Team Moscova

Overall Objectives
Scientific Foundations
Application Domains
New Results
Other Grants and Activities

Section: Software


Participants : Jade Alglave, Luc Maranget, Susmit Sekar [ U. of Cambridge ] .

The tools suite of the Weak Memory Models (cf. the relevant section) comprises now three items:


(Ocaml, 25000 locs [lines of code]) The model checker. It has been rewritten for flexibility and now integrates about 10 different models for three architectures (X86, Power and ARM).


(Ocaml, 4800 locs + C, 500 locs) It takes the same input as memevents and produces executables that exercices the memory model of X86 and Power machines. This year, we gained access to 2 supercomputers, which lead us to add a cross compilation feature to litmus , and to consider running several instances of the same test concurrently. The throughput of test outcomes have been multiplied by a factor of about 100, allowing us to consider running 1000 tests 109 times in a reasonable time.


(Ocaml, 3500 locs) It is s a new tool, developed by J. Alglave and L. Maranget. diy produces inputs for litmus and memevents from concise specification.

The combination of access to supercomputers hosted at IDRIS (CNRS) and HPCX center (UK), improvements of litmus and the design of diy allows us to tackle in-deep testing of the Power architecture machines.

L. Maranget is now the main developer of this tools suite.


Logo Inria