Neural networks are both computationally intensive and memory intensive, making them difficult to deploy on embedded systems with limited hardware resources. To address this limitation, we introduce "deep compression", a three stage pipeline: pruning, trained quantization and Huffman coding, that work together to reduce the storage requirement of neural networks by 35x to 49x without affecting their accuracy. Our method first prunes the network by learning only the important connections. Next, we quantize the weights to enforce weight sharing, finally, we apply Huffman coding. After the first two steps we retrain the network to fine tune the remaining connections and the quantized centroids. Pruning, reduces the number of connections by 9x to 13x; Quantization then reduces the number of bits that represent each connection from 32 to 5. On the ImageNet dataset, our method reduced the storage required by AlexNet by 35x, from 240MB to 6.9MB, without loss of accuracy. Our method reduced the size of VGG-16 by 49x from 552MB to 11.3MB, again with no loss of accuracy. This allows fitting the model into on-chip SRAM cache rather than off-chip DRAM memory. Our compression method also facilitates the use of complex neural networks in mobile applications where application size and download bandwidth are constrained. Benchmarked on CPU, GPU and mobile GPU, compressed network has 3x to 4x layerwise speedup and 3x to 7x better energy efficiency.

https://forum.stanford.edu/events/posterslides/DeepCompressionCompressingDeepNeuralNetworkswithPruning,TrainedQuantizationandHuffmanCoding.pdf

Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Coding

Recent research on deep neural networks has focused primarily on improving accuracy. For a given accuracy level, it is typically possible to identify multiple DNN architectures that achieve that accuracy level. With equivalent accuracy, smaller DNN architectures offer at least three advantages: (1) Smaller DNNs require less communication across servers during distributed training. (2) Smaller DNNs require less bandwidth to export a new model from the cloud to an autonomous car. (3) Smaller DNNs are more feasible to deploy on FPGAs and other hardware with limited memory. To provide all of these advantages, we propose a small DNN architecture called SqueezeNet. SqueezeNet achieves AlexNet-level accuracy on ImageNet with 50x fewer parameters. Additionally, with model compression techniques we are able to compress SqueezeNet to less than 0.5MB (510x smaller than AlexNet). 
The SqueezeNet architecture is available for download here: this https URL

/pdf/squeezenet-alexnet-level-accuracy-with-50x-fewer-parameters-175af9qwgu.pdf

SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size

Many architects believe that major improvements in cost-energy-performance must now come from domain-specific hardware. This paper evaluates a custom ASIC---called a Tensor Processing Unit (TPU)---deployed in datacenters since 2015 that accelerates the inference phase of neural networks (NN). The heart of the TPU is a 65,536 8-bit MAC matrix multiply unit that offers a peak throughput of 92 TeraOps/second (TOPS) and a large (28 MiB) software-managed on-chip memory. The TPU's deterministic execution model is a better match to the 99th-percentile response-time requirement of our NN applications than are the time-varying optimizations of CPUs and GPUs (caches, out-of-order execution, multithreading, multiprocessing, prefetching, ...) that help average throughput more than guaranteed latency. The lack of such features helps explain why, despite having myriad MACs and a big memory, the TPU is relatively small and low power. We compare the TPU to a server-class Intel Haswell CPU and an Nvidia K80 GPU, which are contemporaries deployed in the same datacenters. Our workload, written in the high-level TensorFlow framework, uses production NN applications (MLPs, CNNs, and LSTMs) that represent 95% of our datacenters' NN inference demand. Despite low utilization for some applications, the TPU is on average about 15X - 30X faster than its contemporary GPU or CPU, with TOPS/Watt about 30X - 80X higher. Moreover, using the GPU's GDDR5 memory in the TPU would triple achieved TOPS and raise TOPS/Watt to nearly 70X the GPU and 200X the CPU.

