Enabling multi-threaded applications on hybrid shared memory manycore architectures

As the number of cores per chip increases, maintaining cache coherence becomes prohibitive for both power and performance. Non Coherent Cache (NCC) architectures do away with hardware-based cache coherence, but become difficult to program. Some existing architectures provide a middle ground by providing some shared memory in the hardware. Specifically, the 48-core Intel Single-chip Cloud Computer (SCC) provides some off-chip (DRAM) shared memory and some on-chip (SRAM) shared memory. We call such architectures Hybrid Shared Memory, or HSM, manycore architectures. However, how to efficiently execute multi-threaded programs on HSM architectures is an open problem. To be able to execute a multi-threaded program correctly on HSM architectures, the compiler must: i) identify all the shared data and map it to the shared memory, and ii) map the frequently accessed shared data to the on-chip shared memory. In this paper, we present a source-to-source translator written using CETUS (Dave et al. [1]) that identifies a conservative superset of all the shared data in a multi-threaded application, and maps it to the off-chip shared memory such that it enables execution on HSM architectures. This improves the performance of our benchmarks by 32x. Following, we identify and map the frequently accessed shared data to the on-chip shared memory. This further improves the performance of our benchmarks by 8x on average.