leaderboard.md

layout	title	permalink
page	Leaderboard	/leaderboard/

<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>

Leaderboard

Metric to display by:

<script src="/SOSDLeaderboard/scripts/sorttable.js" type="text/javascript"></script> Latency (ns) Build time (μs) Size

Select a subset of indexes to display: Show options

<script type="text/javascript"> function changeDisplay() { if ($("#display").is(":visible")) { document.getElementById("display").style.display = "None"; document.getElementById("displayToggle").innerHTML = "Show options"; } else { document.getElementById("display").style.display = "Block"; document.getElementById("displayToggle").innerHTML = "Hide table"; } } </script>

Quick selections: All indexes Read-only indexes Updatable indexes

Displaying results on datasets: --- 200M Datasets --- -- 64-Bit datasets -- All 64-Bit Datasets - Real-world datasets - All real datasets Books Facebook OSM Wiki - Synthetic datasets - All synthetic datasets Lognormal Normal Uniform dense Uniform sparse -- 32-Bit datasets -- All 32-Bit Datasets - Real-world datasets - Books - Synthetic datasets - All synthetic datasets Lognormal Normal Uniform dense Uniform sparse --- Larger Datasets --- Books (400M) Books (600M) Books (800M) OSM (400M) OSM (600M) OSM (800M)

{% assign rows = site.data.latency | map: 'Name' | uniq %} {% for name in rows %} {% endfor %}

{{ name }}

<script src="/SOSDLeaderboard/scripts/checkbox.js" type="text/javascript"></script>

Results below are by model. Click on a header to sort by that measure. Click on an index name to open the relevant GitHub repo.

This leaderboard displays the average lookup time for a randomly selected key in a sorted dataset of size 200M.

Learned indexes are marked in bold.

{% for row in site.data.latency %} {% tablerow pair in row %} {{ pair[1] }} {% endtablerow %} {% endfor %}

	Index / Index Size	XS Up to 0.01% of data size	S Up to 0.1% of data size	M Up to 1% of data size	L Up to 10% of data size	XL No limit	Dataset

Top approach in each column is bold and green. Yellow indicates 2-3x degraded performance. Orange/red indicates > 3x degraded performance. <script src="/SOSDLeaderboard/scripts/annotate.js" type="text/javascript"></script>

{% for row in site.data.buildtimes %} {% tablerow pair in row %} {{ pair[1] }} {% endtablerow %} {% endfor %}

	Index / Index Size	XS Up to 0.01% of data size	S Up to 0.1% of data size	M Up to 1% of data size	L Up to 10% of data size	XL No limit

{% for row in site.data.sizes %} {% tablerow pair in row %} {{ pair[1] }} {% endtablerow %} {% endfor %}

	Index / Index Size	XS Up to 0.01% of data size	S Up to 0.1% of data size	M Up to 1% of data size	L Up to 10% of data size	XL No limit

Plots

Here we provide a plot of index size against lookup latency on SOSD data. Indexes can be added and removed from the plot using the corresponding buttons on the table.

Clear Plot

<script src="https://cdnjs.cloudflare.com/ajax/libs/PapaParse/5.3.0/papaparse.min.js"></script> <script src="https://cdn.jsdelivr.net/npm/chart.js@2.8.0"></script> <script type="text/javascript" src="https://cdnjs.cloudflare.com/ajax/libs/chosen/1.8.7/chosen.jquery.min.js"></script> <script type="text/javascript" src="/SOSDLeaderboard/scripts/graphs/graph_latency.js"></script> <script type="text/javascript" src="/SOSDLeaderboard/scripts/graphs/selector.js"></script> <script type="text/javascript" src="/SOSDLeaderboard/scripts/onStart.js"></script>