Conversation
|
run benchmarks |
|
🤖 Benchmark completed (GKE) | trigger Details
Resource Usagetpch — base (merge-base)
tpch — branch
|
|
🤖 Benchmark completed (GKE) | trigger Details
Resource Usagetpcds — base (merge-base)
tpcds — branch
|
|
🤖 Benchmark completed (GKE) | trigger Details
Resource Usageclickbench_partitioned — base (merge-base)
clickbench_partitioned — branch
|
|
run benchmarks |
|
🤖 Benchmark completed (GKE) | trigger Details
Resource Usagetpch — base (merge-base)
tpch — branch
|
|
🤖 Benchmark completed (GKE) | trigger Details
Resource Usagetpcds — base (merge-base)
tpcds — branch
|
|
🤖 Benchmark completed (GKE) | trigger Details
Resource Usageclickbench_partitioned — base (merge-base)
clickbench_partitioned — branch
|
|
run benchmarks |
|
🤖 Benchmark completed (GKE) | trigger Details
Resource Usagetpch — base (merge-base)
tpch — branch
|
|
run benchmark tpch10 |
Replace the linked-list chain storage in JoinHashMap with contiguous flat storage using inline single-match optimization. First insert per key is stored inline in the hash table entry (no extra allocation). Duplicate keys go to an overflow buffer that is flattened into contiguous storage in a single O(n) sequential pass. Probe uses batched map.find() calls (4 at a time) for better instruction-level parallelism, and iterates contiguous slices instead of chasing linked-list pointers. Benchmarks (8192 probes, 200K keys, shuffled build): - Unique keys: 20% faster probe, ~0 flatten cost - 4 dups/key: 8% faster probe - 16 dups/key: 5.2x faster probe (cache-friendly vs scattered pointers) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
|
🤖 Benchmark completed (GKE) | trigger Details
Resource Usagetpcds — base (merge-base)
tpcds — branch
|
|
🤖 Benchmark completed (GKE) | trigger Details
Resource Usageclickbench_partitioned — base (merge-base)
clickbench_partitioned — branch
|
|
🤖 Benchmark completed (GKE) | trigger Details
Resource Usagetpch10 — base (merge-base)
tpch10 — branch
|
Which issue does this PR close?
Rationale for this change
What changes are included in this PR?
Are these changes tested?
Are there any user-facing changes?