In December, we will be sponsored by TWT (https://www.twt.de/) and do an evening with two talks.
• Modern SQL: Evolution of a dinosaur (by Markus Winand)
• Have a text similarity tool running on GPUs (by Jörg Bartnick)
+++ Agenda +++
• 18:30 - 19:00: Arrival, get a drink and socialize
• 19:10 - ~19:25: Talk #1 (Have a text similarity tool running on GPUs) + Q&A
• 19:35 - ~20:25: Talk #2 (Modern SQL: Evolution of a dinosaur) + Q&A
• 20:25 - Open End: Socialising
+++ Modern SQL: Evolution of a dinosaur (by Markus Winand) +++
SQL has evolved considerably in recent years. The purely relational dogma was already abandoned in 1999. What followed was the continuous extension through the incorporation of handy features to address problems that are hard to solve with relational algebra. Today, SQL is Turing complete, can process graphs, has a semantic understanding of XML and JSON, is able to automatically keep historic versions, can analyze time series using regular expressions and much more.
This session walks through the SQL standards from 1999 to the current version of 2016. It demonstrates some selected features on the basis of common problems and compares the traditional SQL-92 solution to the modern SQL approach. Of course, the presentation also shows how long these features are supported by various databases so that you immediately know whether the shown solutions work in your environment or not.
In this session, developers and software architects of all levels gain a better understanding of where SQL is today so that they are able to make more educated decisions. A lot has happened since SQL-92!
Markus Winand is an independent author, trainer, and consultant on all things SQL. His book “SQL Performance Explained” has just been translated into a fifth language. It can be purchased as a print edition, and it’s also available to read for free on his use-the-index-luke.comwebsite. Markus is currently working on his second book centered on the idea that things have changed a lot since SQL-92. It is likewise available free of charge online (modern-sql.com). Markus Winand is regularly speaking at international conferences and user groups.
+++ Have a text similarity tool running on GPUs (by Jörg Bartnick) +++
Everybody knows these situations: You are building a web shop, and suddenly you realise that you didn’t include any machine learning yet! This talk comes to the rescue, we claim that a text similarity computation is machine learning. Then we can use it for all kinds of situations: for example, for having a similar item recommendation for our webshop users. And because we want to ensure that they are always served at the highest speed, we would put it on graphics cards and use the GPU power.
In this talk, Jörg Bartnick will take the artificial example from above to share his experiences with matrix multiplications on GPUs.
Dr Jörg Bartnick is a physicist, a software engineer and he works for trivago.
For his PhD in statistical mechanics, he wrote computer simulations for GPU clusters and predicted experimental results for many-body-systems.
At trivago, he works in the Java backend team on GPU algorithms that help to deliver results a bit faster. He enjoys solving problems with Python, Java, C++ and CUDA.