Conversation
|
run benchmarks |
|
🤖 Hi @Dandandan, thanks for the request (#20796 (comment)).
Please choose one or more of these with You can also set environment variables on subsequent lines: |
|
run benchmarks |
|
🤖 |
Pre-reserve hashtable capacity to stabilize bucket layout, then sort indices by rotating hashes so bucket-determining low bits become the primary sort key. Use next_power_of_two() to recover actual bucket count from capacity() (which returns elements, not buckets). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
|
🤖: Benchmark completed Details
|
|
🤖 |
|
run benchmarks |
|
🤖: Benchmark completed Details
|
|
run benchmarks |
|
🤖 |
|
🤖: Benchmark completed Details
|
|
🤖 |
|
🤖: Benchmark completed Details
|
…, and reserve based on unique hash count - Add count sort optimization to GroupValuesPrimitive for cache-friendly hash table probing - Skip count sort when hash table fits in L2 cache (<256KB) across all implementations - Use count_unique on sorted hashes instead of reserve(n_rows) to avoid over-allocating - Extract shared count_sort and count_unique helpers to group_values/mod.rs Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
|
run benchmarks |
|
🤖 |
|
🤖: Benchmark completed Details
|
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?