From 1b7de686f00d22c8b20d26ef6905acd2c5441f8a Mon Sep 17 00:00:00 2001 From: "opensearch-trigger-bot[bot]" <98922864+opensearch-trigger-bot[bot]@users.noreply.github.com> Date: Fri, 6 Dec 2024 16:29:55 -0600 Subject: [PATCH] Add vectorsearch README update regarding recall accuracy (#516) (#521) (cherry picked from commit bb212c172886571ae49e80bd002e6b7a57f5f389) Signed-off-by: Ian Hoang Signed-off-by: github-actions[bot] Co-authored-by: github-actions[bot] --- vectorsearch/README.md | 3 +++ 1 file changed, 3 insertions(+) diff --git a/vectorsearch/README.md b/vectorsearch/README.md index f3058287..6267761c 100644 --- a/vectorsearch/README.md +++ b/vectorsearch/README.md @@ -36,6 +36,9 @@ opensearch-benchmark execute-test \ --pipeline benchmark-only \ --kill-running-processes ``` +### Best Recall Results + +For accurate recall results, we recommend users to keep `search_clients` equal to or less than the number of CPU cores on the load generation host that is running the benchmark. ## Current Procedures