ヴァンサン・ジョルダンのホームページVincent Jordan's homepage 
page personnelle de Vincent Jordan ![french flag](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="18" height="12"><rect fill-opacity="0"/></svg>)
ProfilePhD student at KDE laboratory (more about this lab.).
My research topic is "XML query processing using GPU". See ![pdf icon](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="16" height="16"><rect fill-opacity="0"/></svg>)
Presentation for entrance exam (2011.02.02) and Research proposal.
I belong to:
Research topic
eXtensible Markup Language (XML) is a both human- and machine-readable language widely used in computer world for the transmission and the storage of data.
This simplification of the SGML language was designed to ease and spread the usage of an interoperable language over the internet.
This aim has been reached nowadays since the XML language is used by a huge amount of applications and most programmation languages feature an XML parser.
XML Path (XPath) language has been created to query XML data. Consequently of the broad XML usage, the need to locate specific data in XML became high and
revealed a trade off: Keep data in convenient XML format with slow querying or convert data into more efficient database format with faster querying?
The biggest is the amount a XML data, the most time-consuming is the conversion option. My research is about finding a way to speedup XPath queries on big XML data.
Graphic Processing Unit (GPU) is a processor designed from the beginning for efficient and massive parallel execution. According to the increase of graphic rendering complexity,
these chips gained a more general purpose design. GPGPU accronym refers to these new GPU architectures and stands for General Purpose GPU.
My current strategy is to execute on GPU, several instances of the same query processing algorithm on different data partitions. This is data parallelism
(opposed to task parallelism where one algorithm processes several data in the same time). Althrough this hardware is called "general purpose", several limitations have to be overcome.
The practical target of this research is to evaluate the opportunity of using GPU as XML query coprocessor for software dealing with big amount of XML data.
In comparison with their parallel processing power, GPU are cheaper than most dedicated chips.
Making good use of GPU hardware, because it is still very graphic computing oriented, is a great challenge:
To learn more about this topic, see
research project homepage,
related links,
HTML master thesis or
PDF master thesis and
PDF presentation (English, 2010.07.20)
PDF presentation (French, 2010.12.14).
Curriculum Vitæ 
Contact (for PhD) (during 6-month internship)
![map icon](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="12" height="16"><rect fill-opacity="0"/></svg>)
show on