Building Heterogeneous Unified Virtual Memories (UVMs) without the Overhead

Konstantinos Koukos, Alberto Ros, Erik Hagersten, Stefanos Kaxiras
2016 ACM Transactions on Architecture and Code Optimization (TACO)  
This work proposes a novel scheme to facilitate heterogeneous systems with unified virtual memory. Research proposals, implement coherence protocols for sequential consistency (SC) between CPU cores, and between devices. Such mechanisms introduce severe bottlenecks in the system; therefore, we adopt the heterogeneous-race-free (HRF) memory model. The use of HRF simplifies the coherency protocol and the GPU memory management unit (MMU). Our protocol optimizes CPU and GPU demands separately, with
more » ... the GPU part being simpler while the CPU is more elaborate and latency-aware. We achieve an average 45% speedup and 45% energy-delay product reduction (20% energy) over the corresponding SC implementation.
doi:10.1145/2889488 fatcat:cx5535ifhfgnxe3yocrc6h77sq