Project description
Recalculating AI matrix multiplication
AI tasks often involve processing large amounts of data, which are represented as matrices. Multiplying these matrices consumes both time and energy because the computer needs to perform numerous calculations to multiply each element of the matrices together. As the size of the matrices increases, so does the amount of computing power and energy required. Despite industry efforts to find a solution, the current cubic-time algorithms remain inefficient. In this context, the ERC-funded FMMF-AI project aims to revolutionise matrix multiplication. The project’s methods promise significant speedups in both software and hardware, thus addressing long-standing inefficiencies. It aims to seize promising business opportunities to make AI computation faster and more energy-efficient.
Objective
Matrix multiplication consumes huge amount of resources: computing time and energy, primarily in AI applications. The industry has recognized the need for faster and more energy-efficient matrix multiplication with state-of-the-art solutions in software (e.g. DGEMM of Intel's math kernel library (MKL) for CPU and NVIDIA's CUDA for GPU) and hardware (e.g. Google's TPU and Intel / Habana labs Gaudi accelerator). Unfortunately, all present solutions employ a wasteful cubic-time algorithm. We have developed methods that provide speedup for matrix multiplication in SW and in HW. The novel developments of Prof. Oded Schwartz and his strong team are based on years of research, and are protected by several patents. The funds are requested to pursue business opportunity.
Fields of science (EuroSciVoc)
CORDIS classifies projects with EuroSciVoc, a multilingual taxonomy of fields of science, through a semi-automatic process based on NLP techniques.
CORDIS classifies projects with EuroSciVoc, a multilingual taxonomy of fields of science, through a semi-automatic process based on NLP techniques.
- natural sciencescomputer and information sciencessoftware
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringcomputer hardwarecomputer processors
You need to log in or register to use this function
Programme(s)
- HORIZON.1.1 - European Research Council (ERC) Main Programme
Funding Scheme
HORIZON-ERC-POC - HORIZON ERC Proof of Concept GrantsHost institution
91904 Jerusalem
Israel