Search arXivSearch

arXiv · 1606.06483

A Soft Processor Overlay with Tightly-coupled FPGA Accelerator

Abstract

FPGA overlays are commonly implemented as coarse-grained reconfigurable architectures with a goal to improve designers' productivity through balancing flexibility and ease of configuration of the underlying fabric. To truly facilitate full application acceleration, it is often necessary to also include a highly efficient processor that integrates and collaborates with the accelerators while maintaining the benefits of being implemented within the same overlay framework. This paper presents an open-source soft processor that is designed to tightly-couple with FPGA accelerators as part of an overlay framework. RISC-V is chosen as the instruction set for its openness and portability, and the soft processor is designed as a 4-stage pipeline to balance resource consumption and performance when implemented on FPGAs. The processor is generically implemented so as to promote design portability and compatibility across different FPGA platforms. Experimental results show that integrated software-hardware applications using the proposed tightly-coupled architecture achieve comparable performance as hardware-only accelerators while the proposed architecture provides additional run-time flexibility. The processor has been synthesized to both low-end and high-performance FPGA families from different vendors, achieving the highest frequency of 268.67MHz and resource consumption comparable to existing RISC-V designs.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Ho-Cheung Ng, Cheng Liu, Hayden Kwok-Hay So. 2016-06-21. A Soft Processor Overlay with Tightly-coupled FPGA Accelerator. https://arxiv.org/abs/1606.06483

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

Bi-SamplerZ: A Rejection-Aware Cooperative Gaussian Sampling Framework for Falcon Signature Hardware

We present Bi-SamplerZ, a rejection-aware cooperative sampling framework that converts this idle capacity into useful computation. After an asymmetric accept/reject outcome, Bi- SamplerZ latches the completed logical result and dynamically reassigns the released physical datapath to the unfinished target. The two paths then evaluate fresh independent candidates for the same remaining distribution. We show that this post-rejection cooperation increases the assisted-round completion probability without modifying the underlying candidate distribution or Bernoulli acceptance rule, and we state the randomness-allocation conditions required to preserve the joint output distribution of the original pair of logical sampler calls

cs.AR

A Multi-Engine Dataflow for MoE Decoding on Scratchpad-Based Tensor Accelerators

Mixture-of-Experts (MoE) decoding on scratchpad-based tensor accelerators (STA) is dominated by moving expert weights while the compute engines sit idle. This traffic is hard to hide, because the experts are known only after routing, and hard to shrink without losing quality or adding critical-path work. We present CARDAN, which represents each expert-weight matrix as a vector-quantized component plus a shared-basis low-rank component and co-designs this representation with a multi-engine decoding dataflow. The representation separates expert-common from expert-private work, so the dataflow overlaps DMA with computation on several engines. Across five MoE families on AWS Trainium3, CARDAN matches or improves BF16-teacher perplexity across all five models and speeds up batch-one decoding by 1.15-1.31x over AWS dense MoE megakernels, rising to 1.7x at batch size 16.

cs.AR

Dissecting How Die Scaling Breaks GPU Fine-grained Scheduling

Modern GPUs are no longer physically symmetric. Die scaling leads to both manufacturing-driven floorsweeping and cache and memory partitioning. The former creates chip-specific compute topologies, while the latter causes non-uniform memory access. These asymmetries are substantial. Topology-oblivious compute unit allocation can lead to up to 1.33x performance variation, while remote accesses increase HBM latency by up to 67% and nearly double L2 latency. However, these asymmetries are hidden behind the GPU's logical resource abstractions and can vary across chips. We develop lightweight characterization methods to uncover per-chip compute topology and memory affinity. We then use the discovered information to make existing fine-grained scheduling asymmetry-aware, considering not only how many resources are allocated but also which physical resources are assigned. Across full-GPU kernel execution, intra-application multiplexing, and inter-application co-location, asymmetry-aware scheduling improves mainstream kernels by up to 1.22x, multiplexed LLM inference by up to 14.3%, and avoids up to 1.33x performance variation.

cs.AR