Deep learning recommendation systems serve personalized content under diverse tail-latency targets and input-query loads. In order to do so, state-of-the-art recommendation models rely on terabyte-scale embedding tables to learn user preferences over large bodies of contents. The reliance on a fixed embedding representation of embedding tables not only imposes significant memory capacity and bandwidth requirements but also limits the scope of compatible system solutions. This paper challenges the assumption of fixed embedding representations by showing how synergies between embedding representations and hardware platforms can lead to improvements in both algorithmic- and system performance. Based on our characterization of various embedding representations, we propose a hybrid embedding representation that achieves higher quality embeddings at the cost of increased memory and compute requirements. To address the system performance challenges of the hybrid representation, we propose MP-Rec -- a co-design technique that exploits heterogeneity and dynamic selection of embedding representations and underlying hardware platforms.

On real system hardware, we demonstrate how matching custom accelerators, i.e., GPUs, TPUs, and IPUs, with compatible embedding representations can lead to 16.65x performance speedup. Additionally, in query-serving scenarios, MP-Rec achieves 2.49x and 3.76x higher correct prediction throughput and 0.19% and 0.22% better model quality on a CPU-GPU system for the Kaggle and Terabyte datasets, respectively.

Deep learning recommendation systems must provide high quality, personalized content under strict tail-latency targets and high system loads. This paper presents RecPipe, a system to jointly optimize recommendation quality and inference performance. Central to RecPipe is decomposing recommendation models into multi-stage pipelines to maintain quality while reducing compute complexity and exposing distinct parallelism opportunities. RecPipe implements an inference scheduler to map multi-stage recommendation engines onto commodity, heterogeneous platforms (e.g., CPUs, GPUs).While the hardware-aware scheduling improves ranking efficiency, the commodity platforms suffer from many limitations requiring specialized hardware. Thus, we design RecPipeAccel (RPAccel), a custom accelerator that jointly optimizes quality, tail-latency, and system throughput. RPAc-cel is designed specifically to exploit the distinct design space opened via RecPipe. In particular, RPAccel processes queries in sub-batches to pipeline recommendation stages, implements dual static and dynamic embedding caches, a set of top-k filtering units, and a reconfigurable systolic array. Com-pared to prior-art and at iso-quality, we demonstrate that RPAccel improves latency and throughput by 3x and 6x.

Neural personalized recommendationmodelsareusedacrossawide Samuel Hsia Harvard University Cambridge, Massachusetts, USA shsia@g.harvard.edu David Brooks Harvard University Cambridge, Massachusetts, USA dbrooks@eecs.harvard.edu USA. ACM, New York, NY, USA, 13 pages. https://doi.org/10.1145/3445814. 3446763 variety of datacenter applications including search, social media, and entertainment. State-of-the-art models comprise large embedding tables that have billions of parameters requiring large memory capacities. Unfortunately, large and fast DRAM-based memories levy high infrastructure costs. Conventional SSD-based storage solutions offer an order of magnitude larger capacity, but have worse read latency and bandwidth, degrading inference performance. RecSSD is a near data processing based SSD memory system customized for neural recommendation inference that reduces end-to-end model inference latency by 2× compared to using COTS SSDs across eight industry-representative models.

Deep learning based recommendation systems form the backbone of most personalized cloud services. Though the computer architecture community has recently started to take notice of deep recommendation inference, the resulting solutions have taken wildly different approaches - ranging from near memory processing to at-scale optimizations. To better design future hardware systems for deep recommendation inference, we must first systematically examine and characterize the underlying systems-level impact of design decisions across the different levels of the execution stack. In this paper, we characterize eight industry-representative deep recommendation models at three different levels of the execution stack: algorithms and software, systems platforms, and hardware microarchitectures. Through this cross-stack characterization, we first show that system deployment choices (i.e., CPUs or GPUs, batch size granularity) can give us up to 15x speedup. To better understand the bottlenecks for further optimization, we look at both software operator usage breakdown and CPU frontend and backend microarchitectural inefficiencies. Finally, we model the correlation between key algorithmic model architecture features and hardware bottlenecks, revealing the absence of a single dominant algorithmic component behind each hardware bottleneck.

Personalized recommendation systems leverage deep learning models and account for the majority of data center AI cycles. Their performance is dominated by memory-bound sparse embedding operations with unique irregular memory access patterns that pose a fundamental challenge to accelerate. This paper proposes a lightweight, commodity DRAM compliant, near-memory processing solution to accelerate personalized recommendation inference. The in-depth characterization of production-grade recommendation models shows that embedding operations with high model-, operator- and data-level parallelism lead to memory bandwidth saturation, limiting recommendation inference performance. We propose RecNMP which provides a scalable solution to improve system throughput, supporting a broad range of sparse embedding models. RecNMP is specifically tailored to production environments with heavy co-location of operators on a single server. Several hardware/software co-optimization techniques such as memory-side caching, table-aware packet scheduling, and hot entry profiling are studied, resulting in up to 9.8x memory latency speedup over a highly-optimized baseline. Overall, RecNMP offers 4.2x throughput improvement and 45.8% memory energy savings.

The widespread application of deep learning has changed the landscape of computation in the data center. In particular, personalized recommendation for content ranking is now largely accomplished leveraging deep neural networks. However, despite the importance of these models and the amount of compute cycles they consume, relatively little research attention has been devoted to systems for recommendation. To facilitate research and to advance the understanding of these workloads, this paper presents a set of real-world, production-scale DNNs for personalized recommendation coupled with relevant performance metrics for evaluation. In addition to releasing a set of open-source workloads, we conduct in-depth analysis that underpins future system design and optimization for at-scale recommendation: Inference latency varies by 60% across three Intel server generations, batching and co-location of inferences can drastically improve latency-bounded throughput, and the diverse composition of recommendation models leads to different optimization strategies.

Deep learning based recommendation systems form the backbone of most personalized cloud services. Though the computer architecture community has recently started to take notice of deep recommendation inference, the resulting solutions have taken wildly different approaches - ranging from near memory processing to at-scale optimizations. To better design future hardware systems for deep recommendation inference, we must first systematically examine and characterize the underlying systems-level impact of design decisions across the different levels of the execution stack. In this paper, we characterize eight industry-representative deep recommendation models at three different levels of the execution stack: algorithms and software, systems platforms, and hardware microarchitectures. Through this cross-stack characterization, we first show that system deployment choices (i.e., CPUs or GPUs, batch size granularity) can give us up to 15x speedup. To better understand the bottlenecks for further optimization, we look at both software operator usage breakdown and CPU frontend and backend microarchitectural inefficiencies. Finally, we model the correlation between key algorithmic model architecture features and hardware bottlenecks, revealing the absence of a single dominant algorithmic component behind each hardware bottleneck.

Neural personalized recommendation is the corner-stone of a wide collection of cloud services and products, constituting significant compute demand of the cloud infrastructure. Thus, improving the execution efficiency of neural recommendation directly translates into infrastructure capacity saving. In this paper, we devise a novel end-to-end modeling infrastructure, DeepRecInfra, that adopts an algorithm and system co-design methodology to custom-design systems for recommendation use cases. Leveraging the insights from the recommendation characterization, a new dynamic scheduler, DeepRecSched, is proposed to maximize latency-bounded throughput by taking into account characteristics of inference query size and arrival patterns, recommendation model architectures, and underlying hardware systems. By doing so, system throughput is doubled across the eight industry-representative recommendation models. Finally, design, deployment, and evaluation in at-scale production datacenter shows over 30% latency reduction across a wide variety of recommendation models running on hundreds of machines.

Performance analysis and optimization are essential tasks for hardware and software engineers. In the age of datacenter-scale computing, it is particularly important to conduct comparative performance analysis to understand discrepancies and limitations among different hardware systems and applications. However, there is a distinct lack of productive visualization tools for these comparisons. We present CHAMPVis [1], a web-based, interactive visualization tool that leverages the hierarchical organization of hardware systems to enable productive performance analysis. With CHAMPVis, users can make definitive performance comparisons across applications or hardware platforms. In addition, CHAMPVis provides methods to rank and cluster based on performance metrics to identify common optimization opportunities. Our thorough task analysis reveals three types of datacenter-scale performance analysis tasks: summarization, detailed comparative analysis, and interactive performance bottleneck identification. We propose techniques for each class of tasks including (1) 1-D feature space projection for similarity analysis; (2) Hierarchical parallel coordinates for comparative analysis; and (3) User interactions for rapid diagnostic queries to identify optimization targets. We evaluate CHAMPVis by analyzing standard datacenter applications and machine learning benchmarks in two different case studies.

