اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدMulti-GPU SNN Simulation with Static Load Balancing
175
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
We present a SNN simulator which scales to millions of neurons, billions of synapses, and 8 GPUs. This is made possible by 1) a novel, cache-aware spike transmission algorithm 2) a model parallel multi-GPU distribution scheme and 3) a static, yet very effective load balancing strategy. The simulator further features an easy to use API and the ability to create custom models. We compare the proposed simulator against two state of the art ones on a series of benchmarks using three well-established models. We find that our simulator is faster, consumes less memory, and scales linearly with the number of GPUs.
قيم البحث
اقرأ أيضاً
We present a clock-driven Spiking Neural Network simulator which is up to 3x faster than the state of the art while, at the same time, being more general and requiring less programming effort on both the users and maintainers side. This is made possi
ble by designing our pipeline around work queues which act as interfaces between stages and greatly reduce implementation complexity. We evaluate our work using three well-established SNN models on a series of benchmarks.
We present two novel optimizations that accelerate clock-based spiking neural network (SNN) simulators. The first one targets spike timing dependent plasticity (STDP). It combines lazy- with event-driven plasticity and efficiently facilitates the com
putation of pre- and post-synaptic spikes using bitfields and integer intrinsics. It offers higher bandwidth than event-driven plasticity alone and achieves a 1.5x-2x speedup over our closest competitor. The second optimization targets spike delivery. We partition our graph representation in a way that bounds the number of neurons that need be updated at any given time which allows us to perform said update in shared memory instead of global memory. This is 2x-2.5x faster than our closest competitor. Both optimizations represent the final evolutionary stages of years of iteration on STDP and spike delivery inside Spice (/spaIk/), our state of the art SNN simulator. The proposed optimizations are not exclusive to our graph representation or pipeline but are applicable to a multitude of simulator designs. We evaluate our performance on three well-established models and compare ourselves against three other state of the art simulators.
We consider the load balancing problem in large-scale heterogeneous systems with multiple dispatchers. We introduce a general framework called Local-Estimation-Driven (LED). Under this framework, each dispatcher keeps local (possibly outdated) estima
tes of queue lengths for all the servers, and the dispatching decision is made purely based on these local estimates. The local estimates are updated via infrequent communications between dispatchers and servers. We derive sufficient conditions for LED policies to achieve throughput optimality and delay optimality in heavy-traffic, respectively. These conditions directly imply delay optimality for many previous local-memory based policies in heavy traffic. Moreover, the results enable us to design new delay optimal policies for heterogeneous systems with multiple dispatchers. Finally, the heavy-traffic delay optimality of the LED framework directly resolves a recent open problem on how to design optimal load balancing schemes using delayed information.
Recently, fog computing has been introduced as a modern distributed paradigm and complement to cloud computing to provide services. Fog system extends storing and computing to the edge of the network, which can solve the problem about service computi
ng of the delay-sensitive applications remarkably besides enabling the location awareness and mobility support. Load balancing is an important aspect of fog networks that avoids a situation with some under-loaded or overloaded fog nodes. Quality of Service (QoS) parameters such as resource utilization, throughput, cost, response time, performance, and energy consumption can be improved with load balancing. In recent years, some researches in load balancing techniques in fog networks have been carried out, but there is no systematic review to consolidate these studies. This article reviews the load-balancing mechanisms systematically in fog computing in four classifications, including approximate, exact, fundamental, and hybrid methods (published between 2013 and August 2020). Also, this article investigates load balancing metrics with all advantages and disadvantages related to chosen load balancing mechanisms in fog networks. The evaluation techniques and tools applied for each reviewed study are explored as well. Additionally, the essential open challenges and future trends of these mechanisms are discussed.
GPUs are used for training, inference, and tuning the machine learning models. However, Deep Neural Network (DNN) vary widely in their ability to exploit the full power of high-performance GPUs. Spatial sharing of GPU enables multiplexing several DNN
s on the GPU and can improve GPU utilization, thus improving throughput and lowering latency. DNN models given just the right amount of GPU resources can still provide low inference latency, just as much as dedicating all of the GPU for their inference task. An approach to improve DNN inference is tuning of the DNN model. Autotuning frameworks find the optimal low-level implementation for a certain target device based on the trained machine learning model, thus reducing the DNNs inference latency and increasing inference throughput. We observe an interdependency between the tuned model and its inference latency. A DNN model tuned with specific GPU resources provides the best inference latency when inferred with close to the same amount of GPU resources. While a model tuned with the maximum amount of the GPUs resources has poorer inference latency once the GPU resources are limited for inference. On the other hand, a model tuned with an appropriate amount of GPU resources still achieves good inference latency across a wide range of GPU resource availability. We explore the causes that impact the tuning of a model at different amounts of GPU resources. We present many techniques to maximize resource utilization and improve tuning performance. We enable controlled spatial sharing of GPU to multiplex several tuning applications on the GPU. We scale the tuning server instances and shard the tuning model across multiple client instances for concurrent tuning of different operators of a model, achieving better GPU multiplexing. With our improvements, we decrease DNN autotuning time by up to 75 percent and increase throughput by a factor of 5.
الأسئلة المقترحة
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
