An analysis and simulation tool of real-time communications in on-chip networks

Abstract

This paper presents Real-Time Network-on-chip-based architecture Analysis and Simulation tool (ReTiNAS), with a special focus on real-time communications. It allows fast and precise exploration of real-time design choices onto NoC architectures. ReTiNAS is an event-based simulator written in Python. It implements different real-time communication protocols and tracks the communications within the NoC at cycle level. Its modularity allows activating and deactivating different NoC components and easily extending the implemented protocols for more customized simulations and analysis. Further, we use ReTiNAS to perform a comparative study of analysis and simulation for different communication protocols using a wide set of synthetic experiments.

Task and Communication Allocation for Real-time Tasks to Networks-on-Chip Multiprocessors

“…Hence, DAG and intersub-tasks communications are generated by TGFF [7]. In the architecture side, we map these tasks on a 3 × 3 2D-Mesh NoC with routers, containing for each of them 6 V C with the following TDMA slot configuration : [4,2,3,5,3,3]. To avoid untractable hyper-periods, the period of every task are randomly generated from a list of values between the interval of 1000 and 10000.…”

“…The first uses TDMA (Time-division multiple access) to regulate the medium communication access, while the second assign a priority to each message, and a scheduling policy is applied. A comparative study of these techniques are reported in [4] using simulation and analysis. [10], [13] proposed techniques for TDMA quantum assignment to minimize communication latency, i.e find a near-optimal VC slot assignment using heuristics.…”

“…Mainly they can be classified to: (i) priority-based and TDMA-based. We have reported an experimental and analytical comparison between communication protocols for real-time systems onto NoCs in [4].…”

Memory-processor co-scheduling for real-time tasks on network-on-chip manycore architectures

“…The first category concerns the use of TDMA (Time-division multiple access) to share the communication medium among communication flows by time quantum reservation, whereas the second category consist of assigning priority to flows, and therefore, the arbitration follows the priority map [10]. A comparative study of both strategies is reported in [1] through the simulation and analysis of several scenarios. Moreover, [11], [12] proposed heuristics to find the optimal TDMA quantum assignment for a minimum communication latency.…”

“…Several works have proposed different techniques to allocate real-time workload on many-core architecture, while a lot of others have focused on estimating the worst-case communication time (latency) under different network topology and NoC-design assumptions. Although the proposed solutions are efficient and very interesting, they provide poor performances when combined [1]. In this work, we tackle both task-to-core allocation and inter-task communication at the same time, to avoid increasing the pessimism that can be accounted in the two steps.…”