Section: New Results

Run-time environments for multicore architectures

In the recent past, we acquired a solid expertise in multicore systems through the PhD of Jean-Pierre Lozi [60] and Florian David [38]. This expertise has led us to initiate several collaborations with industry partners, in the form of CIFRE PhD support. We first targeted real-time multicore systems with the goal of improving resource usage, through a cooperation with Renault and the PhD of Antoine Blin. Recently, we have started another cooperation on multicore real-time systems for avionics and space with Thales TRT, that is the topic of the PhD of Cédric Courtaud.

The PhD of Jean-Pierre Lozi [60] was on improving the performance locks on large multicore architectures. In an paper published at Usenix ATC 2012 [58], and more recently in an article published in 2016 in ACM Transactions on Computer Systems (TOCS) [10], we proposed a new locking technique, Remote Core Locking (RCL), that aims to accelerate the execution of critical sections in legacy applications on multicore architectures. RCL is currently one of the most efficient locking technique and the ATC 2012 paper has currently 67 citations on Google scholar. The idea of RCL is to replace lock acquisitions by optimized remote procedure calls to a dedicated server hardware thread. RCL limits the performance collapse observed with other lock algorithms when many threads try to acquire a lock concurrently and removes the need to transfer lock-protected shared data to the hardware thread acquiring the lock because such data can typically remain in the server's cache. Eighteen applications were used to evaluate RCL from standard multicore benchmark suites, such as SPLASH-2 and Phoenix 2. By using RCL instead of Linux POSIX locks, performance is improved by up to 2.5 times on Memcached, and up to 11.6 times on Berkeley DB with the TPC-C client. On a SPARC machine with two Sun Ultrasparc T2+ processors and 128 hardware threads, performance is improved by up to 1.3 times with respect to Solaris POSIX locks on Memcached, and up to 7.9 times on Berkeley DB with the TPC-C client.

The PhD of Antoine Blin is on modern complex embedded systems that involve a mix of real-time and best-effort applications. The recent emergence of low-cost multicore processors raises the possibility of running both kinds of applications on a single machine, with virtualization ensuring isolation. Nevertheless, memory contention can introduce other sources of delay, that can lead to missed deadlines. We first investigated the source of memory contention for the Mibench benchmark in a paper published at NETYS 2016 [12]. Then, in a paper published at ECRTS 2016 [11], we present a combined offline/online memory bandwidth monitoring approach. Our approach estimates and limits the impact of the memory contention incurred by the best-effort applications on the execution time of the real-time application. Using our approach, the system designer can limit the overhead on the real-time application to under 5% of its expected execution time, while still enabling progress of the best-effort applications.