Matching Items (16)
Filtering by
- All Subjects: Computer Engineering
- All Subjects: Reliability
- Creators: Cao, Yu
Description
Memories play an integral role in today's advanced ICs. Technology scaling has enabled high density designs at the price paid for impact due to variability and reliability. It is imperative to have accurate methods to measure and extract the variability in the SRAM cell to produce accurate reliability projections for future technologies. This work presents a novel test measurement and extraction technique which is non-invasive to the actual operation of the SRAM memory array. The salient features of this work include i) A single ended SRAM test structure with no disturbance to SRAM operations ii) a convenient test procedure that only requires quasi-static control of external voltages iii) non-iterative method that extracts the VTH variation of each transistor from eight independent switch point measurements. With the present day technology scaling, in addition to the variability with the process, there is also the impact of other aging mechanisms which become dominant. The various aging mechanisms like Negative Bias Temperature Instability (NBTI), Channel Hot Carrier (CHC) and Time Dependent Dielectric Breakdown (TDDB) are critical in the present day nano-scale technology nodes. In this work, we focus on the impact of NBTI due to aging in the SRAM cell and have used Trapping/De-Trapping theory based log(t) model to explain the shift in threshold voltage VTH. The aging section focuses on the following i) Impact of Statistical aging in PMOS device due to NBTI dominates the temporal shift of SRAM cell ii) Besides static variations , shifting in VTH demands increased guard-banding margins in design stage iii) Aging statistics remain constant during the shift, presenting a secondary effect in aging prediction. iv) We have investigated to see if the aging mechanism can be used as a compensation technique to reduce mismatch due to process variations. Finally, the entire test setup has been tested in SPICE and also validated with silicon and the results are presented. The method also facilitates the study of design metrics such as static, read and write noise margins and also the data retention voltage and thus help designers to improve the cell stability of SRAM.
ContributorsRavi, Venkatesa (Author) / Cao, Yu (Thesis advisor) / Bakkaloglu, Bertan (Committee member) / Clark, Lawrence (Committee member) / Arizona State University (Publisher)
Created2013
Description
Non-volatile memories (NVM) are widely used in modern electronic devices due to their non-volatility, low static power consumption and high storage density. While Flash memories are the dominant NVM technology, resistive memories such as phase change access memory (PRAM) and spin torque transfer random access memory (STT-MRAM) are gaining ground. All these technologies suffer from reliability degradation due to process variations, structural limits and material property shift. To address the reliability concerns of these NVM technologies, multi-level low cost solutions are proposed for each of them. My approach consists of first building a comprehensive error model. Next the error characteristics are exploited to develop low cost multi-level strategies to compensate for the errors. For instance, for NAND Flash memory, I first characterize errors due to threshold voltage variations as a function of the number of program/erase cycles. Next a flexible product code is designed to migrate to a stronger ECC scheme as program/erase cycles increases. An adaptive data refresh scheme is also proposed to improve memory reliability with low energy cost for applications with different data update frequencies. For PRAM, soft errors and hard errors models are built based on shifts in the resistance distributions. Next I developed a multi-level error control approach involving bit interleaving and subblock flipping at the architecture level, threshold resistance tuning at the circuit level and programming current profile tuning at the device level. This approach helped reduce the error rate significantly so that it was now sufficient to use a low cost ECC scheme to satisfy the memory reliability constraint. I also studied the reliability of a PRAM+DRAM hybrid memory system and analyzed the tradeoffs between memory performance, programming energy and lifetime. For STT-MRAM, I first developed an error model based on process variations. I developed a multi-level approach to reduce the error rates that consisted of increasing the W/L ratio of the access transistor, increasing the voltage difference across the memory cell and adjusting the current profile during write operation. This approach enabled use of a low cost BCH based ECC scheme to achieve very low block failure rates.
ContributorsYang, Chengen (Author) / Chakrabarti, Chaitali (Thesis advisor) / Cao, Yu (Committee member) / Ogras, Umit Y. (Committee member) / Bakkaloglu, Bertan (Committee member) / Arizona State University (Publisher)
Created2014
Description
Coarse Grain Reconfigurable Arrays (CGRAs) are promising accelerators capable of
achieving high performance at low power consumption. While CGRAs can efficiently
accelerate loop kernels, accelerating loops with control flow (loops with if-then-else
structures) is quite challenging. Techniques that handle control flow execution in
CGRAs generally use predication. Such techniques execute both branches of an
if-then-else structure and select outcome of either branch to commit based on the
result of the conditional. This results in poor utilization of CGRA s computational
resources. Dual-issue scheme which is the state of the art technique for control flow
fetches instructions from both paths of the branch and selects one to execute at
runtime based on the result of the conditional. This technique has an overhead in
instruction fetch bandwidth. In this thesis, to improve performance of control flow
execution in CGRAs, I propose a solution in which the result of the conditional
expression that decides the branch outcome is communicated to the instruction fetch
unit to selectively issue instructions from the path taken by the branch at run time.
Experimental results show that my solution can achieve 34.6% better performance
and 52.1% improvement in energy efficiency on an average compared to state of the
art dual issue scheme without imposing any overhead in instruction fetch bandwidth.
achieving high performance at low power consumption. While CGRAs can efficiently
accelerate loop kernels, accelerating loops with control flow (loops with if-then-else
structures) is quite challenging. Techniques that handle control flow execution in
CGRAs generally use predication. Such techniques execute both branches of an
if-then-else structure and select outcome of either branch to commit based on the
result of the conditional. This results in poor utilization of CGRA s computational
resources. Dual-issue scheme which is the state of the art technique for control flow
fetches instructions from both paths of the branch and selects one to execute at
runtime based on the result of the conditional. This technique has an overhead in
instruction fetch bandwidth. In this thesis, to improve performance of control flow
execution in CGRAs, I propose a solution in which the result of the conditional
expression that decides the branch outcome is communicated to the instruction fetch
unit to selectively issue instructions from the path taken by the branch at run time.
Experimental results show that my solution can achieve 34.6% better performance
and 52.1% improvement in energy efficiency on an average compared to state of the
art dual issue scheme without imposing any overhead in instruction fetch bandwidth.
ContributorsRajendran Radhika, Shri Hari (Author) / Shrivastava, Aviral (Thesis advisor) / Christen, Jennifer Blain (Committee member) / Cao, Yu (Committee member) / Arizona State University (Publisher)
Created2014
Description
Static CMOS logic has remained the dominant design style of digital systems for
more than four decades due to its robustness and near zero standby current. Static
CMOS logic circuits consist of a network of combinational logic cells and clocked sequential
elements, such as latches and flip-flops that are used for sequencing computations
over time. The majority of the digital design techniques to reduce power, area, and
leakage over the past four decades have focused almost entirely on optimizing the
combinational logic. This work explores alternate architectures for the flip-flops for
improving the overall circuit performance, power and area. It consists of three main
sections.
First, is the design of a multi-input configurable flip-flop structure with embedded
logic. A conventional D-type flip-flop may be viewed as realizing an identity function,
in which the output is simply the value of the input sampled at the clock edge. In
contrast, the proposed multi-input flip-flop, named PNAND, can be configured to
realize one of a family of Boolean functions called threshold functions. In essence,
the PNAND is a circuit implementation of the well-known binary perceptron. Unlike
other reconfigurable circuits, a PNAND can be configured by simply changing the
assignment of signals to its inputs. Using a standard cell library of such gates, a technology
mapping algorithm can be applied to transform a given netlist into one with
an optimal mixture of conventional logic gates and threshold gates. This approach
was used to fabricate a 32-bit Wallace Tree multiplier and a 32-bit booth multiplier
in 65nm LP technology. Simulation and chip measurements show more than 30%
improvement in dynamic power and more than 20% reduction in core area.
The functional yield of the PNAND reduces with geometry and voltage scaling.
The second part of this research investigates the use of two mechanisms to improve
the robustness of the PNAND circuit architecture. One is the use of forward and reverse body biases to change the device threshold and the other is the use of RRAM
devices for low voltage operation.
The third part of this research focused on the design of flip-flops with non-volatile
storage. Spin-transfer torque magnetic tunnel junctions (STT-MTJ) are integrated
with both conventional D-flipflop and the PNAND circuits to implement non-volatile
logic (NVL). These non-volatile storage enhanced flip-flops are able to save the state of
system locally when a power interruption occurs. However, manufacturing variations
in the STT-MTJs and in the CMOS transistors significantly reduce the yield, leading
to an overly pessimistic design and consequently, higher energy consumption. A
detailed analysis of the design trade-offs in the driver circuitry for performing backup
and restore, and a novel method to design the energy optimal driver for a given yield is
presented. Efficient designs of two nonvolatile flip-flop (NVFF) circuits are presented,
in which the backup time is determined on a per-chip basis, resulting in minimizing
the energy wastage and satisfying the yield constraint. To achieve a yield of 98%,
the conventional approach would have to expend nearly 5X more energy than the
minimum required, whereas the proposed tunable approach expends only 26% more
energy than the minimum. A non-volatile threshold gate architecture NV-TLFF are
designed with the same backup and restore circuitry in 65nm technology. The embedded
logic in NV-TLFF compensates performance overhead of NVL. This leads to the
possibility of zero-overhead non-volatile datapath circuits. An 8-bit multiply-and-
accumulate (MAC) unit is designed to demonstrate the performance benefits of the
proposed architecture. Based on the results of HSPICE simulations, the MAC circuit
with the proposed NV-TLFF cells is shown to consume at least 20% less power and
area as compared to the circuit designed with conventional DFFs, without sacrificing
any performance.
more than four decades due to its robustness and near zero standby current. Static
CMOS logic circuits consist of a network of combinational logic cells and clocked sequential
elements, such as latches and flip-flops that are used for sequencing computations
over time. The majority of the digital design techniques to reduce power, area, and
leakage over the past four decades have focused almost entirely on optimizing the
combinational logic. This work explores alternate architectures for the flip-flops for
improving the overall circuit performance, power and area. It consists of three main
sections.
First, is the design of a multi-input configurable flip-flop structure with embedded
logic. A conventional D-type flip-flop may be viewed as realizing an identity function,
in which the output is simply the value of the input sampled at the clock edge. In
contrast, the proposed multi-input flip-flop, named PNAND, can be configured to
realize one of a family of Boolean functions called threshold functions. In essence,
the PNAND is a circuit implementation of the well-known binary perceptron. Unlike
other reconfigurable circuits, a PNAND can be configured by simply changing the
assignment of signals to its inputs. Using a standard cell library of such gates, a technology
mapping algorithm can be applied to transform a given netlist into one with
an optimal mixture of conventional logic gates and threshold gates. This approach
was used to fabricate a 32-bit Wallace Tree multiplier and a 32-bit booth multiplier
in 65nm LP technology. Simulation and chip measurements show more than 30%
improvement in dynamic power and more than 20% reduction in core area.
The functional yield of the PNAND reduces with geometry and voltage scaling.
The second part of this research investigates the use of two mechanisms to improve
the robustness of the PNAND circuit architecture. One is the use of forward and reverse body biases to change the device threshold and the other is the use of RRAM
devices for low voltage operation.
The third part of this research focused on the design of flip-flops with non-volatile
storage. Spin-transfer torque magnetic tunnel junctions (STT-MTJ) are integrated
with both conventional D-flipflop and the PNAND circuits to implement non-volatile
logic (NVL). These non-volatile storage enhanced flip-flops are able to save the state of
system locally when a power interruption occurs. However, manufacturing variations
in the STT-MTJs and in the CMOS transistors significantly reduce the yield, leading
to an overly pessimistic design and consequently, higher energy consumption. A
detailed analysis of the design trade-offs in the driver circuitry for performing backup
and restore, and a novel method to design the energy optimal driver for a given yield is
presented. Efficient designs of two nonvolatile flip-flop (NVFF) circuits are presented,
in which the backup time is determined on a per-chip basis, resulting in minimizing
the energy wastage and satisfying the yield constraint. To achieve a yield of 98%,
the conventional approach would have to expend nearly 5X more energy than the
minimum required, whereas the proposed tunable approach expends only 26% more
energy than the minimum. A non-volatile threshold gate architecture NV-TLFF are
designed with the same backup and restore circuitry in 65nm technology. The embedded
logic in NV-TLFF compensates performance overhead of NVL. This leads to the
possibility of zero-overhead non-volatile datapath circuits. An 8-bit multiply-and-
accumulate (MAC) unit is designed to demonstrate the performance benefits of the
proposed architecture. Based on the results of HSPICE simulations, the MAC circuit
with the proposed NV-TLFF cells is shown to consume at least 20% less power and
area as compared to the circuit designed with conventional DFFs, without sacrificing
any performance.
ContributorsYang, Jinghua (Author) / Vrudhula, Sarma (Thesis advisor) / Barnaby, Hugh (Committee member) / Cao, Yu (Committee member) / Seo, Jae-Sun (Committee member) / Arizona State University (Publisher)
Created2018
Description
The aging mechanism in devices is prone to uncertainties due to dynamic stress conditions. In AMS circuits these can lead to momentary fluctuations in circuit voltage that may be missed by a compact model and hence cause unpredictable failure. Firstly, multiple aging effects in the devices may have underlying correlations. The generation of new traps during TDDB may significantly accelerate BTI, since these traps are close to the dielectric-Si interface in scaled technology. Secondly, the prevalent reliability analysis lacks a direct validation of the lifetime of devices and circuits. The aging mechanism of BTI causes gradual degradation of the device leading to threshold voltage shift and increasing the failure rate. In the 28nm HKMG technology, contribution of BTI to NMOS degradation has become significant at high temperature as compared to Channel Hot Carrier (CHC). This requires revising the End of Lifetime (EOL) calculation based on contribution from induvial aging effects especially in feedback loops. Conventionally, aging in devices is extrapolated from a short-term measurement, but this practice results in unreliable prediction of EOL caused by variability in initial parameters and stress conditions. To mitigate the extrapolation issues and improve predictability, this work aims at providing a new approach to test the device to EOL in a fast and controllable manner. The contributions of this thesis include: (1) based on stochastic trapping/de-trapping mechanism, new compact BTI models are developed and verified with 14nm FinFET and 28nm HKMG data. Moreover, these models are implemented into circuit simulation, illustrating a significant increase in failure rate due to accelerated BTI, (2) developing a model to predict accelerated aging under special conditions like feedback loops and stacked inverters, (3) introducing a feedback loop based test methodology called Adaptive Accelerated Aging (AAA) that can generate accurate aging data till EOL, (4) presenting simulation and experimental data for the models and providing test setup for multiple stress conditions, including those for achieving EOL in 1 hour device as well as ring oscillator (RO) circuit for validation of the proposed methodology, and (5) scaling these models for finding a guard band for VLSI design circuits that can provide realistic aging impact.
ContributorsPatra, Devyani (Author) / Cao, Yu (Thesis advisor) / Barnaby, Hugh (Thesis advisor) / Seo, Jae-Sun (Committee member) / Arizona State University (Publisher)
Created2017
Description
Hardware implementation of deep neural networks is earning significant importance nowadays. Deep neural networks are mathematical models that use learning algorithms inspired by the brain. Numerous deep learning algorithms such as multi-layer perceptrons (MLP) have demonstrated human-level recognition accuracy in image and speech classification tasks. Multiple layers of processing elements called neurons with several connections between them called synapses are used to build these networks. Hence, it involves operations that exhibit a high level of parallelism making it computationally and memory intensive. Constrained by computing resources and memory, most of the applications require a neural network which utilizes less energy. Energy efficient implementation of these computationally intense algorithms on neuromorphic hardware demands a lot of architectural optimizations. One of these optimizations would be the reduction in the network size using compression and several studies investigated compression by introducing element-wise or row-/column-/block-wise sparsity via pruning and regularization. Additionally, numerous recent works have concentrated on reducing the precision of activations and weights with some reducing to a single bit. However, combining various sparsity structures with binarized or very-low-precision (2-3 bit) neural networks have not been comprehensively explored. Output activations in these deep neural network algorithms are habitually non-binary making it difficult to exploit sparsity. On the other hand, biologically realistic models like spiking neural networks (SNN) closely mimic the operations in biological nervous systems and explore new avenues for brain-like cognitive computing. These networks deal with binary spikes, and they can exploit the input-dependent sparsity or redundancy to dynamically scale the amount of computation in turn leading to energy-efficient hardware implementation. This work discusses configurable spiking neuromorphic architecture that supports multiple hidden layers exploiting hardware reuse. It also presents design techniques for minimum-area/-energy DNN hardware with minimal degradation in accuracy. Area, performance and energy results of these DNN and SNN hardware is reported for the MNIST dataset. The Neuromorphic hardware designed for SNN algorithm in 28nm CMOS demonstrates high classification accuracy (>98% on MNIST) and low energy (51.4 - 773 (nJ) per classification). The optimized DNN hardware designed in 40nm CMOS that combines 8X structured compression and 3-bit weight precision showed 98.4% accuracy at 33 (nJ) per classification.
ContributorsKolala Venkataramanaiah, Shreyas (Author) / Seo, Jae-Sun (Thesis advisor) / Chakrabarti, Chaitali (Committee member) / Cao, Yu (Committee member) / Arizona State University (Publisher)
Created2018
Description
The rapid improvement in computation capability has made deep convolutional neural networks (CNNs) a great success in recent years on many computer vision tasks with significantly improved accuracy. During the inference phase, many applications demand low latency processing of one image with strict power consumption requirement, which reduces the efficiency of GPU and other general-purpose platform, bringing opportunities for specific acceleration hardware, e.g. FPGA, by customizing the digital circuit specific for the deep learning algorithm inference. However, deploying CNNs on portable and embedded systems is still challenging due to large data volume, intensive computation, varying algorithm structures, and frequent memory accesses. This dissertation proposes a complete design methodology and framework to accelerate the inference process of various CNN algorithms on FPGA hardware with high performance, efficiency and flexibility.
As convolution contributes most operations in CNNs, the convolution acceleration scheme significantly affects the efficiency and performance of a hardware CNN accelerator. Convolution involves multiply and accumulate (MAC) operations with four levels of loops. Without fully studying the convolution loop optimization before the hardware design phase, the resulting accelerator can hardly exploit the data reuse and manage data movement efficiently. This work overcomes these barriers by quantitatively analyzing and optimizing the design objectives (e.g. memory access) of the CNN accelerator based on multiple design variables. An efficient dataflow and hardware architecture of CNN acceleration are proposed to minimize the data communication while maximizing the resource utilization to achieve high performance.
Although great performance and efficiency can be achieved by customizing the FPGA hardware for each CNN model, significant efforts and expertise are required leading to long development time, which makes it difficult to catch up with the rapid development of CNN algorithms. In this work, we present an RTL-level CNN compiler that automatically generates customized FPGA hardware for the inference tasks of various CNNs, in order to enable high-level fast prototyping of CNNs from software to FPGA and still keep the benefits of low-level hardware optimization. First, a general-purpose library of RTL modules is developed to model different operations at each layer. The integration and dataflow of physical modules are predefined in the top-level system template and reconfigured during compilation for a given CNN algorithm. The runtime control of layer-by-layer sequential computation is managed by the proposed execution schedule so that even highly irregular and complex network topology, e.g. GoogLeNet and ResNet, can be compiled. The proposed methodology is demonstrated with various CNN algorithms, e.g. NiN, VGG, GoogLeNet and ResNet, on two different standalone FPGAs achieving state-of-the art performance.
Based on the optimized acceleration strategy, there are still a lot of design options, e.g. the degree and dimension of computation parallelism, the size of on-chip buffers, and the external memory bandwidth, which impact the utilization of computation resources and data communication efficiency, and finally affect the performance and energy consumption of the accelerator. The large design space of the accelerator makes it impractical to explore the optimal design choice during the real implementation phase. Therefore, a performance model is proposed in this work to quantitatively estimate the accelerator performance and resource utilization. By this means, the performance bottleneck and design bound can be identified and the optimal design option can be explored early in the design phase.
As convolution contributes most operations in CNNs, the convolution acceleration scheme significantly affects the efficiency and performance of a hardware CNN accelerator. Convolution involves multiply and accumulate (MAC) operations with four levels of loops. Without fully studying the convolution loop optimization before the hardware design phase, the resulting accelerator can hardly exploit the data reuse and manage data movement efficiently. This work overcomes these barriers by quantitatively analyzing and optimizing the design objectives (e.g. memory access) of the CNN accelerator based on multiple design variables. An efficient dataflow and hardware architecture of CNN acceleration are proposed to minimize the data communication while maximizing the resource utilization to achieve high performance.
Although great performance and efficiency can be achieved by customizing the FPGA hardware for each CNN model, significant efforts and expertise are required leading to long development time, which makes it difficult to catch up with the rapid development of CNN algorithms. In this work, we present an RTL-level CNN compiler that automatically generates customized FPGA hardware for the inference tasks of various CNNs, in order to enable high-level fast prototyping of CNNs from software to FPGA and still keep the benefits of low-level hardware optimization. First, a general-purpose library of RTL modules is developed to model different operations at each layer. The integration and dataflow of physical modules are predefined in the top-level system template and reconfigured during compilation for a given CNN algorithm. The runtime control of layer-by-layer sequential computation is managed by the proposed execution schedule so that even highly irregular and complex network topology, e.g. GoogLeNet and ResNet, can be compiled. The proposed methodology is demonstrated with various CNN algorithms, e.g. NiN, VGG, GoogLeNet and ResNet, on two different standalone FPGAs achieving state-of-the art performance.
Based on the optimized acceleration strategy, there are still a lot of design options, e.g. the degree and dimension of computation parallelism, the size of on-chip buffers, and the external memory bandwidth, which impact the utilization of computation resources and data communication efficiency, and finally affect the performance and energy consumption of the accelerator. The large design space of the accelerator makes it impractical to explore the optimal design choice during the real implementation phase. Therefore, a performance model is proposed in this work to quantitatively estimate the accelerator performance and resource utilization. By this means, the performance bottleneck and design bound can be identified and the optimal design option can be explored early in the design phase.
ContributorsMa, Yufei (Author) / Vrudhula, Sarma (Thesis advisor) / Seo, Jae-Sun (Thesis advisor) / Cao, Yu (Committee member) / Barnaby, Hugh (Committee member) / Arizona State University (Publisher)
Created2018
Description
The Resistive Random Access Memory (ReRAM) is an emerging non-volatile memory
technology because of its attractive attributes, including excellent scalability (< 10 nm), low
programming voltage (< 3 V), fast switching speed (< 10 ns), high OFF/ON ratio (> 10),
good endurance (up to 1012 cycles) and great compatibility with silicon CMOS technology [1].
However, ReRAM suffers from larger write latency, energy and reliability issue compared to
Dynamic Random Access Memory (DRAM). To improve the energy-efficiency, latency efficiency and reliability of ReRAM storage systems, a low cost cross-layer approach that spans device, circuit, architecture and system levels is proposed.
For 1T1R 2D ReRAM system, the effect of both retention and endurance errors on
ReRAM reliability is considered. Proposed approach is to design circuit-level and architecture-level techniques to reduce raw Bit Error Rate significantly and then employ low cost Error Control Coding to achieve the desired lifetime.
For 1S1R 2D ReRAM system, a cross-point array with “multi-bit per access” per subarray
is designed for high energy-efficiency and good reliability. The errors due to cell-level as well
as array-level variations are analyzed and a low cost scheme to maintain reliability and latency
with low energy consumption is proposed.
For 1S1R 3D ReRAM system, access schemes which activate multiple subarrays with
multiple layers in a subarray are used to achieve high energy efficiency through activating fewer
subarray, and good reliability is achieved through innovative data organization.
Finally, a novel ReRAM-based accelerator design is proposed to support multiple
Convolutional Neural Networks (CNN) topologies including VGGNet, AlexNet and ResNet.
The multi-tiled architecture consists of 9 processing elements per tile, where each tile
implements the dot product operation using ReRAM as computation unit. The processing
elements operate in a systolic fashion, thereby maximizing input feature map reuse and
minimizing interconnection cost. The system-level evaluation on several network benchmarks
show that the proposed architecture can improve computation efficiency and energy efficiency
compared to a state-of-the-art ReRAM-based accelerator.
technology because of its attractive attributes, including excellent scalability (< 10 nm), low
programming voltage (< 3 V), fast switching speed (< 10 ns), high OFF/ON ratio (> 10),
good endurance (up to 1012 cycles) and great compatibility with silicon CMOS technology [1].
However, ReRAM suffers from larger write latency, energy and reliability issue compared to
Dynamic Random Access Memory (DRAM). To improve the energy-efficiency, latency efficiency and reliability of ReRAM storage systems, a low cost cross-layer approach that spans device, circuit, architecture and system levels is proposed.
For 1T1R 2D ReRAM system, the effect of both retention and endurance errors on
ReRAM reliability is considered. Proposed approach is to design circuit-level and architecture-level techniques to reduce raw Bit Error Rate significantly and then employ low cost Error Control Coding to achieve the desired lifetime.
For 1S1R 2D ReRAM system, a cross-point array with “multi-bit per access” per subarray
is designed for high energy-efficiency and good reliability. The errors due to cell-level as well
as array-level variations are analyzed and a low cost scheme to maintain reliability and latency
with low energy consumption is proposed.
For 1S1R 3D ReRAM system, access schemes which activate multiple subarrays with
multiple layers in a subarray are used to achieve high energy efficiency through activating fewer
subarray, and good reliability is achieved through innovative data organization.
Finally, a novel ReRAM-based accelerator design is proposed to support multiple
Convolutional Neural Networks (CNN) topologies including VGGNet, AlexNet and ResNet.
The multi-tiled architecture consists of 9 processing elements per tile, where each tile
implements the dot product operation using ReRAM as computation unit. The processing
elements operate in a systolic fashion, thereby maximizing input feature map reuse and
minimizing interconnection cost. The system-level evaluation on several network benchmarks
show that the proposed architecture can improve computation efficiency and energy efficiency
compared to a state-of-the-art ReRAM-based accelerator.
ContributorsMao, Manqing (Author) / Chakrabariti, Chaitali (Thesis advisor) / Yu, Shimeng (Committee member) / Cao, Yu (Committee member) / Orgas, Umit (Committee member) / Arizona State University (Publisher)
Created2019
Description
In recent years, Artificial Intelligence (AI) (e.g., Deep Neural Networks (DNNs), Transformer) has shown great success in real-world applications due to its superior performance in various cognitive tasks. The impressive performance achieved by AI models normally accompanies the cost of enormous model size and high computational complexity, which significantly hampers their implementation on resource-limited Cyber-Physical Systems (CPS), Internet-of-Things (IoT), or Edge systems due to their tightly constrained energy, computing, size, and memory budget. Thus, the urgent demand for enhancing the \textbf{Efficiency} of DNN has drawn significant research interests across various communities.
