Tailoring instruction-set extensions for an ultra-low power tightly-coupled cluster of OpenRISC cores

Abstract

Baseline RISC instruction sets for ultra-low power processors are constantly being tuned to reduce cycle count when executing computation-intensive applications. Performance improvements often come at a non-negligible price in terms of area and critical path length and imply deeper pipelines and complex memory interfaces. This penalizes control-intensive code execution and significantly increases cost and complexity of building multi-core clusters. In addition, some extensions are not easily exploited by compilers and may increase code development effort, especially when considering parallel applications. In this paper we describe our efforts in enhancing a baseline open ISA (OpenRISC) and its LLVM compiler back-end to significantly reduce execution cycles while minimizing the impact on core micro-architecture complexity, number of pipeline stages, area and power. In addition, we improved the core micro-architecture to streamline its integration in a tightly-coupled cluster, sharing instruction cache and data memory, thereby further enhancing parallel execution efficiency. The combined effect of ISA, compiler and micro-architecture evolution gives an average energy efficiency boost of 59% on vector intensive code and 41% otherwise, at an area and power increase of 2.3% and 18% on a four-core processor cluster.

The Quest for Energy-Efficient I$ Design in Ultra-Low-Power Clustered Many-Cores

“…The PULP cluster (see Figure 3), that we consider as a baseline for the proposed exploration, includes a configurable number of processing elements (PEs), based on OpenRISC OR10N cores [34] [35], each featuring a private instruction cache. The OR10N cores are based on an inorder, single-issue, four stage pipeline micro-architecture without branch prediction, improved with extensions for higher throughput and energy efficiency in parallel signal processing workloads [34]. No data caches are present, therefore avoiding memory coherency overhead and additional area penalties [30].…”

A Heterogeneous Multicore System on Chip for Energy Efficient Brain Inspired Computing

“…The proposed SoC implements the third generation PULP (Parallel Ultra-Low-Power) platform 1 extended with a dedicated accelerator for convolution intensive processing [16]. The programmable computing engine of the SoC is based on a tightly coupled cluster of 4 OpenRISC ISA cores called OR10N enhanced for energy efficient digital signal processing [17]. The cluster features a shared 4kB latch-based Standard Cell Memory (SCM) [18] instruction cache that, coupled with a private per-core L0 buffer, increases energy efficiency by 30% with respect to an SRAM-based private cache architecture [15].…”

A sub-10mW real-time implementation for EMG hand gesture recognition based on a multi-core biomedical SoC

“…1) MCU: The employed MCU is an implementation derived from the PULP (Parallel-Ultra-Low-Power) Platform [23], [24]. It comes with 4 general-purpose openRISC cores sharing the level 1 (L1) memory -tightly coupled data memory (TCDM).…”