Leonardo Kuffo
@LeonardoKuffo
Followers
624
Following
6K
Media
951
Statuses
33K
🇪🇨 🇳🇱 Hago música y programo, also: Database Researcher at CWI ;)
Amsterdam
Joined June 2010
Prateek Gaur and co at @Snowflake reproduced the (great) results for the ALP encoding algorithm from @cwi_da / @afroozeh3 / @peterabcz. ALP achieves ZSTD levels of compression and much faster decode. We are discussing adding it to @ApacheParquet: https://t.co/gxwF5QqtNO
0
10
78
Me enferme, me despidieron y llevo años esperando justicia... Les comparto esta historia... 1/ 💪Desde hace años trabajé con toda dedicación en un colegio privado de la ciudad. Di lo mejor de mí por mis estudiantes, con pasión, tiempo y compromiso. ...
1
2
0
@SIGMODConf Berlin is a wrap! Many 🙏 to the organizers! Next stop for @cwi_da is @VLDBconf London to present: - https://t.co/2STsOKrktp v0.1 - spilling multi-operator joins (via @duckdblabs) - the SQLStorm benchmark consisting of 30k LLM-generated complex queries (via TUM)
0
2
11
@LeonardoKuffo presenting his SIGMOD2025 paper on PDX. PDX is a vertical layout that can accelerate vector search in principle in any vector index technique (it makes the distance calculation faster, using better SIMD + pruning). https://t.co/z8dYKxDn4Z
https://t.co/eja3WJ6axc
0
1
4
And.. Azim Afroozeh put a lot of effort in open-sourcing the ALP floating point compressor ( https://t.co/gw0qhFeDSN). Leonardo Kuffo had written with him the SIGMOD2024 paper which now won a reproducibility award! + 🙏🙏 to the reproducibility committee - this is a ton of work
But @cwi_da has no reason to complain, here in Berlin. Leonardo Kuffo at the preceding DaMoN2025 workshop won the Best Paper Award for a study that showed that for vector databases, it matters a lot which AWS CPU you pick. Congratulations to him!
1
2
12
But @cwi_da has no reason to complain, here in Berlin. Leonardo Kuffo at the preceding DaMoN2025 workshop won the Best Paper Award for a study that showed that for vector databases, it matters a lot which AWS CPU you pick. Congratulations to him!
SIGMOD2025 for the 1st time used a schedule where most papers are presented as posters only Tips for next time - gather user interest data prior to deciding poster/paper & room assignment - present posters in a (high ceiling) room with good acoustics & allot enough space + time
0
1
13
Today we're launching DuckLake, an integrated data lake and catalog format powered by SQL. DuckLake unlocks next-generation data warehousing where compute is local, consistency central, and storage scales till infinity. DuckLake is an open standard and we've implemented it in
21
200
689
Aunque parezca fuerte no quiere decir que nada me lastima
0
0
3
This paper benchmarks different cloud CPUs across vector search methods and quantization to find the best cost-performance balance. Methods 🔧: → Different CPU microarchitectures like AWS Graviton 3 and AMD Zen 4 were evaluated. → Benchmarks used Inverted Files (IVF) and
1
9
13
New paper from CWI on price-performance for different vector search architectures on different cloud CPU architectures 👀 https://t.co/IwVPDal7yv
arxiv.org
Vector databases have emerged as a new type of systems that support efficient querying of high-dimensional vectors. Many of these offer their database as a service in the cloud. However, the...
0
1
3
Cloud, Vector Search Speed & Costs, and the Importance of SIMD An interesting study was just published on @arXiv - comparing @Meta’s FAISS to @Unum_Cloud USearch. Looks realistic and is the perfect starting point for my next wave of optimizations 😊 https://t.co/fNqRrhNkOk
3
8
50
Bang for the Buck: Vector Search on Cloud CPUs @LeonardoKuffo et al. reveal that AWS Graviton3 provides best queries-per-dollar for vector search, even outperforming newer architectures, while CPU performance varies significantly. 📝 https://t.co/CBbWxJq5Bo
arxiv.org
Vector databases have emerged as a new type of systems that support efficient querying of high-dimensional vectors. Many of these offer their database as a service in the cloud. However, the...
0
4
14
Are you doing Vector Search in the cloud? Choose your instances carefully! A cheaper one may be 3x faster! In this article, we explore which CPUs available in AWS are the best for vector search algorithms (and why) We will present this work at #DAMON2025 @SIGMODConf @peterabcz
Bang for the Buck: Vector Search on Cloud CPUs @LeonardoKuffo et al. reveal that AWS Graviton3 provides best queries-per-dollar for vector search, even outperforming newer architectures, while CPU performance varies significantly. 📝 https://t.co/CBbWxJq5Bo
1
1
4
Que vuelvan las corridas de toros a Baleares ya es una vergüenza, pero sentirse orgulloso de que los niños puedan disfrutar del maltrato animal es de absoluto degenerado.
Es para mí un honor que el esfuerzo realizado durante tantos años haya dado sus frutos y que los niños puedan volver a disfrutar de la tauromaquia. Gracias a la empresa “Balears Cambio de Tercio” por su espontáneo reconocimiento. Que triunfe la libertad es un logro de todos. Nos
17
202
2K
PDX: A Data Layout for Vector Similarity Search Introduces a vertical data layout for vectors that accelerates similarity search through dimension-by-dimension processing, outperforming SIMD-optimized kernels. 📝 https://t.co/bDHFcg4oDn 👨🏽💻 https://t.co/9ZBcQDN2aG
github.com
⚡ Faster similarity search with PDX: A vertical data layout for vectors - cwida/PDX
1
9
19
🚨🇪🇨 ¡EL VIDEO DE MR. BEAST EN ECUADOR! El famoso youtuber, #MrBeast, visitó hace algunos meses nuestro país y muchos desconocían el motivo. Pues hoy lanzó un video donde ayudó a 2000 personas con amputaciones, para que volvieran a caminar. 50 de ellas fueron en Ecuador.
62
1K
6K
Who's ready? #AGDQ2025 begins tomorrow! Last year, @GamesDoneQuick raised over $2.5 million for @preventcancer! That's $2.5 million for research, free cancer screenings & more—across the U.S. & around the 🌎. Tune in tomorrow beginning at 11:30 a.m. ET. ⬇️
twitch.tv
Fright Fatales Day 1 - Tomb Raider 2 The Haunted Mansion Any% by @NondescriptMidnight !FF !GDQueer !BAF !AGDQ
1
9
26
Buckle up because we're crashing into the new year with my annual database retrospective: License change blowbacks! @databricks vs. @SnowflakeDB gangwar! @DuckDB shotgun weddings! Buying a college quarterback with database money for your new lover!
cs.cmu.edu
Andy rises from the ashes of his dead startup and discusses what happened in 2024 in the database game.
18
165
737
Introducing Willow, our new state-of-the-art quantum computing chip with a breakthrough that can reduce errors exponentially as we scale up using more qubits, cracking a 30-year challenge in the field. In benchmark tests, Willow solved a standard computation in <5 mins that would
3K
12K
77K
The video for my "What Goes Around Comes Around... And Around" talk at @cwi_da is now available: https://t.co/ss3G2UAfEF 📊Slides: https://t.co/NudKne8NlF 📄Paper:
4
34
195