I'm working at INRIA being financed to work on improving MALT a memory profiler I developped during my first post-doc and maintained since ten years now.
I worked as a research ingineer at INRIA in the AIRSEA team to apply the PSyClone Fortran source-to-source compiler to addapt the CROCO ocean simulation code to GPU and think about a DSL approach. I also passed 2 months at the ECCC (Environment and Climate Change Canada) at Montreal.
I worked for one year at DDN on the RED object storage developement effort.
I was previously working in an R&D effort at ATOS BULL on bust buffers for supercomputer file systems (ATOS BULL Data Management Tool [PDF].
I worked under the supervision of Philippe Couvée which I want to thank for the opportunity to work in his team and for his patience.
I was also working on H2020 european projects:
I was previously a fellow at CERN working on the R&D of the future data acquisition system (DAQ) of LHCb in collaboration with Intel. The challenge of this project is making a DAQ without any hardware triggers which will be a first time for such a large detector.
Removing the first level hardware trigger will imply the capability of handling 40 Tb/s in the DAQ. In order to manage this high throughput we currently study high-speed fabrics coming from the HPC field : Intel Omni-Path and Mellanox InfiniBand. Both run at 100 Gb/s.
This work was done under the supervision of Niko Neufeld which I would thanks a lot for those nice 3 years there with him and all the online team.
The final system will need to scale on 500 servers to fulfil the required bandwidth considering 80 Gb/s per node. After being aggregated on those 500 server, the data will need to be dispatched on 4000 server for triggering decision before reaching long term storage for physics analysis.
As a side project I derived a NUMA profiling tool from MALT during this post-doc, now also open-sourced: NUMAPROF (NUMA Profiler)
I made one year post-doc in the MAQAO team at the Exascale Computing Research Lab. During this post-doc I mostly developped my memory profiling tool MALT (MALoc Tracker)
I made my PhD. from 2010 to 2014 at the CEA working on memory management for supercomputers. I mostly developed a parallel memory allocator for NUMA architectures. I also worked on a kernel patch prototype to avoid the cost of memory clearing on page first touch.
Title: Contribution à l'amélioration des méthodes d'optimisation de la gestion de la mémoire dans le cadre du Calcul Haute Performance.
You can find the documents here:
This work was supervised by Marc Pérache (CEA) and William Jalby (Université de Versailles Saint-Quentin-en-Yvelines), thanks very much to both of them for leading me on this path especially to Marc for the day to day supervision and all the passionate discussions.
I'm mostly interested in:
As side projects I also looking on:
As a computer scientist of course I like programing, but mostly thinking on the way of programing. For this I like working around the given langages:
What you do should be beautiful, code should be poetry, and don't forget the code <-> humans' interactions.
Today we make complex sotware in complex environment. Optimization is already first getting something clean and working. Then optimization should be at the global level, not local (not only). Software engineering is the path.
First optimize your time of debugging, do not try to stroke faster on your keyboard, take time to think !
How do we convert our research in a more environment aware path by consuming less and not more ?
Sadly I work in a domain which has exponential growth and which should certainly a lot to rethink itself.
The earth and its life is far more important than our research, and by far more important than the industry it creates and justify. How to tackle peacefully all together this requirement ?
During my PhD, first postdoc and recent positions I had the chance to teach as assistant :
There is lab life and outside life.. I'm also involved in some extra assciations on other kind of topics:
