Accurate neuron resilience prediction for a flexible reliability management in neural network accelerators

Schorn, Christoph; Guntoro, Andre; Ascheid, Gerd

doi:10.23919/date.2018.8342151

Cited by 54 publications

(23 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In contrast, since thread-level ABFT generates checksums across small, thread-level matrix multiplications, it can detect one fault per thread in the kernel, resulting in higher overall fault tolerance guarantees. As prior work has illustrated that different layers in NNs are more susceptible to faults than others [52,71], one might also consider varying the approach to ABFT used depending on the fault tolerance requirement of a given layer.…”

Section: Discussion 71 Additional Opportunities For Adaptive Abftmentioning

confidence: 99%

Arithmetic-Intensity-Guided Fault Tolerance for Neural Network Inference on GPUs

Kosaian,

Rashmi

2021

Preprint

View full text Add to dashboard Cite

Neural networks (NNs) are increasingly employed in domains that require high reliability, such as scientific computing and safetycritical systems, as well as in environments more prone to unreliability (e.g., soft errors), such as on spacecraft. As recent work has shown that faults in NN inference can lead to mispredictions and safety hazards, it is critical to impart fault tolerance to NN inference. Algorithm-based fault tolerance (ABFT) is emerging as an appealing approach for efficient fault tolerance in NNs.In this work, we identify new, unexploited opportunities for low-overhead ABFT for NN inference: current inference-optimized GPUs have high compute-to-memory-bandwidth ratios, while many layers of current and emerging NNs have low arithmetic intensity. This leaves many convolutional and fully-connected layers in NNs memory-bandwidth-bound. These layers thus exhibit stalls in computation that could be filled by redundant execution, but that current approaches to ABFT for NN inference cannot exploit.To reduce execution-time overhead for such memory-bandwidthbound layers, we first investigate thread-level ABFT schemes for inference-optimized GPUs that exploit this fine-grained compute underutilization. We then propose intensity-guided ABFT, an adaptive, arithmetic-intensity-guided approach to ABFT that selects the best ABFT scheme for each individual layer between traditional approaches to ABFT, which are suitable for compute-bound layers, and thread-level ABFT, which is suitable for memory-bandwidthbound layers. Through this adaptive approach, intensity-guided ABFT reduces execution-time overhead by 1.09-5.3× across a variety of NNs, lowering the cost of fault tolerance for current and future NN inference workloads.

show abstract

Section: Discussion 71 Additional Opportunities For Adaptive Abftmentioning

confidence: 99%

Arithmetic-Intensity-Guided Fault Tolerance for Neural Network Inference on GPUs

Kosaian,

Rashmi

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…In this work, we make use of the framework presented in [19], where different methods to identify relevant input features in CNNs are implemented. This relevance is defined as the contribution of a certain pixel in the overall prediction of the network [18]. Hereby, we extend this concept to identify network elements where additional approximation can be applied with negligible CNN accuracy degradation.…”

Section: B Kernel-wise Optimizationmentioning

confidence: 99%

“…To select these resilient kernels, we propose a novel approach. This solution is based on understandable neural networks research [18], [19]. The focus of this research is to determine the relationships between neural network predictions and input feature maps.…”

Section: B Kernel-wise Optimizationmentioning

confidence: 99%

Exploiting Resiliency for Kernel-Wise CNN Approximation Enabled by Adaptive Hardware Design

Parra

El-Yamany

Soliman

et al. 2021

2021 IEEE International Symposium on Circuits and Systems (ISCAS)

Self Cite

View full text Add to dashboard Cite

Efficient low-power accelerators for Convolutional Neural Networks (CNNs) largely benefit from quantization and approximation, which are typically applied layer-wise for efficient hardware implementation. In this work, we present a novel strategy for efficient combination of these concepts at a deeper level, which is at each channel or kernel. We first apply layerwise, low bit-width, linear quantization and truncation-based approximate multipliers to the CNN computation. Then, based on a state-of-the-art resiliency analysis, we are able to apply a kernelwise approximation and quantization scheme with negligible accuracy losses, without further retraining. Our proposed strategy is implemented in a specialized framework for fast design space exploration. This optimization leads to a boost in estimated power savings of up to 34% in residual CNN architectures for image classification, compared to the base quantized architecture.

show abstract

“…Previous work [26,33] incorporates circuit-level error detection techniques with architectural error bypassing/masking to mitigate the effect of timing and SRAM errors in voltage underscaled DNN accelerators. The irregular resilience behavior of DNN weights and activations is noted in [3,27], and the described characteristic is utilized to improve the resilience of neural network processing systems through reliability-aware resource allocation techniques. An error-correcting output coding scheme is utilized in [19] to enhance the self-correcting capability of neural networks.…”

Section: Related Workmentioning

confidence: 99%

“…While DNNs are known to exhibit a noticeable immunity against imprecision and small numerical perturbations, the numerical influence of bit-errors could manifest in quite dramatic terms in digital systems due to the asymmetric weight of the bits in the number representations. The consequent degradation in the accuracy of DNN processing systems has been established in the literature on the topic [3,4,17,21,[25][26][27]33]; additional reliability measures are called for therefore, particularly for critical applications.…”

Section: Introductionmentioning

confidence: 99%

Just say zero

Ozen

Orailoğlu

2020

Proceedings of the 39th International Conference on Computer-Aided Design

View full text Add to dashboard Cite

Accurate neuron resilience prediction for a flexible reliability management in neural network accelerators

Cited by 54 publications

References 13 publications

Arithmetic-Intensity-Guided Fault Tolerance for Neural Network Inference on GPUs

Arithmetic-Intensity-Guided Fault Tolerance for Neural Network Inference on GPUs

Exploiting Resiliency for Kernel-Wise CNN Approximation Enabled by Adaptive Hardware Design

Just say zero

Contact Info

Product

Resources

About