draw quad
+probe (0, 0) f32(0.5, 0.4, 1.0, 0.9) 1
 uniform 0 float 0.9
-todo(sm<4 | msl & sm>=6) draw quad
-probe (0, 0) rgba (1.0, 0.9, 1.0, 0.6) 1
+todo(msl & sm>=6) draw quad
+probe (0, 0) f32(1.0, 0.9, 1.0, 0.6) 1
 
 [pixel shader]
 float func(out float o)
@@ -184,7 +184,7 @@ float4 main() : sv_target
 draw quad
 probe (0, 0) rgba (0.4, 0.3, 0.3, 0.9) 1
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 uniform float f;
 
 float func(out float o)
@@ -235,26 +235,26 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.3, 0.2, 0.3, 0.3) 1
+draw quad
+probe (0, 0) f32(0.3, 0.2, 0.3, 0.3) 1
 
 uniform 0 float 0.1
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.3, 0.3, 0.3, 0.3) 1
+draw quad
+probe (0, 0) f32(0.3, 0.3, 0.3, 0.3) 1
 
 uniform 0 float 0.3
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.3, 0.5, 0.3, 0.3) 1
+draw quad
+probe (0, 0) f32(0.3, 0.5, 0.3, 0.3) 1
 
 uniform 0 float 0.7
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.3, 0.9, 0.7, 0.6) 1
+draw quad
+probe (0, 0) f32(0.3, 0.9, 0.7, 0.6) 1
 
 uniform 0 float 0.9
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.4, 0.1, 0.7, 0.6) 1
+draw quad
+probe (0, 0) f32(0.4, 0.1, 0.7, 0.6) 1
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 
 uniform float4 f[3];
 
@@ -291,24 +291,24 @@ float4 main() : sv_target
 uniform 0 float4 0.3 0.0 0.0 0.0
 uniform 4 float4 0.0 0.0 0.0 0.0
 uniform 8 float4 0.1 0.0 0.0 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.3, 0.2, 0.6, 0.6) 1
+draw quad
+probe (0, 0) f32(0.3, 0.2, 0.6, 0.6) 1
 
 uniform 4 float4 0.35 0.0 0.0 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.3, 0.3, 0.6, 0.6) 1
+draw quad
+probe (0, 0) f32(0.3, 0.3, 0.6, 0.6) 1
 
 uniform 8 float4 0.5 0.0 0.0 0.0
-todo(sm<4) draw quad
+draw quad
 % Apparently fixed by Mesa commit 9785fa460c41b9498c24a82b98069655a91224c5,
 % which only adds optimization; so it might just be hiding the bug rather than
 % solving it; see also https://gitlab.freedesktop.org/mesa/mesa/-/issues/13092
 bug(mesa<25.1 & llvm>=16) probe (0, 0) f32(0.3, 0.5, 0.6, 0.6) 1
 
 uniform 0 float4 1.0 0.0 0.0 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.3, 0.5, 0.6, 0.6) 1
+draw quad
+probe (0, 0) f32(0.3, 0.5, 0.6, 0.6) 1
 
 uniform 4 float4 2.0 0.0 0.0 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.4, 0.1, 0.6, 0.6) 1
+draw quad
+probe (0, 0) f32(0.4, 0.1, 0.6, 0.6) 1
diff --git a/tests/hlsl/hard-copy-prop.shader_test b/tests/hlsl/hard-copy-prop.shader_test
index 779111b5b..20780f55e 100644
--- a/tests/hlsl/hard-copy-prop.shader_test
+++ b/tests/hlsl/hard-copy-prop.shader_test
@@ -1,4 +1,4 @@
-[pixel shader todo(sm<4)]
+[pixel shader]
 float cond;
 
 float4 main() : sv_target
@@ -17,14 +17,13 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (2.0, 2.0, 2.0, 2.0)
+draw quad
+probe (0, 0) f32(2.0, 2.0, 2.0, 2.0)
 uniform 0 float 1.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (-2.0, -2.0, -2.0, -2.0)
+draw quad
+probe (0, 0) f32(-2.0, -2.0, -2.0, -2.0)
 
