An official website of the United States government
Processing in-memory (PIM) promises to unleash unprecedented computing capabilities for high-data-rate applications. Computation using PIM is performed by breaking down computationally expensive operations into in-memory kernels that can be efficiently executed using non-volatile memory. Logic styles such as MAGIC require that each output memory cell is prepared for evaluation before executing the functional logic operation. State-of-the-art synthesis algorithms perform the preparation immediately after memory cells have expired. Unfortunately, this results in columns of cells being prepared greedily, instead of leveraging efficient parallel data preparation instructions. In this paper, we propose the PREP framework that maximizes the opportunities for parallel column preparation using execution sequence optimization. The key idea of the framework is to postpone data preparation instructions until there are no available prepared cells. Next, the accumulated memory cells are prepared in parallel to release the memory for functional evaluations. The framework is capable of exploring a frontier of area-performance solutions. The PREP framework is evaluated using 15 benchmarks from the SuiteSparse library. Compared with state-of-the-art synthesis tools, energy consumption and latency are respectively reduced by 27% and 25%, with no additional cost in crossbar memory.
more »
« less
LOGIC: Logic Synthesis for Digital In-Memory Computing
In-memory processing offers a promising solution for enhancing the performance of data-intensive applications. While analog in-memory computing demonstrates remarkable efficiency, its limited precision is suitable only for approximate computing tasks. In contrast, digital in-memory computing delivers the deterministic precision necessary to accelerate high-assurance applications. Current digital in-memory computing methods typically involve manually breaking down arithmetic operations into in-memory compute kernels. In contrast, traditional digital circuits are synthesized through intricate and automated design workflows. In this article, we introduce a logic synthesis framework called LOGIC, which facilitates the translation of high-level applications into digital in-memory compute kernels that can be executed using non-volatile memory. We propose techniques for decomposing element-wise arithmetic operations into in-memory kernels while minimizing the number of in-memory operations. Additionally, we optimize the sequence of in-memory operations to reduce non-volatile memory utilization. To address the NP-hard execution sequencing optimization problem, we have developed twolook-aheadalgorithms that offer practical solutions. Additionally, we leverage data layout reorganization to efficiently accelerate applications that heavily rely on sparse matrix-vector multiplication operations. Our experimental evaluations demonstrate that our proposed synthesis approach improves the area and latency of fixed-point multiplication by 84% and 20% compared to the state-of-the-art, respectively. Moreover, when applied to scientific computing applications sourced from the SuiteSparse Matrix Collection, our design achieves remarkable improvements in area, latency, and energy efficiency by factors of 4.8×, 2.6×, and 11×, respectively.
more »
« less
- PAR ID:
- 10604069
- Publisher / Repository:
- Association for Computing Machinery (ACM)
- Date Published:
- Journal Name:
- ACM Transactions on Design Automation of Electronic Systems
- Volume:
- 30
- Issue:
- 2
- ISSN:
- 1084-4309
- Format(s):
- Medium: X Size: p. 1-27
- Size(s):
- p. 1-27
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
This paper presents a Ferroelectric FET (FeFET) based processing-in-memory (PIM) architecture to accelerate inference of deep neural networks (DNNs). We propose a digital in-memory vector-matrix multiplication (VMM) engine design utilizing the FeFET crossbar to enables bit-parallel computation and eliminate analog-to-digital conversion in prior mixed-signal PIM designs. A dedicated hierarchical network-on-chip (H-NoC) is developed for input broadcasting and on-the-fly partial results processing, reducing the data transmission volume and latency. Simulations in 28nm CMOS technology show 115x and 6.3x higher computing efficiency (GOPs/W) over desktop GPU (Nvidia GTX 1080Ti) and ReRAM based design, respectively.more » « less
-
null (Ed.)The increasingly central role of speech based human computer interaction necessitates on-device, low-latency, low-power, high-accuracy key word spotting (KWS). State-of-the-art accuracies on speech-related tasks have been achieved by long short-term memory (LSTM) neural network (NN) models. Such models are typically computationally intensive because of their heavy use of Matrix vector multiplication (MVM) operations. Compute-in-Memory (CIM) architectures, while well suited to MVM operations, have not seen widespread adoption for LSTMs. In this paper we adapt resistive random access memory based CIM architectures for KWS using LSTMs. We find that a hybrid system composed of CIM cores and digital cores achieves 90% test accuracy on the google speech data set at the cost of 25 uJ/decision. Our optimized architecture uses 5-bit inputs, and analog weights to produce 6-bit outputs. All digital computation are performed with 8-bit precision leading to a 3.7× improvement in computational efficiency compared to equivalent digital systems at that accuracy.more » « less
-
With the prosperous development of Deep Neural Network (DNNs), numerous Process-In-Memory (PIM) designs have emerged to accelerate DNN models with exceptional throughput and energy-efficiency. PIM accelerators based on Non-Volatile Memory (NVM) or volatile memory offer distinct advantages for computational efficiency and performance. NVM based PIM accelerators, demonstrated success in DNN inference, face limitations in on-device learning due to high write energy, latency, and instability. Conversely, fast volatile memories, like SRAM, offer rapid read/write operations for DNN training, but suffer from significant leakage currents and large memory footprints. In this paper, for the first time, we present a fully-digital sparse processing in hybrid NVM-SRAM design, synergistically combines the strengths of NVM and SRAM, tailored for on-device continual learning. Our designed NVM and SRAM based PIM circuit macros could support both storage and processing of N:M structured sparsity pattern, significantly improving the storage and computing efficiency. Exhaustive experiments demonstrate that our hybrid system effectively reduces area and power consumption while maintaining high accuracy, offering a scalable and versatile solution for on-device continual learning.more » « less
-
Abstract The constant drive to achieve higher performance in deep neural networks (DNNs) has led to the proliferation of very large models. Model training, however, requires intensive computation time and energy. Memristor‐based compute‐in‐memory (CIM) modules can perform vector‐matrix multiplication (VMM) in place and in parallel, and have shown great promises in DNN inference applications. However, CIM‐based model training faces challenges due to non‐linear weight updates, device variations, and low‐precision. In this work, a mixed‐precision training scheme is experimentally implemented to mitigate these effects using a bulk‐switching memristor‐based CIM module. Low‐precision CIM modules are used to accelerate the expensive VMM operations, with high‐precision weight updates accumulated in digital units. Memristor devices are only changed when the accumulated weight update value exceeds a pre‐defined threshold. The proposed scheme is implemented with a system‐onchip of fully integrated analog CIM modules and digital sub‐systems, showing fast convergence of LeNet training to 97.73%. The efficacy of training larger models is evaluated using realistic hardware parameters and verifies that CIM modules can enable efficient mix‐precision DNN training with accuracy comparable to full‐precision software‐trained models. Additionally, models trained on chip are inherently robust to hardware variations, allowing direct mapping to CIM inference chips without additional re‐training.more » « less
