nak: Set variable_latency=0 for !needs_scoreboard

This simplifies usage of estimate_variable_latency a little in that we
can just use it directly in our max() expressions instead of guarding it
with an if.

Reviewed-by: Mary Guillemard <mary@mary.zone>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38939>

src/nouveau/compiler/nak/opt_instr_sched_common.rs

@@ -253,7 +253,11 @@ pub fn estimate_block_weight(cfg: &CFG<BasicBlock>, block_idx: usize) -> u64 {
 /// Memory instructions were copied from L1 data cache latencies.
 /// For instructions not mentioned in the paper, I made up numbers.
 /// This could probably be improved.
-pub fn estimate_variable_latency(sm: u8, op: &Op) -> u32 {
+pub fn estimate_variable_latency(sm: &dyn ShaderModel, op: &Op) -> u32 {
+    if !sm.op_needs_scoreboard(op) {
+        return 0;
+    }
+
     match op {
         // Multi-function unit
         Op::Rro(_) | Op::MuFu(_) => 15,
@@ -265,7 +269,7 @@ pub fn estimate_variable_latency(sm: u8, op: &Op) -> u32 {
         // Integer ALU
         Op::BRev(_) | Op::Flo(_) | Op::PopC(_) => 15,
         Op::IMad(_) | Op::IMul(_) => {
-            assert!(sm < 70);
+            assert!(sm.sm() < 70);
             86
         }
 

src/nouveau/compiler/nak/opt_instr_sched_postpass.rs

@@ -89,12 +89,10 @@ fn generate_dep_graph(sm: &dyn ShaderModel, instrs: &[Instr]) -> DepGraph {
                 } else {
                     sm.raw_latency(&instr.op, i, &instrs[r_ip].op, r_src_idx)
                 };
-                if sm.op_needs_scoreboard(&instr.op) {
+
-                    latency = max(
+                latency =
-                        latency,
+                    max(latency, estimate_variable_latency(sm, &instr.op));
-                        estimate_variable_latency(sm.sm(), &instr.op),
+
-                    );
-                }
                 g.add_edge(ip, r_ip, EdgeLabel { latency });
             }
         });
@@ -126,11 +124,11 @@ fn generate_dep_graph(sm: &dyn ShaderModel, instrs: &[Instr]) -> DepGraph {
             .map(|i| sm.worst_latency(&instr.op, i))
             .max()
             .unwrap_or(0);
-        if sm.op_needs_scoreboard(&instr.op) {
+
-            let var_latency = estimate_variable_latency(sm.sm(), &instr.op)
+        let var_latency = estimate_variable_latency(sm, &instr.op)
-                + sm.exec_latency(&instrs[instrs.len() - 1].op);
+            + sm.exec_latency(&instrs[instrs.len() - 1].op);
-            ready_cycle = max(ready_cycle, var_latency);
+        ready_cycle = max(ready_cycle, var_latency);
-        }
+
         let label = &mut g.nodes[ip].label;
         label.exec_latency = sm.exec_latency(&instr.op);
         label.ready_cycle = ready_cycle;