-
-[pixel shader todo(sm<4)]
+[pixel shader]
 float cond;
 
 float4 main() : sv_target
@@ -43,14 +42,13 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (2.0, 2.0, 2.0, 2.0)
+draw quad
+probe (0, 0) f32(2.0, 2.0, 2.0, 2.0)
 uniform 0 float 1.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (20.0, 20.0, 20.0, 20.0)
+draw quad
+probe (0, 0) f32(20.0, 20.0, 20.0, 20.0)
 
-
-[pixel shader todo(sm<4)]
+[pixel shader]
 float cond;
 
 float4 main() : sv_target
@@ -69,11 +67,11 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (1.0, 4.0, 0.0, 0.0)
+draw quad
+probe (0, 0) f32(1.0, 4.0, 0.0, 0.0)
 uniform 0 float 1.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (1.0, 40.0, 0.0, 0.0)
+draw quad
+probe (0, 0) f32(1.0, 40.0, 0.0, 0.0)
 
 [pixel shader]
 float x[6];
@@ -206,7 +204,7 @@ float4 main() : sv_target
     return 0;
 }
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 float a;
 
 float4 main() : sv_target
@@ -221,8 +219,8 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float 11.0
-todo(sm<4) draw quad
-probe (0, 0) rgba(1, 2, 3, 5)
+draw quad
+probe (0, 0) f32(1, 2, 3, 5)
 
 [require]
 shader model < 5.1
diff --git a/tests/hlsl/return.shader_test b/tests/hlsl/return.shader_test
index 594a032a4..f0e4c12dc 100644
--- a/tests/hlsl/return.shader_test
+++ b/tests/hlsl/return.shader_test
@@ -25,7 +25,7 @@ void main(out float4 ret : sv_target)
 draw quad
 probe (0, 0) rgba (0.1, 0.2, 0.3, 0.4)
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 uniform float f;
 
 float4 main() : sv_target
@@ -37,13 +37,13 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float 0.2
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.1, 0.2, 0.3, 0.4)
+draw quad
+probe (0, 0) f32(0.1, 0.2, 0.3, 0.4)
 uniform 0 float 0.8
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.5, 0.6, 0.7, 0.8)
+draw quad
+probe (0, 0) f32(0.5, 0.6, 0.7, 0.8)
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 uniform float f;
 
 void main(out float4 ret : sv_target)
@@ -63,13 +63,13 @@ void main(out float4 ret : sv_target)
 
 [test]
 uniform 0 float 0.2
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.3, 0.4, 0.5, 0.6)
+draw quad
+probe (0, 0) f32(0.3, 0.4, 0.5, 0.6)
 uniform 0 float 0.8
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.1, 0.2, 0.3, 0.4)
+draw quad
+probe (0, 0) f32(0.1, 0.2, 0.3, 0.4)
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 uniform float f;
 
 void main(out float4 ret : sv_target)
@@ -89,16 +89,16 @@ void main(out float4 ret : sv_target)
 
 [test]
 uniform 0 float 0.1
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.1, 0.2, 0.3, 0.4) 1
+draw quad
+probe (0, 0) f32(0.1, 0.2, 0.3, 0.4) 1
 uniform 0 float 0.5
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.2, 0.3, 0.4, 0.5) 1
+draw quad
+probe (0, 0) f32(0.2, 0.3, 0.4, 0.5) 1
 uniform 0 float 0.9
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.5, 0.6, 0.7, 0.8) 1
+draw quad
+probe (0, 0) f32(0.5, 0.6, 0.7, 0.8) 1
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 uniform float f;
 
 void main(out float4 ret : sv_target)
@@ -115,14 +115,14 @@ void main(out float4 ret : sv_target)
 
 [test]
 uniform 0 float 0.1
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.1, 0.2, 0.3, 0.4) 1
+draw quad
+probe (0, 0) f32(0.1, 0.2, 0.3, 0.4) 1
 uniform 0 float 0.5
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.5, 0.6, 0.7, 0.8) 1
+draw quad
+probe (0, 0) f32(0.5, 0.6, 0.7, 0.8) 1
 uniform 0 float 0.9
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.4, 0.5, 0.6, 0.7) 1
+draw quad
+probe (0, 0) f32(0.4, 0.5, 0.6, 0.7) 1
 