Motivated by the aforementioned concerns, this doctoral research has been mainly focusing on Enabling Deep Learning at Edge: From Efficient and Dynamic Inference to On-Device Learning. Specifically, from the inference perspective, this dissertation begins by investigating a hardware-friendly model compression method that effectively reduces the size of AI model while simultaneously achieving improved speed on edge devices. Additionally, due to the fact that diverse resource constraints of different edge devices, this dissertation further explores dynamic inference, which allows for real-time tuning of inference model size, computation, and latency to accommodate the limitations of each edge device. Regarding efficient on-device learning, this dissertation starts by analyzing memory usage during transfer learning training. Based on this analysis, a novel framework called "Reprogramming Network'' (Rep-Net) is introduced that offers a fresh perspective on the on-device transfer learning problem. The Rep-Net enables on-device transferlearning by directly learning to reprogram the intermediate features of a pre-trained model. Lastly, this dissertation studies an efficient continual learning algorithm that facilitates learning multiple tasks without the risk of forgetting previously acquired knowledge. In practice, through the exploration of task correlation, an interesting phenomenon is observed that the intermediate features are highly correlated between tasks with the self-supervised pre-trained model. Building upon this observation, a novel approach called progressive task-correlated layer freezing is proposed to gradually freeze a subset of layers with the highest correlation ratios for each task leading to training efficiency.
ContributorsYang, Li (Author) / Fan, Deliang (Thesis advisor) / Seo, Jae-Sun (Committee member) / Zhang, Junshan (Committee member) / Cao, Yu (Committee member) / Arizona State University (Publisher)
Created2023
Description
Convolutional neural networks(CNNs) achieve high accuracy on large datasets but requires significant computation and storage requirement for training/testing. While many applications demand low latency and energy-efficient processing of the images, deploying these complex algorithms on the hardware is a challenging task. This dissertation first presents a compiler-based CNN training accelerator using DDR3 and HBM2 memory. An optimized RTL library is implemented to perform training-specific tasks and an RTL compiler is developed to generate FPGA-synthesizable RTL based on user-defined constraints. High Bandwidth Memory(HBM) provides efficient off-chip communication and improves the training performance. The impact of HBM2 on CNN training workloads is analyzed and compressively compared with DDR3. For training ResNet-20/VGG-like CNNs for the CIFAR-10 dataset, the proposed CNN training accelerator on Stratix-10 GX FPGA(DDR3) demonstrates 479 GOPS performance, and on Stratix-10 MX FPGA(HBM) shows 4.5/9.7 X energy-efficiency improvement compared to Tesla V100 GPU. Next, the FPGA online learning accelerator is presented. Adopting model segmentation techniques from Progressive Segmented Training(PST), the online learning accelerator achieved a 4.2X reduction in training latency. Furthermore, this dissertation presents an 8-bit floating-point (FP8) training processor which implements (1) Highly parallel tensor cores that maintain high PE utilization, (2) Hardware-efficient channel gating for dynamic output activation sparsity (3) Dynamic weight sparsity based on group Lasso (4) Gradient skipping based on FP prediction error. The 28nm prototype chip demonstrates significant improvements in FLOPs reduction (7.3×), energy efficiency (16.4 TFLOPS/W), and overall training latency speedup (4.7×) for both supervised training and self-supervised training tasks. In addition to the training accelerators, this dissertation also presents a CNN inference accelerator on ASIC(FixyNN) and FPGA(FixyFPGA). FixyNN consists of a fixed-weight feature extractor that generates ubiquitous CNN features and a conventional programmable CNN accelerator. In the fixed-weight feature extractor, the network weights are hard-coded into hardware and used as a fixed operand for the multiplication. Experimental results demonstrate FixyNN can achieve very high energy efficiencies up to 26.6 TOPS/W, and FixyFPGA achieves $2.34\times$ higher GOPS on ImageNet classification. In summary, this dissertation comprehensively discusses novel architectures of high-performance and energy-efficient ASIC/FPGA CNN inference/training accelerators.
ContributorsKolala Venkataramaniah, Shreyas (Author) / Seo, Jae-Sun (Thesis advisor) / Cao, Yu (Committee member) / Chakrabarti, Chaitali (Committee member) / Fan, Deliang (Committee member) / Arizona State University (Publisher)
Created2022