From 9df3499dc60ec51287659df1d2d152d38595b3ef Mon Sep 17 00:00:00 2001
From: Aria Shrimpton <me@aria.rip>
Date: Tue, 19 Mar 2024 21:32:59 +0000
Subject: improvements to nsplit

---
 src/crates/candelabra/src/profiler/info.rs | 50 ++++++++++--------------------
 1 file changed, 16 insertions(+), 34 deletions(-)

(limited to 'src')

diff --git a/src/crates/candelabra/src/profiler/info.rs b/src/crates/candelabra/src/profiler/info.rs
index 30e7df7..fe66c3c 100644
--- a/src/crates/candelabra/src/profiler/info.rs
+++ b/src/crates/candelabra/src/profiler/info.rs
@@ -105,48 +105,30 @@ impl UsageProfile {
             return None;
         }
 
-        // calculate cost of switching
         let before = &top_by_partition[0].0;
         let after = &top_by_partition[split_idx].0;
 
-        // halfway between the two partitions
+        // figure out when to switch
         let copy_n = {
             let before = self.0[split_idx - 1].avg_n;
             let after = self.0[split_idx].avg_n;
-            (before + after) / 2.0
+            ((before + after) / 2.0).min(before + (before / 2.0))
         };
-        let after_model = candidates.get(after).unwrap();
-        let switching_cost = copy_n * after_model.by_op.get("insert")?.estimatef(copy_n);
 
-        // see if it's "worth it"
-        let before_costs = &costs_by_partitions
-            .iter()
-            .find(|(name, _)| **name == before)
-            .unwrap()
-            .1;
-        let after_costs = &costs_by_partitions
-            .iter()
-            .find(|(name, _)| **name == after)
-            .unwrap()
-            .1;
-        let not_switching_cost = before_costs[split_idx..].iter().sum::<f64>()
-            - after_costs[split_idx..].iter().sum::<f64>();
-
-        debug!("Estimated switching cost: {}", switching_cost);
-        debug!("Estimated not switching cost: {}", not_switching_cost);
-
-        if not_switching_cost < switching_cost {
-            None
-        } else {
-            Some((
-                ContainerSelection::Split {
-                    before: before.to_string(),
-                    threshold: copy_n as usize,
-                    after: after.to_string(),
-                },
-                top_by_partition.iter().map(|(_, v)| v).sum::<f64>() + switching_cost,
-            ))
-        }
+        // calculate cost of switching
+        let before_model = candidates.get(before).unwrap();
+        let after_model = candidates.get(after).unwrap();
+        let switching_cost = (copy_n * after_model.by_op.get("insert")?.estimatef(copy_n))
+            + before_model.by_op.get("clear")?.estimatef(copy_n);
+
+        Some((
+            ContainerSelection::Split {
+                before: before.to_string(),
+                threshold: copy_n as usize,
+                after: after.to_string(),
+            },
+            top_by_partition.iter().map(|(_, v)| v).sum::<f64>() + switching_cost,
+        ))
     }
 
     /// Estimate the cost of using the implementation with the given cost model
-- 
cgit v1.2.3