In-Datacenter Performance Analysis of a Tensor Processing Unit

Many architects believe that major improvements in cost-energy-performance must now come from domain-specific hardware. This paper evaluates a custom ASIC---called a Tensor Processing Unit (TPU) --- deployed in datacenters since 2015 that accelerates the inference phase of neural networks (NN). The heart of the TPU is a 65,536 8-bit MAC matrix multiply unit that offers a peak throughput of 92 TeraOps/second (TOPS) and a large (28 MiB) software-managed on-chip memory. The TPU's deterministic execution model is a better match to the 99th-percentile response-time requirement of our NN applications than are the time-varying optimizations of CPUs and GPUs that help average throughput more than guaranteed latency. The lack of such features helps explain why, despite having myriad MACs and a big memory, the TPU is relatively small and low power. We compare the TPU to a server-class Intel Haswell CPU and an Nvidia K80 GPU, which are contemporaries deployed in the same datacenters. Our workload, written in the high-level TensorFlow framework, uses production NN applications (MLPs, CNNs, and LSTMs) that represent 95% of our datacenters' NN inference demand. Despite low utilization for some applications, the TPU is on average about 15X -- 30X faster than its contemporary GPU or CPU, with TOPS/Watt about 30X -- 80X higher. Moreover, using the CPU's GDDR5 memory in the TPU would triple achieved TOPS and raise TOPS/Watt to nearly 70X the GPU and 200X the CPU.

/pdf/in-datacenter-performance-analysis-of-a-tensor-processing-4pmp83zqw4.pdf

Deep neural networks (DNNs) are currently widely used for many artificial intelligence (AI) applications including computer vision, speech recognition, and robotics. While DNNs deliver state-of-the-art accuracy on many AI tasks, it comes at the cost of high computational complexity. Accordingly, techniques that enable efficient processing of DNNs to improve energy efficiency and throughput without sacrificing application accuracy or increasing hardware cost are critical to the wide deployment of DNNs in AI systems. This article aims to provide a comprehensive tutorial and survey about the recent advances toward the goal of enabling efficient processing of DNNs. Specifically, it will provide an overview of DNNs, discuss various hardware platforms and architectures that support DNNs, and highlight key trends in reducing the computation cost of DNNs either solely via hardware design changes or via joint hardware design and DNN algorithm changes. It will also summarize various development resources that enable researchers and practitioners to quickly get started in this field, and highlight important benchmarking metrics and design considerations that should be used for evaluating the rapidly growing number of DNN hardware designs, optionally including algorithmic codesigns, being proposed in academia and industry. The reader will take away the following concepts from this article: understand the key design considerations for DNNs; be able to evaluate different DNN hardware implementations with benchmarks and comparison metrics; understand the tradeoffs between various hardware architectures and platforms; be able to evaluate the utility of various DNN design techniques for efficient processing; and understand recent implementation trends and opportunities.

Efficient Processing of Deep Neural Networks: A Tutorial and Survey

State-of-the-art deep neural networks (DNNs) have hundreds of millions of connections and are both computationally and memory intensive, making them difficult to deploy on embedded systems with limited hardware resources and power budgets. While custom hardware helps the computation, fetching weights from DRAM is two orders of magnitude more expensive than ALU operations, and dominates the required power.Previously proposed 'Deep Compression' makes it possible to fit large DNNs (AlexNet and VGGNet) fully in on-chip SRAM. This compression is achieved by pruning the redundant connections and having multiple connections share the same weight. We propose an energy efficient inference engine (EIE) that performs inference on this compressed network model and accelerates the resulting sparse matrix-vector multiplication with weight sharing. Going from DRAM to SRAM gives EIE 120× energy saving; Exploiting sparsity saves 10×; Weight sharing gives 8×; Skipping zero activations from ReLU saves another 3×. Evaluated on nine DNN benchmarks, EIE is 189× and 13× faster when compared to CPU and GPU implementations of the same DNN without compression. EIE has a processing power of 102 GOPS working directly on a compressed network, corresponding to 3 TOPS on an uncompressed network, and processes FC layers of AlexNet at 1.88×104 frames/sec with a power dissipation of only 600mW. It is 24,000× and 3,400× more energy efficient than a CPU and GPU respectively. Compared with DaDianNao, EIE has 2.9×, 19× and 3× better throughput, energy efficiency and area efficiency.

