Full system simulation of many-core heterogeneous SoCs using GPU and QEMU semihosting

Modern system-on-chips are evolving towards complex and heterogeneous platforms with general purpose processors cou- pled with massively parallel manycore accelerator fabrics (e.g. embedded GPUs). Platform developers are looking for ecient full-system simulators capable of simulating com- plex applications, middleware and operating systems on these heterogeneous targets. Unfortunately current virtual plat- forms are not able to tackle the complexity and heteroge- neity of state-of-the-art SoCs. Software emulators, such as the open-source QEMU project, cope quite well in terms of simulation speed and functional accuracy with homoge- neous coarse-grained multi-cores. The main contribution of this paper is the introduction of a novel virtual prototy- ping technique which exploits the heterogeneous accelera- tors available in commodity PCs to tackle the heterogeneity challenge in full-SoC system simulation. In a nutshell, our approach makes it possible to partition simulation between the host CPU and GPU. More specically, QEMU runs on the host CPU and the simulation of manycore accelerators is ooaded, through semi-hosting, to the host GPU. Our experimental results conrm the exibility and eciency of our enhanced QEMU environment.