Making computer engineering a science;
Systematizing program and system analysis and optimization using auto-tuning, machine learning and social networking.
If you are interested in preview, tutorial or presentation about cTuning/Collective Mind crowd-tuning and machine learning technology, its current industrial usages or new publication model, do not hesitate to get in touch with Grigori Fursin!
cTuning-related recent and upcoming events and news:
- 2013.06.05: Grigori will give a keynote about cTuning/Collective Mind technology at iWAPT'13 @ ICCS'13
- 2013.05.02: cTuning/cMind thematic session on "making computer engineering a science" was held at HiPEAC computing systems week/ACM ECRC in Paris - thanks to your feedback we plan to continue this collaborative activity!
- 2013.03.27: Grigori gave a keynote on statistical crowdsourcing of auto-tuning and tutorial on Collective Mind Framework at the Conference on Advanced Topics and Auto Tuning in High Performance Scientific Computing in NTU, Taiwan
- 2013.02.27: Grigori was an external PhD examiner at Politecnico di Milano for PhD students: Michele Tartara, Ettore Speziale, Simone Corbetta and Paolo Grassi
- 2013.01.22: We organized ADAPT workshop at HiPEAC 2013 sponsored by NVidia and Microsoft Research
cTuning Virtual Lab
cTuning mission
The main aim of cTuning virtual lab is to promote collaborative and reproducible research in computer engineering to build efficient, adaptive, self-tuning computer systems.
Our original research on semiconductor neural networks for future adaptive, neural computers has been considerably slowed down in 1993-1996 due to slow and inefficient computer systems. It forced us to switch to computer engineering to understand how to build fast, reliable and power-efficient computer systems and to apply our background in physics and AI to systematize and automate design and optimization of computer systems.
We decided to open cTuning public center and virtual lab to collaboratively develop tools, repositories, models and share data and interfaces to gradually learn how to build efficient computers that should allow us to get back to original research on neural computers one day.
Since 1997, we have been involved in multiple collaborative projects while releasing our open-source tools and data from our publications to the community. We are advocating for a new publication model that should favor reproducible research and sharing of data, models, tools and interfaces.
We are grateful to all our collaborators for interesting discussions, support and feedback!
cTuning concept and coordination (2005-cur.): Grigori Fursin