EIE: efficient inference engine on compressed deep neural network

State-of-the-art deep neural networks (DNNs) have hundreds of millions of connections and are both computationally and memory intensive, making them difficult to deploy on embedded systems with limited hardware resources and power budgets. While custom hardware helps the computation, fetching weights from DRAM is two orders of magnitude more expensive than ALU operations, and dominates the required power. 
Previously proposed 'Deep Compression' makes it possible to fit large DNNs (AlexNet and VGGNet) fully in on-chip SRAM. This compression is achieved by pruning the redundant connections and having multiple connections share the same weight. We propose an energy efficient inference engine (EIE) that performs inference on this compressed network model and accelerates the resulting sparse matrix-vector multiplication with weight sharing. Going from DRAM to SRAM gives EIE 120x energy saving; Exploiting sparsity saves 10x; Weight sharing gives 8x; Skipping zero activations from ReLU saves another 3x. Evaluated on nine DNN benchmarks, EIE is 189x and 13x faster when compared to CPU and GPU implementations of the same DNN without compression. EIE has a processing power of 102GOPS/s working directly on a compressed network, corresponding to 3TOPS/s on an uncompressed network, and processes FC layers of AlexNet at 1.88x10^4 frames/sec with a power dissipation of only 600mW. It is 24,000x and 3,400x more energy efficient than a CPU and GPU respectively. Compared with DaDianNao, EIE has 2.9x, 19x and 3x better throughput, energy efficiency and area efficiency.

EIE: Efficient Inference Engine on Compressed Deep Neural Network

Reconfigurable architectures have gained popularity in recent years as they allow the design of energy-efficient accelerators. Fine-grain fabrics (e.g. FPGAs) have traditionally suffered from performance and power inefficiencies due to bit-level reconfigurable abstractions. Both fine-grain and coarse-grain architectures (e.g. CGRAs) traditionally require low level programming and suffer from long compilation times. We address both challenges with Plasticine, a new spatially reconfigurable architecture designed to efficiently execute applications composed of parallel patterns. Parallel patterns have emerged from recent research on parallel programming as powerful, high-level abstractions that can elegantly capture data locality, memory access patterns, and parallelism across a wide range of dense and sparse applications.We motivate Plasticine by first observing key application characteristics captured by parallel patterns that are amenable to hardware acceleration, such as hierarchical parallelism, data locality, memory access patterns, and control flow. Based on these observations, we architect Plasticine as a collection of Pattern Compute Units and Pattern Memory Units. Pattern Compute Units are multi-stage pipelines of reconfigurable SIMD functional units that can efficiently execute nested patterns. Data locality is exploited in Pattern Memory Units using banked scratchpad memories and configurable address decoders. Multiple on-chip address generators and scatter-gather engines make efficient use of DRAM bandwidth by supporting a large number of outstanding memory requests, memory coalescing, and burst mode for dense accesses. Plasticine has an area footprint of 113 mm2 in a 28nm process, and consumes a maximum power of 49 W at a 1 GHz clock. Using a cycle-accurate simulator, we demonstrate that Plasticine provides an improvement of up to 76.9x in performance-per-Watt over a conventional FPGA over a wide range of dense and sparse applications.

https://dl.acm.org/doi/pdf/10.1145/3079856.3080256

Plasticine: A Reconfigurable Architecture For Parallel Paterns

Industry is increasingly turning to reconfigurable architectures like FPGAs and CGRAs for improved performance and energy efficiency. Unfortunately, adoption of these architectures has been limited by their programming models. HDLs lack abstractions for productivity and are difficult to target from higher level languages. HLS tools are more productive, but offer an ad-hoc mix of software and hardware abstractions which make performance optimizations difficult. In this work, we describe a new domain-specific language and compiler called Spatial for higher level descriptions of application accelerators. We describe Spatial's hardware-centric abstractions for both programmer productivity and design performance, and summarize the compiler passes required to support these abstractions, including pipeline scheduling, automatic memory banking, and automated design tuning driven by active machine learning. We demonstrate the language's ability to target FPGAs and CGRAs from common source code. We show that applications written in Spatial are, on average, 42% shorter and achieve a mean speedup of 2.9x over SDAccel HLS when targeting a Xilinx UltraScale+ VU9P FPGA on an Amazon EC2 F1 instance.

https://dl.acm.org/doi/pdf/10.1145/3192366.3192379

Spatial: a language and compiler for application accelerators

Unlike traditional dark silicon works that attack the computing logic, this article puts a focus on the memory part, which dissipates most of the energy for memory-bound CPU applications. This article discusses the dark memory state and present Pareto curves for compute units, accelerators, and on-chip memory, and motivates the need for HW/SW codesign for parallelism and locality. –Muhammad Shafique, Vienna University of Technology

Ardavan Pedram

Papers

EIE: efficient inference engine on compressed deep neural network

EIE: Efficient Inference Engine on Compressed Deep Neural Network

Plasticine: A Reconfigurable Architecture For Parallel Paterns

Spatial: a language and compiler for application accelerators

Dark Memory and Accelerator-Rich System Optimization in the Dark Silicon Era