solidangle
Joined 11 years ago
Karma
578
Submissions
119
Member since
May 9
Recent submissions
Comment · 3 years ago
Comment · 3 years ago
> The ANN index IVF implemented in pgvector has very poor performance, with only around 50% recall.
My understanding is that this mostly due to the default settings that pgvector uses (nprobes = 3) and not due to the usage of IVF. The recall would improve significantly with better defaults. This of course would also increase the latency of vector searches, but that is the trade-off of using IVF instead of HNSW (worse latency at high recall, but much lower storage/memory costs).
Comment · 3 years ago
This sounds like an interesting contrarian take. Would you care to elaborate?
Comment · 3 years ago
But you don't have to write map-reduce jobs at all? You can just write SQL queries or Pandas programs, and they automatically get parallelized by Databricks. Databricks is a data warehouse (just like Snowflake).
Comment · 4 years ago
TIS-100 brought back some of the same joy that I had when I began with programming many years ago. Since then programming has become boring for me, but the weird limitations of TIS-100 almost made it feel like I was starting fresh again. I loved learning tons of small tricks that I could apply to beat the harder levels.
Shenzhen I/O otoh felt like work.
Comment · 4 years ago
2-day delivery would be uncompetitive in The Netherlands. Next day delivery is the standard for webshops here, and most offer same day delivery as a paid upgrade when you either order before noon or live in a large city.
Comment · 4 years ago
Wow, $14.99/month is a stark difference with the €2.99/month that they charge for a Prime subscription here in The Netherlands. I wonder if they will increase the subscription fee here to the same level once they reach the desired market share in the video streaming and e-commerce spaces here.
Comment · 4 years ago
For those wondering what the differences between Raft and Paxos are: I can highly recommend the following paper https://arxiv.org/abs/2004.05074 It presents Paxos similarly to how Raft was originally presented.
Comment · 4 years ago
You might be interested in this paper [1] which presents Paxos in the same way as Raft is presented in the Raft paper. In my view it really illustrates that most of the simplicity of Raft comes from its excellent presentation and not necessarily from its algorithmic simplicity.
Comment · 4 years ago
(Multi-)Paxos and Raft are incredibly similar. I would even argue that Raft is just a variant of Paxos in the same way that e.g. Fast Paxos and EPaxos are variants of Paxos. Most of the Raft algorithm is identical to the Paxos algorithm. The main difference between the two algorithms is how leadership election works. In my view the main reason why Raft is considered to be simpler is because it was presented in a really clear paper. Paxos becomes equally simple when using a similar representation as used in that paper. See this paper: https://arxiv.org/pdf/2004.05074
Comment · 4 years ago
Netflix is also the most expensive service that I am subscribed to. Netflix is €14 and Amazon Prime is only €6. I basically only watch Netflix for a few of their originals (which are arguably aren't that great).
Comment · 4 years ago
Does anyone know why this post is no longer present on the first page (or the second) despite racking up over 60 points in less than 30 minutes?
Comment · 5 years ago
Dutch: honderdvijfenzeventig (hundred five and seventy)
Comment · 5 years ago
When was the last time you used Databricks? You should definitely try it again. Their product offering has improved a lot in the past few years.
> broad feature set
My experience is that the feature sets of Snowflake and Databricks are very similar. Both have time travel support. Snowflake has materialized views, but Databricks has Delta Live Tables. Databricks has a distributed Pandas API, but Snowflake recently introduced Snowpark. Databricks also has autoscaling and they recently launched a serverless offering that makes autoscaling super fast aswell.
[1] https://www.cidrdb.org/cidr2023/papers/p66-wolde.pdf