Prof. Gao and Prof. Giorgi.

EU and USA cooperating on the dataflow execution model

Prof. Gao, one of most renown scientist for dataflow research joins the TERAFLUX project

Giving Publicity

Prof. Valero hosting world class research including TERAFLUX

TERAFLUX at work

Some TERAFLUX scientists at work

TERAFLUX architecture

The TERAFLUX architecture has been developed to enable TERAdevice computing (i.e. 1000 billion transistor on a chip or package) by 2020.

TERAFLUX extensions for HP COTSon simulator

TERAFLUX applications and Legacy applications can run on top of a Linux system: COTSon enables Dataflow Execution through DataFlow-Threads (DF-Threads) and/or Dataflow-Codelets

Double Execution

TERAFLUX can continue to run programs even in presence of faulty cores, nodes, or interconnects -- without requiring "whole system" checkpointing, thanks to the DataFlow execution.


OpenStream enables streaming programming on TERAFLUX. Parallel programs in StarSs/OmpSs can be interfaced to OpenStream too. OpenStream generates DF-Threads and T* Instructions


Look at the Deliverables section: many of them are publicly available and show the outcomes of the TERAFLUX project


The Papers section lists all the scientific work published during the TERAFLUX project

previous pauseresume next

TERAFLUX: Exploiting Dataflow Parallelism in Teradevice Computing

The TERAFLUX “Transformation Hierarchy”

What is TERAFLUX about?

  • Multi-cores and Many-cores
  • Multithreaded Execution
  • Programming Models
  • Reliability


For further details on the Technical Approach, please click: