Modern deep learning systems rely on (a) a handtuned
neural network topology, (b) massive amounts of labelled
training data, and (c) extensive training over large-scale compute
resources to build a system that can perform efficient image
classification or speech recognition. Unfortunately, we are still
far away from implementing adaptive general purpose intelligent
systems which would need to learn autonomously in unknown
environments and may not have access to some or any of
these three components. Reinforcement learning and evolutionary
algorithm (EA) based methods circumvent this problem by
continuously interacting with the environment and updating
the models based on obtained rewards. However, deploying
these algorithms on ubiquitous autonomous agents at the edge
(robots/drones) demands extremely high energy-efficiency due
to (i) tight power and energy budgets, (ii) continuous / lifelong
interaction with the environment, (iii) intermittent or no
connectivity to the cloud to offloadheavy-weight processing.
To address this need, we present GENESYS, a HW-SW
prototype of a EA-based learning system, that comprises of
a closed loop learning engine called EvE and an inference
engine called ADAM. EvE can evolve the topology and weights
of neural networks completely in hardware for the task at
hand, without requiring hand-optimization or backpropogation
training. ADAM continuously interacts with the environment and
is optimized for efficiently running the irregular neural networks
generated by EvE. GENESYS identifies and leverages multiple
unique avenues of parallelism unique to EAs that we term “gene”-
level parallelism, and “population”-level parallelism. We ran
GENESYS with a suite of environments from OpenAI gym and
observed 2-5 orders of magnitude higher energy-efficiency over
state-of-the-art embedded and desktop CPU and GPU systems.
more »
« less
Hierarchical Network Connectivity and Partitioning for Reconfigurable Large-Scale Neuromorphic Systems
We present an efficient and scalable partitioning method for mapping large-scale neural network models with locally dense and globally sparse connectivity onto reconfigurable neuromorphic hardware. Scalability in computational efficiency, i.e., amount of time spent in actual computation, remains a huge challenge in very large networks. Most partitioning algorithms also struggle to address the scalability in network workloads in finding a globally optimal partition and efficiently mapping onto hardware. As communication is regarded as the most energy and time-consuming part of such distributed processing, the partitioning framework is optimized for compute-balanced, memory-efficient parallel processing targeting low-latency execution and dense synaptic storage, with minimal routing across various compute cores. We demonstrate highly scalable and efficient partitioning for connectivity-aware and hierarchical address-event routing resource-optimized mapping, significantly reducing the total communication volume recursively when compared to random balanced assignment. We showcase our results working on synthetic networks with varying degrees of sparsity factor and fan-out, small-world networks, feed-forward networks, and a hemibrain connectome reconstruction of the fruit-fly brain. The combination of our method and practical results suggest a promising path toward extending to very large-scale networks and scalable hardware-aware partitioning.
more »
« less
- Award ID(s):
- 1823366
- NSF-PAR ID:
- 10406024
- Date Published:
- Journal Name:
- Frontiers in Neuroscience
- Volume:
- 15
- ISSN:
- 1662-453X
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
With the end of Dennard scaling, power constraints have led to increasing compute specialization in the form of differently specialized accelerators integrated at various levels of the general-purpose system hierarchy. The result is that the most common general-purpose computing platform is now a heterogeneous mix of architectures even within a single die. Consequently, mapping application code regions into available execution engines has become a challenge due to different interfaces and increased software complexity. At the same time, the energy costs of data movement have become increasingly dominant relative to computation energy. This has inspired a move towards data-centric systems, where computation is brought to data, in contrast to traditional processing-centric models. However, enabling compute nearer memory entails its own challenges, including the interactions between distance-specialization and compute-specialization. The granularity of any offload to near(er) memory logic would impact the potential data transmission reduction, as smaller offloads will not be able to amortize the transmission costs of invocation and data return, while very large offloads can only be mapped onto logic that can support all of the necessary operations within kernel-scale codes, which exacerbates both area and power constraints. For better energy efficiency, each set of related operations should be mapped onto the execution engine that, among those capable of running the set of operations, best balances the data movement and the degree of compute specialization of that engine for this code. Further, this offload should proceed in a decentralized way that keeps both the data and control movement low for all transitions among engines and transmissions of operands and results. To enable such a decentralized offload model, we propose an architecture interface that enables a common offload model for accelerators across the memory hierarchy and a tool chain to automatically identify (in a distance-aware fashion) and map profitable code regions on specialized execution engines. We evaluate the proposed architecture for a wide range of workloads and show energy reduction compared to an energy-efficient in-order core. We also demonstrate better area efficiency compared to kernel-scale offloads.more » « less
-
To deliver scalable performance to large-scale scientific and data analytic applications, HPC cluster architectures adopt the distributed-memory model. The performance and scalability of parallel applications on such systems are limited by the communication cost across compute nodes. Therefore, projecting the minimum communication cost and maximum scalability of the user applications plays a critical role in assessing the benefits of porting these applications to HPC clusters as well as developing efficient distributed-memory implementations. Unfortunately, this task is extremely challenging for end users, as it requires comprehensive knowledge of the target application and hardware architecture and demands significant effort and time for manual system analysis. To streamline the process of porting user applications to HPC clusters, this paper presents CommAnalyzer, an automated framework for estimating the communication cost on distributed-memory models from sequential code. CommAnalyzer uses novel dynamic program analyses and graph algorithms to capture the inherent flow of program values (information) in sequential code to estimate the communication when this code is ported to HPC clusters. Therefore, CommAnalyzer makes it possible to project the efficiency/scalability upper-bound (i.e., Roofline) of the effective distributed-memory implementation before even developing one. The experiments with real-world, regular and irregular HPC applications demonstrate the utility of CommAnalyzer in estimating the minimum communication of sequential applications on HPC clusters. In addition, the optimized MPI+X implementations achieve more than 92% of the efficiency upper-bound across the different workloads.more » « less
-
Due to the developments of topographic techniques, clear satellite imagery, and various means for collecting information, geospatial datasets are growing in volume, complexity, and heterogeneity. For efficient execution of spatial computations and analytics on large spatial data sets, parallel processing is required. To exploit fine-grained parallel processing in large scale compute clusters, partitioning in a load-balanced way is necessary for skewed datasets. In this work, we focus on spatial join operation where the inputs are two layers of geospatial data. Our partitioning method for spatial join uses Adaptive Partitioning (ADP) technique, which is based on Quadtree partitioning. Unlike existing partitioning techniques, ADP partitions the spatial join workload instead of partitioning the individual datasets separately to provide better load-balancing. Based on our experimental evaluation, ADP partitions spatial data in a more balanced way than Quadtree partitioning and Uniform grid partitioning. ADP uses an output-sensitive duplication avoidance technique which minimizes duplication of geometries that are not part of spatial join output. In a distributed memory environment, this technique can reduce data communication and storage requirements compared to traditional methods. To improve the performance of ADP, an MPI+Threads based parallelization is presented. With ParADP, a pair of real world datasets, one with 717 million polylines and another with 10 million polygons, is partitioned into 65,536 grid cells within 7 seconds. ParADP performs well with both good weak scaling up to 4,032 CPU cores and good strong scaling up to 4,032 CPU cores.more » « less
-
The increasing popularity and ubiquity of various large graph datasets has caused renewed interest for graph partitioning. Existing graph partitioners either scale poorly against large graphs or disregard the impact of the underlying hardware topology. A few solutions have shown that the nonuniform network communication costs may affect the performance greatly. However, none of them considers the impact of resource contention on the memory subsystems (e.g., LLC and Memory Controller) of modern multicore clusters. They all neglect the fact that the bandwidth of modern high-speed networks (e.g., Infiniband) has become comparable to that of the memory subsystems. In this paper, we provide an in-depth analysis, both theoretically and experimentally, on the contention issue for distributed workloads. We found that the slowdown caused by the contention can be as high as 11x. We then design an architecture-aware graph partitioner, ARGO , to allow the full use of all cores of multicore machines without suffering from either the contention or the communication heterogeneity issue. Our experimental study showed (1) the effectiveness of ARGO , achieving up to 12x speedups on three classic workloads: Breadth First Search, Single Source Shortest Path, and PageRank; and (2) the scalability of ARGO in terms of both graph size and the number of partitions on two billion-edge real-world graphs.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.3389/fnins.2021.797654
