Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userPerformance evaluation over HW/SW co-design SoC memory transfers for a CNN accelerator
75
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
Many FPGAs vendors have recently included embedded processors in their devices, like Xilinx with ARM-Cortex A cores, together with programmable logic cells. These devices are known as Programmable System on Chip (PSoC). Their ARM cores (embedded in the processing system or PS) communicates with the programmable logic cells (PL) using ARM-standard AXI buses. In this paper we analyses the performance of exhaustive data transfers between PS and PL for a Xilinx Zynq FPGA in a co-design real scenario for Convolutional Neural Networks (CNN) accelerator, which processes, in dedicated hardware, a stream of visual information from a neuromorphic visual sensor for classification. In the PS side, a Linux operating system is running, which recollects visual events from the neuromorphic sensor into a normalized frame, and then it transfers these frames to the accelerator of multi-layered CNNs, and read results, using an AXI-DMA bus in a per-layer way. As these kind of accelerators try to process information as quick as possible, data bandwidth becomes critical and maintaining a good balanced data throughput rate requires some considerations. We present and evaluate several data partitioning techniques to improve the balance between RX and TX transfer and two different ways of transfers management: through a polling routine at the userlevel of the OS, and through a dedicated interrupt-based kernellevel driver. We demonstrate that for longer enough packets, the kernel-level driver solution gets better timing in computing a CNN classification example. Main advantage of using kernel-level driver is to have safer solutions and to have tasks scheduling in the OS to manage other important processes for our application, like frames collection from sensors and their normalization.
rate research
Read More
To meet the extreme compute demands for deep learning across commercial and scientific applications, dataflow accelerators are becoming increasingly popular. While these domain-specific accelerators are not fully programmable like CPUs and GPUs, they retain varying levels of flexibility with respect to data orchestration, i.e., dataflow and tiling optimizations to enhance efficiency. There are several challenges when designing new algorithms and mapping approaches to execute the algorithms for a target problem on new hardware. Previous works have addressed these challenges individually. To address this challenge as a whole, in this work, we present a HW-SW co-design ecosystem for spatial accelerators called Union within the popular MLIR compiler infrastructure. Our framework allows exploring different algorithms and their mappings on several accelerator cost models. Union also includes a plug-and-play library of accelerator cost models and mappers which can easily be extended. The algorithms and accelerator cost models are connected via a novel mapping abstraction that captures the map space of spatial accelerators which can be systematically pruned based on constraints from the hardware, workload, and mapper. We demonstrate the value of Union for the community with several case studies which examine offloading different tensor operations(CONV/GEMM/Tensor Contraction) on diverse accelerator architectures using different mapping schemes.
Hardware/Software (HW/SW) co-designed processors provide a promising solution to the power and complexity problems of the modern microprocessors by keeping their hardware simple. Moreover, they employ several runtime optimizations to improve the performance. One of the most potent optimizations, vectorization, has been utilized by modern microprocessors, to exploit the data level parallelism through SIMD accelerators. Due to their hardware simplicity, these accelerators have evolved in terms of width from 64-bit vectors in Intel MMX to 512-bit wide vector units in Intel Xeon Phi and AVX-512. Although SIMD accelerators are simple in terms of hardware design, code generation for them has always been a challenge. Moreover, increasing vector lengths with each new generation add to this complexity. This paper explores the scalability of SIMD accelerators from the code generation point of view. We discover that the SIMD accelerators remain underutilized at higher vector lengths mainly due to: a) reduced dynamic instruction stream coverage for vectorization and b) increase in permutations. Both of these factors can be attributed to the rigidness of the SIMD architecture. We propose a novel SIMD architecture that possesses the flexibility needed to support higher vector lengths. Furthermore, we propose Variable Length Vectorization and Selective Writing in a HW/SW co-designed environment to transparently target the flexibility of the proposed architecture. We evaluate our proposals using a set of SPECFP2006 and Physicsbench applications. Our experimental results show an average dynamic instruction reduction of 31% and 40% and an average speed up of 13% and 10% for SPECFP2006 and Physicsbench respectively, for 512-bit vector length, over the scalar baseline code.
We designed a device for Convolution Neural Network applications with non-volatile MRAM memory and computing-in-memory co-designed architecture. It has been successfully fabricated using 22nm technology node CMOS Si process. More than 40MB MRAM density with 9.9TOPS/W are provided. It enables multiple models within one single chip for mobile and IoT device applications.
CUDA Unified Memory improves the GPU programmability and also enables GPU memory oversubscription. Recently, two advanced memory features, memory advises and asynchronous prefetch, have been introduced. In this work, we evaluate the new features on two platforms that feature different CPUs, GPUs, and interconnects. We derive a benchmark suite for the experiments and stress the memory system to evaluate both in-memory and oversubscription performance. The results show that memory advises on the Intel-Volta/Pascal-PCIe platform bring negligible improvement for in-memory executions. However, when GPU memory is oversubscribed by about 50%, using memory advises results in up to 25% performance improvement compared to the basic CUDA Unified Memory. In contrast, the Power9-Volta-NVLink platform can substantially benefit from memory advises, achieving up to 34% performance gain for in-memory executions. However, when GPU memory is oversubscribed on this platform, using memory advises increases GPU page faults and results in considerable performance loss. The CUDA prefetch also shows different performance impact on the two platforms. It improves performance by up to 50% on the Intel-Volta/Pascal-PCI-E platform but brings little benefit to the Power9-Volta-NVLink platform.
End-to-end performance estimation and measurement of deep neural network (DNN) systems become more important with increasing complexity of DNN systems consisting of hardware and software components. The methodology proposed in this paper aims at a reduced turn-around time for evaluating different design choices of hardware and software components of DNN systems. This reduction is achieved by moving the performance estimation from the implementation phase to the concept phase by employing virtual hardware models instead of gathering measurement results from physical prototypes. Deep learning compilers introduce hardware-specific transformations and are, therefore, considered a part of the design flow of virtual system models to extract end-to-end performance estimations. To validate the run-time accuracy of the proposed methodology, a system processing the DilatedVGG DNN is realized both as virtual system model and as hardware implementation. The results show that up to 92 % accuracy can be reached in predicting the processing time of the DNN inference.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