Machine learning (ML) needs industry-standard performance benchmarks to support design and competitive evaluation of the many emerging software and hardware solutions for ML. But ML training presents three unique benchmarking challenges absent from other domains: optimizations that improve training throughput can increase the time to solution, training is stochastic and time to solution exhibits high variance, and software and hardware systems are so diverse that fair benchmarking with the same binary, code, and even hyperparameters is difficult. We therefore present MLPerf, an ML benchmark that overcomes these challenges. Our analysis quantitatively evaluates MLPerf's efficacy at driving performance and scalability improvements across two rounds of results from multiple vendors.

Deeply embedded applications require low-power, low-cost hardware that fits within stringent area constraints. Deep learning has many potential uses in these domains, but introduces significant inefficiencies stemming from off-chip DRAM accesses of model weights. Ideally, models would fit entirely on-chip. However, even with compression, memory requirements for state-of-the-art mod- els make on-chip inference impractical. Due to increased density, emerging eNVMs are one promising solution. We present MaxNVM, a principled co-design of sparse encodings, protective logic, and fault-prone MLC eNVM technologies (i.e.,RRAM and CTT) to enable highly-efficient DNN inference. We find bit reduction techniques (e.g., clustering and sparse compression) increase weight vulnerability to faults. This limits the capabilities of MLC eNVM. To circumvent this limitation, we improve storage den- sity (i.e., bits-per-cell) with minimal overhead using protective logic. Tradeoffs between density and reliability result in a rich design space. We show that by balancing these techniques, the weights of large networks are able to reasonably fit on-chip. Compared to a naive, single-level-cell eNVM solution, our highly-optimized MLC memory systems reduce weight area by up to 29×. We compare our technique against NVDLA, a state-of-the-art industry-grade CNN accelerator, and demonstrate up to 3.2× reduced power and up to 3.5× reduced energy per ResNet50 inference.

Recurrent neural networks (RNNs) are becoming the de facto solution for speech recognition. RNNs exploit long-term temporal relationships in data by applying repeated, learned transformations. Unlike fully-connected (FC) layers with single vector matrix operations, RNN layers consist of hundreds of such operations chained over time. This poses challenges unique to RNNs that are not found in convolutional neural networks (CNNs) or FC models, namely large dynamic activation. In this paper we present MASR, a principled and modular architecture that accelerates bidirectional RNNs for on-chip ASR. MASR is designed to exploit sparsity in both dynamic activations and static weights. The architecture is enhanced by a series of dynamic activation optimizations that enable compact storage, ensure no energy is wasted computing null operations, and maintain high MAC utilization for highly parallel accelerator designs. In comparison to current state-of-the-art sparse neural network accelerators (e.g., EIE), MASR provides 2x area 3x energy, and 1.6x performance benefits. The modular nature of MASR enables designs that efficiently scale from resource-constrained low-power IoT applications to large-scale, highly parallel datacenter deployments.

This paper presents a 25mm^2 SoC in 16nm FinFET technology targeting flexible acceleration of compute intensive kernels in DNN, DSP and security algorithms. The SoC includes an always-on sub-system, a dual-core Arm A53 CPU cluster, an embedded FPGA array, and a quad-core cache-coherent accelerator cluster. Measurement results demonstrate the following observations: 1) moving DSP/cryptography kernels from A53 to eFPGA increases energy efficiency between 5.5× - 28.9×, 2) the use of cache coherency for datapath accelerators increases throughput by 2.94×, and 3) accelerator flexibility-efficiency (GOPS/W) range spans from 3.1× (A53+S1MD), to 16.5× (eFPGA), to 54.5× (CCA) compared to the dual-core CPU baseline on comparable tasks. The energy per inference on MobileNet-128 CNN shows a peak improvement of 47.6×.

