You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Adds infino, a Rust engine that runs SQL, full-text (BM25), and vector search over one copy of your data on object storage. The data is spec-compliant Parquet on S3, Azure, GCS, or local disk, so anything that reads Parquet can read it. This entry runs the SQL path.
Some Pointers:
The harness is a single Rust binary that pulls only the published infino crate (0.1.10), with a committed lockfile and a plain release build, so a fresh VM reproduces it with nothing else to install. queries.sql is the same as the datafusion entry: the two type fixes datafusion does inline (day count to DATE, binary to UTF-8) are done at load time instead.
infino stays up as a server between queries. start opens the table once and serves it over a unix socket, query is a thin client, and BENCH_RESTARTABLE=yes restarts it before each cold run. That meets the true-cold rule (cold first run, warm second and third), so there is no lukewarm-cold-run tag.
Please Note: load includes infino compacting into its superfile format (256 MB segments). That is part of writing the data and is counted in load_time, not an optimize slipped in before the queries.
Ran the whole thing on a fresh c6a.4xlarge at 100M rows, built from crates.io: load, then the 43-query sweep with row counts checked. Hot sum 33.7s.
Fixed it @alexey-milovidov. install earlier assumed the runner VM had a C toolchain, but infino's build needs cc; it now apt-installs build-essential + pkg-config first. Would be great if you could re-trigger the run. Thanks!
Results for infino are ready for: c6a.4xlarge.
This pull request is from a fork, so the automation cannot push to it; save this result as infino/results/20260729/c6a.4xlarge.json.
Tick "Allow edits by maintainers" on the pull request to let the automation commit the results itself.
I have updated the results for c6a.4xlarge from this run in the PR.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds infino, a Rust engine that runs SQL, full-text (BM25), and vector search over one copy of your data on object storage. The data is spec-compliant Parquet on S3, Azure, GCS, or local disk, so anything that reads Parquet can read it. This entry runs the SQL path.
Some Pointers:
infinocrate (0.1.10), with a committed lockfile and a plain release build, so a fresh VM reproduces it with nothing else to install.queries.sqlis the same as thedatafusionentry: the two type fixes datafusion does inline (day count to DATE, binary to UTF-8) are done at load time instead.startopens the table once and serves it over a unix socket,queryis a thin client, andBENCH_RESTARTABLE=yesrestarts it before each cold run. That meets the true-cold rule (cold first run, warm second and third), so there is nolukewarm-cold-runtag.loadincludes infino compacting into its superfile format (256 MB segments). That is part of writing the data and is counted inload_time, not an optimize slipped in before the queries.Ran the whole thing on a fresh c6a.4xlarge at 100M rows, built from crates.io: load, then the 43-query sweep with row counts checked. Hot sum 33.7s.