 [pixel shader]
 void main(out float4 ret : sv_target)
@@ -141,7 +141,7 @@ void main(out float4 ret : sv_target)
 draw quad
 probe (0, 0) rgba (0.2, 0.4, 0.6, 0.8)
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 uniform float f;
 
 void main(out float4 ret : sv_target)
@@ -160,26 +160,26 @@ void main(out float4 ret : sv_target)
 
 [test]
 uniform 0 float 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.1, 0.1, 0.1, 0.1) 1
+draw quad
+probe (0, 0) f32(0.1, 0.1, 0.1, 0.1) 1
 
 uniform 0 float 0.1
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.2, 0.2, 0.2, 0.2) 1
+draw quad
+probe (0, 0) f32(0.2, 0.2, 0.2, 0.2) 1
 
 uniform 0 float 0.3
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.4, 0.4, 0.4, 0.4) 1
+draw quad
+probe (0, 0) f32(0.4, 0.4, 0.4, 0.4) 1
 
 uniform 0 float 0.7
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.8, 0.8, 0.8, 0.8) 1
+draw quad
+probe (0, 0) f32(0.8, 0.8, 0.8, 0.8) 1
 
 uniform 0 float 0.9
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.9, 0.9, 0.9, 0.9) 1
+draw quad
+probe (0, 0) f32(0.9, 0.9, 0.9, 0.9) 1
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 uniform float f;
 
 void main(out float4 ret : sv_target)
@@ -204,13 +204,13 @@ void main(out float4 ret : sv_target)
 
 [test]
 uniform 0 float 0.2
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.2, 0.2, 0.2, 0.2)
+draw quad
+probe (0, 0) f32(0.2, 0.2, 0.2, 0.2)
 uniform 0 float 0.8
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.5, 0.5, 0.5, 0.5)
+draw quad
+probe (0, 0) f32(0.5, 0.5, 0.5, 0.5)
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 
 uniform float4 f[3];
 
@@ -236,23 +236,23 @@ void main(out float4 ret : sv_target)
 uniform 0 float4 0.3 0.0 0.0 0.0
 uniform 4 float4 0.0 0.0 0.0 0.0
 uniform 8 float4 0.1 0.0 0.0 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.1, 0.1, 0.1, 0.1) 1
+draw quad
+probe (0, 0) f32(0.1, 0.1, 0.1, 0.1) 1
 
 uniform 4 float4 0.35 0.0 0.0 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.2, 0.2, 0.2, 0.2) 1
+draw quad
+probe (0, 0) f32(0.2, 0.2, 0.2, 0.2) 1
 
 uniform 8 float4 0.5 0.0 0.0 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.4, 0.4, 0.4, 0.4) 1
+draw quad
+probe (0, 0) f32(0.4, 0.4, 0.4, 0.4) 1
 
 uniform 0 float4 1.0 0.0 0.0 0.0
-todo(sm<4) draw quad
-probe (0, 0) rgba (0.4, 0.4, 0.4, 0.4) 1
+draw quad
+probe (0, 0) f32(0.4, 0.4, 0.4, 0.4) 1
 
 uniform 4 float4 2.0 0.0 0.0 0.0
-todo(sm<4) draw quad
+draw quad
 % Apparently fixed by Mesa commit 9785fa460c41b9498c24a82b98069655a91224c5,
 % which only adds optimization; so it might just be hiding the bug rather than
 % solving it; see also https://gitlab.freedesktop.org/mesa/mesa/-/issues/13092
diff --git a/tests/hlsl/unroll.shader_test b/tests/hlsl/unroll.shader_test
index cb21079be..8a9977a71 100644
--- a/tests/hlsl/unroll.shader_test
+++ b/tests/hlsl/unroll.shader_test
@@ -1,6 +1,6 @@
 % Unrolling edge cases.
 
-[pixel shader todo(sm<4)]
+[pixel shader]
 float a : register(c0);
 
 float4 main() : sv_target
@@ -19,8 +19,8 @@ float4 main() : sv_target
 
 [test]
 uniform 0 float 1
-todo(sm<4) draw quad
-probe (0,0) rgba(10.0, 1.0, 2.0, 3.0)
+draw quad
+probe (0,0) f32(10.0, 1.0, 2.0, 3.0)
 
 [pixel shader]
 float4 main() : sv_target