Merge pull request #4237 from vespa-engine/vekterli/reduce-to-1M-dataset

Reduce to 1M dataset and use synthetic ID attribute
vespa-engine · Oct 23, 2024 · b4e1dec · b4e1dec
2 parents 0867e3b + 63ecef2
commit b4e1dec
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 2 deletions.
diff --git a/tests/performance/mmap_vs_directio/.gitignore b/tests/performance/mmap_vs_directio/.gitignore
@@ -1,2 +1,2 @@
-enwiki-20240801-pages.6819k.jsonl.zst
+enwiki-20240801-pages.*.jsonl.zst
 squad2-questions.fbench.141k.txt
diff --git a/tests/performance/mmap_vs_directio/mmap_vs_directio.rb b/tests/performance/mmap_vs_directio/mmap_vs_directio.rb
@@ -29,7 +29,7 @@ def test_wikipedia_corpus_search_performance
 
     query_file_name = 'squad2-questions.fbench.141k.txt'
     report_io_stat_deltas do
-      feed_file('enwiki-20240801-pages.6819k.jsonl.zst')
+      feed_file('enwiki-20240801-pages.1M.jsonl.zst')
     end
 
     @search_node.trigger_flush # Shovel everything into a disk index
@@ -119,6 +119,7 @@ def benchmark_queries(query_file, type, clients, warmup = false, runtime = 20)
     run_fbench2(@container,
                 node_file,
                 {:clients => clients,
+                 :append_str => '&presentation.summary=minimal&hits=10',
                  :use_post => false,
                  :runtime => runtime,
                  :result_file => result_file},

diff --git a/tests/performance/mmap_vs_directio/wikimedia.sd b/tests/performance/mmap_vs_directio/wikimedia.sd
@@ -1,6 +1,9 @@
 schema wikimedia {
 
   document wikimedia {
+    field id type int {
+      indexing: attribute | summary
+    }
     field title type string {
       indexing: index | summary
       index: enable-bm25
@@ -21,4 +24,8 @@ schema wikimedia {
     }
   }
 
+  document-summary minimal {
+    summary id {}
+  }
+
 }