Various hardware accelerators have been developed for energy-efficient and real-time inference of neural networks on edge devices. However, most training is done on high-performance GPUs or servers, and the huge memory and computing costs prevent training neural networks on edge devices. This paper proposes a novel tensor-based training framework, which offers orders-of-magnitude memory reduction in the training process. We propose a novel rank-adaptive tensorized neural network model, and design a hardware-friendly low-precision algorithm to train this model. We present an FPGA accelerator to demonstrate the benefits of this training method on edge devices. Our preliminary FPGA implementation achieves 59× speedup and 123× energy reduction compared to embedded CPU, and 292× memory reduction over a standard full-size training.
more »
« less
Accelerating Low Bit-width Neural Networks at the Edge, PIM or FPGA: A Comparative Study
Deep Neural Network (DNN) acceleration with digital Processing-in-Memory (PIM) platforms at the edge is an actively-explored domain with great potential to not only address memory-wall bottlenecks but to offer orders of performance improvement in comparison to the von-Neumann architecture. On the other side, FPGA-based edge computing has been followed as a potential solution to accelerate compute-intensive workloads. In this work, adopting low-bit-width neural networks, we perform a solid and comparative inference performance analysis of a recent processing-in-SRAM tape-out with a low-resource FPGA board and a high-performance GPU to provide a guideline for the research community. We explore and highlight the key architectural constraints of these edge candidates that impact their overall performance. Our experimental data demonstrate that the processing-in-SRAM can obtain up to ~160x speed-up and up to 228x higher efficiency (img/s/W) compared to the under-test FPGA on the CIFAR-10 dataset.
more »
« less
- Award ID(s):
- 2228028
- PAR ID:
- 10476521
- Publisher / Repository:
- ACM
- Date Published:
- Journal Name:
- GLSVLSI '23: Proceedings of the Great Lakes Symposium on VLSI 2023
- ISBN:
- 9798400701252
- Page Range / eLocation ID:
- 625 to 630
- Format(s):
- Medium: X
- Location:
- Knoxville TN USA
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
null (Ed.)PIM (processing-in-memory) based hardware accelerators have shown great potentials in addressing the computation and memory access intensity of modern CNNs (convolutional neural networks). While adopting NVM (non-volatile memory) helps to further mitigate the storage and energy consumption overhead, adopting quantization, e.g., shift-based quantization, helps to tradeoff the computation overhead and the accuracy loss, integrating both NVM and quantization in hardware accelerators leads to sub-optimal acceleration. In this paper, we exploit the natural shift property of DWM (domain wall memory) to devise DWMAcc, a DWM-based accelerator with asymmetrical storage of weight and input data, to speed up the inference phase of shift-based CNNs. DWMAcc supports flexible shift operations to enable fast processing with low performance and area overhead. We then optimize it with zero-sharing , input-reuse , and weight-share schemes. Our experimental results show that, on average, DWMAcc achieves 16.6× performance improvement and 85.6× energy consumption reduction over a state-of-the-art SRAM based design.more » « less
-
The processing demands of current and emerging applications, such as image/video processing, are increasing due to the deluge of data, generated by mobile and edge devices. This raises challenges for a vast range of computing systems, starting from smart-phones and reaching cloud and data centers. Heterogeneous computing demonstrates its ability as an efficient computing model due to its capability to adapt to various workload requirements. Field programmable gate arrays (FPGAs) provide power and performance benefits and have been used in many application domains from embedded systems to the cloud. In this paper, we used a closely coupled CPU-FPGA heterogeneous system to accelerate a sliding window based image processing algorithm, Canny edge detector. We accelerated Canny using two different implementations: Code partitioned and data partitioned. In the data partitioned implementation, we proposed a weighted round-robin based algorithm that partitions input images and distributes the load between the CPU and the FPGA based on latency. The paper also compares the performance of the proposed accelerators with separate CPU and FPGA implementations. Using our hybrid CPU-FPGA based algorithm, we achieved a speedup of up to 4.8× over a CPU-only and up to 2.1× over a FPGA-only implementations. Moreover, the estimated total energy consumption of our algorithm is more efficient than a CPU-only implementation. Our results show a significant reduction in energy-delay product (EDP) compared to the CPU-only implementation, and comparable EDP results to the FPGA-only implementation.more » « less
-
null (Ed.)Large Convolutional Neural Networks (CNNs) are often pruned and compressed to reduce the amount of parameters and memory requirement. However, the resulting irregularity in the sparse data makes it difficult for FPGA accelerators that contains systolic arrays of Multiply-and-Accumulate (MAC) units, such as Intel’s FPGA-based Deep Learning Accelerator (DLA), to achieve their maximum potential. Moreover, FPGAs with low-bandwidth off-chip memory could not satisfy the memory bandwidth requirement for sparse matrix computation. In this paper, we present 1) a sparse matrix packing technique that condenses sparse inputs and filters before feeding them into the systolic array of MAC units in the Intel DLA, and 2) a customization of the Intel DLA which allows the FPGA to efficiently utilize a high bandwidth memory (HBM2) integrated in the same package. For end-to-end inference with randomly pruned ResNet-50/MobileNet CNN models, our experiments demonstrate 2.7x/3x performance improvement compared to an FPGA with DDR4, 2.2x/2.1x speedup against a server-class Intel SkyLake CPU, and comparable performance with 1.7x/2x power efficiency gain as compared to an NVidia V100 GPU.more » « less
-
We present a generic and programmable Processing-in-SRAM (PSRAM) accelerator chip design based on an 8T-SRAM array to accommodate a complete set of Boolean logic operations (e.g., NOR/NAND/XOR, both 2- and 3-input), majority, and full adder, for the first time, all in a single cycle. PSRAM provides the programmability required for in-memory computing platforms that could be used for various applications such as parallel vector operation, neural networks, and data encryption. The prototype design is implemented in a SRAM macro with size of 16 kb, demonstrating one of the fastest programmable in-memory computing system to date operating at 1.23 GHz. The 65nm prototype chip achieves system-level peak throughput of 1.2 TOPS, and energy-efficiency of 34.98 TOPS/W at 1.2V.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.1145/3583781.3590213
