The cTuning foundation is a non-profit research and development organization. It is the outcome of Grigori Fursin's initiative started in 2006 to build faster, smaller, more power efficient and reliable computer systems via open repository of optimization knowledge, artifact sharing, reproducible experimentation, universal and multi-objective autotuning and crowd-tuning (experiment crowdsourcing), machine learning and run-time adaptation. We actively collaborate with major international universities, companies, ACM and dividiti.
The cTuning foundation is developing an open source knowledge management framework and web-based repository to enable collaborative and reproducible experimentation in computer engineering while exposing it to powerful predictive analytics (statistical analysis, machine learning, detection of missing features, improvement of models) and collective intelligence. The latest version of our technology (Collective Knowledge or CK) is available at GitHub, in Google Play Store and in Debian distribution. You can see latest results from various experiment crowdsourcing (such as GCC/LLVM crowd-tuning) in CK live repository. You can see an example of a CK-based interactive article here (OpenCL crowd-tuning and machine-learning based run-time adaptation). Here are a few papers describing our approach: [ DATE'16, CPC'15, JSC'14, TRUST'14@PLDI'14, GCC Summit'09, IJPP'11, TACO'10, PLDI'10, SMART'09, HiPEAC'09 ]
Our tools and techniques have been successfully used in multiple academic and industrial projects helping the community unify, systematize, standardize and accelerate their previously ad-hoc, complex, time-consuming and error-prone process of benchmarking, autotuning (optimization) and co-design of computer systems (software and hardware). Enabling faster, smaller, cheaper, more energy efficient and reliable computer systems help, in turn, boost innovation in science and technology!
For example, our CK-based, universal, customizable and multi-objective auto-tuner can help gain up to 10x performance speedups and 30% energy savings on various OpenCL/CUDA/CPU libraries without sacrificing accuracy (or up to 20x speed-ups with a small accuracy degradation) as shown in a CK-powered interactive graph below for a popular SLAM algorithm (all points can be reproduced/validated on practically any platform using CK):
Since the beginning of his own research in 1997, Grigori Fursin realized that it will not be possible to advance without sharing artifacts, reproducing empirical experiments, validating others' results and enabling fair comparison of techniques. Therefore, he spent a considerable effort sharing all artifacts from his own research and promoting new publication model where results are validated by the community. cTuning foundation continued this effort and recently successfully arranged the first workshop in computer engineering where publications and results have been validated by the community (see ADAPT workshop). Furthermore, together with other colleagues, we managed to persuade several major conferences in computer engineering to start artifact evaluation. We are now a part of the ACM taskforce on reproducible research and experimentation! We also actively participate in various international research projects developing novel techniques to improve machine learning, data mining, knowledge discovery, statistical analysis, feature detection, experiment crowdsourcing and so on.