The large memory requirements of deep neural networks limit their deployment and adoption on many devices. Model compression methods effectively reduce the memory requirements of these models, usually through applying transformations such as weight pruning or quantization. In this paper, we present a novel scheme for lossy weight encoding which complements conventional compression techniques. The encoding is based on the Bloomier filter, a probabilistic data structure that can save space at the cost of introducing random errors. Leveraging the ability of neural networks to tolerate these imperfections and by re-training around the errors, the proposed technique, Weightless, can compress DNN weights by up to 496x with the same model accuracy. This results in up to a 1.51x improvement over the state-of-the-art.

Emerging Internet of Things (IoT) devices necessitate system-on-chips (SoCs) that can scale from ultralow power always-on (AON) operation, all the way up to less frequent high-performance tasks at high energy efficiency. Specialized accelerators are essential to help meet these needs at both ends of the scale, but maintaining workload flexibility remains an important goal. This article presents a 25-mm² SoC in 16-nm FinFET technology which demonstrates targeted, flexible acceleration of key compute-intensive kernels spanning machine learning (ML), DSP, and cryptography. The SMIV SoC includes a dedicated AON sub-system, a dual-core Arm Cortex-A53 CPU cluster, an SoC-attached embedded field-programmable gate array (eFPGA) array, and a quad-core cache-coherent accelerator (CCA) cluster. Measurement results demonstrate: 1) 1236x power envelope, from 1.1 mW (only AON cluster), up to 1.36 W (whole SoC at maximum throughput); 2) 5.5-28.9x energy efficiency gain from offloading compute kernels from A53 to eFPGA; 3) 2.94x latency improvement using coherent memory access (CCA cluster); and 4) 55x MobileNetV1 energy per inference improvement on CCA compared to the CPU baseline. The overall flexibility-efficiency range on SMIV spans measured energy efficiencies of 1x (dual-core A53), 3.1x (A53 with SIMD), 16.5x (eFPGA), 54.9x (CCA), and 256x (AON) at a peak efficiency of 4.8 TOPS/W.

One of the biggest performance bottlenecks of today's neural network (NN) accelerators is off-chip memory accesses [11]. In this paper, we propose a method to use multi-level, embedded nonvolatile memory (eNVM) to eliminate all off-chip weight accesses. The use of multi-level memory cells increases the probability of faults. Therefore, we co-design the weights and memories such that their properties complement each other and the faults result in no noticeable NN accuracy loss. In the extreme case, the weights in fully connected layers can be stored using a single transistor. With weight pruning and clustering, we show our technique reduces the memory area by over an order of magnitude compared to an SRAM baseline. In the case of VGG16 (130M weights), we are able to store all the weights in 4.9 mm2, well within the area allocated to SRAM in modern NN accelerators [6]

As the use of deep neural networks continues to grow, so does the fraction of compute cycles devoted to their execution. This has led the CAD and architecture communities to devote considerable attention to building DNN hardware. Despite these efforts, the fault tolerance of DNNs has generally been overlooked. This paper is the first to conduct a large-scale, empirical study of DNN resilience. Motivated by the inherent algorithmic resilience of DNNs, we are interested in understanding the relationship between fault rate and model accuracy. To do so, we present Ares: a light-weight, DNN-specific fault injection framework validated within 12% of real hardware. We find that DNN fault tolerance varies by orders of magnitude with respect to model, layer type, and structure.

One of the biggest performance bottlenecks of today’s neural network (NN) accelerators is off-chip memory accesses. In this paper, we propose a method to use multi-level, embedded non-volatile memory (eNVM) to eliminate all off-chip weight accesses. The use of multi-level memory cells increases the probability of faults. Therefore, we co-design the weights and memories such that their properties complement each other and the faults result in no noticeable NN accuracy loss. In the extreme case, the weights in fully connected layers can be stored using a single transistor. With weight pruning and clustering, we show our technique reduces the memory area by over an order of magnitude compared to an SRAM baseline. In the case of VGG16 (130M weights), we are able to store all the weights in 4.9 mm2, well within the area allocated to SRAM in modern NN accelerators.