Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userNB-FEB: An Easy-to-Use and Scalable Universal Synchronization Primitive for Parallel Programming
99
0
0.0
(
0
)
Added by
Phuong Ha
Publication date
2008
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
This paper addresses the problem of universal synchronization primitives that can support scalable thread synchronization for large-scale many-core architectures. The universal synchronization primitives that have been deployed widely in conventional architectures like CAS and LL/SC are expected to reach their scalability limits in the evolution to many-core architectures with thousands of cores. We introduce a non-blocking full/empty bit primitive, or NB-FEB for short, as a promising synchronization primitive for parallel programming on may-core architectures. We show that the NB-FEB primitive is universal, scalable, feasible and convenient to use. NB-FEB, together with registers, can solve the consensus problem for an arbitrary number of processes (universality). NB-FEB is combinable, namely its memory requests to the same memory location can be combined into only one memory request, which consequently mitigates performance degradation due to synchronization hot spots (scalability). Since NB-FEB is a variant of the original full/empty bit that always returns a value instead of waiting for a conditional flag, it is as feasible as the original full/empty bit, which has been implemented in many computer systems (feasibility). The original full/empty bit is well-known as a special-purpose primitive for fast producer-consumer synchronization and has been used extensively in the specific domain of applications. In this paper, we show that NB-FEB can be deployed easily as a general-purpose primitive. Using NB-FEB, we construct a non-blocking software transactional memory system called NBFEB-STM, which can be used to handle concurrent threads conveniently. NBFEB-STM is space efficient: the space complexity of each object updated by $N$ concurrent threads/transactions is $Theta(N)$, the optimal.
rate research
Read More
Deep learning has achieved great success in a wide spectrum of multimedia applications such as image classification, natural language processing and multimodal data analysis. Recent years have seen the development of many deep learning frameworks that provide a high-level programming interface for users to design models, conduct training and deploy inference. However, it remains challenging to build an efficient end-to-end multimedia application with most existing frameworks. Specifically, in terms of usability, it is demanding for non-experts to implement deep learning models, obtain the right settings for the entire machine learning pipeline, manage models and datasets, and exploit external data sources all together. Further, in terms of adaptability, elastic computation solutions are much needed as the actual serving workload fluctuates constantly, and scaling the hardware resources to handle the fluctuating workload is typically infeasible. To address these challenges, we introduce SINGA-Easy, a new deep learning framework that provides distributed hyper-parameter tuning at the training stage, dynamic computational cost control at the inference stage, and intuitive user interactions with multimedia contents facilitated by model explanation. Our experiments on the training and deployment of multi-modality data analysis applications show that the framework is both usable and adaptable to dynamic inference loads. We implement SINGA-Easy on top of Apache SINGA and demonstrate our system with the entire machine learning life cycle.
Designing efficient and scalable sparse linear algebra kernels on modern multi-GPU based HPC systems is a daunting task due to significant irregular memory references and workload imbalance across the GPUs. This is particularly the case for Sparse Triangular Solver (SpTRSV) which introduces additional two-dimensional computation dependencies among subsequent computation steps. Dependency information is exchanged and shared among GPUs, thus warrant for efficient memory allocation, data partitioning, and workload distribution as well as fine-grained communication and synchronization support. In this work, we demonstrate that directly adopting unified memory can adversely affect the performance of SpTRSV on multi-GPU architectures, despite linking via fast interconnect like NVLinks and NVSwitches. Alternatively, we employ the latest NVSHMEM technology based on Partitioned Global Address Space programming model to enable efficient fine-grained communication and drastic synchronization overhead reduction. Furthermore, to handle workload imbalance, we propose a malleable task-pool execution model which can further enhance the utilization of GPUs. By applying these techniques, our experiments on the NVIDIA multi-GPU supernode V100-DGX-1 and DGX-2 systems demonstrate that our design can achieve on average 3.53x (up to 9.86x) speedup on a DGX-1 system and 3.66x (up to 9.64x) speedup on a DGX-2 system with 4-GPUs over the Unified-Memory design. The comprehensive sensitivity and scalability studies also show that the proposed zero-copy SpTRSV is able to fully utilize the computing and communication resources of the multi-GPU system.
The computation of first and second-order derivatives is a staple in many computing applications, ranging from machine learning to scientific computing. We propose an algorithm to automatically differentiate algorithms written in a subset of C99 code and its efficient implementation as a Python script. We demonstrate that our algorithm enables automatic, reliable, and efficient differentiation of common algorithms used in physical simulation and geometry processing.
Fast domain propagation of linear constraints has become a crucial component of todays best algorithms and solvers for mixed integer programming and pseudo-boolean optimization to achieve peak solving performance. Irregularities in the form of dynamic algorithmic behaviour, dependency structures, and sparsity patterns in the input data make efficient implementations of domain propagation on GPUs and, more generally, on parallel architectures challenging. This is one of the main reasons why domain propagation in state-of-the-art solvers is single thread only. In this paper, we present a new algorithm for domain propagation which (a) avoids these problems and allows for an efficient implementation on GPUs, and is (b) capable of running propagation rounds entirely on the GPU, without any need for synchronization or communication with the CPU. We present extensive computational results which demonstrate the effectiveness of our approach and show that ample speedups are possible on practically relevant problems: on state-of-the-art GPUs, our geometric mean speed-up for reasonably-large instances is around 10x to 20x and can be as high as 180x on favorably-large instances.
Sensing systems powered by energy harvesting have traditionally been designed to tolerate long periods without energy. As the Internet of Things (IoT) evolves towards a more transient and opportunistic execution paradigm, reducing energy storage costs will be key for its economic and ecologic viability. However, decreasing energy storage in harvesting systems introduces reliability issues. Transducers only produce intermittent energy at low voltage and current levels, making guaranteed task completion a challenge. Existing ad hoc methods overcome this by buffering enough energy either for single tasks, incurring large data-retention overheads, or for one full application cycle, requiring a large energy buffer. We present Julienning: an automated method for optimizing the total energy cost of batteryless applications. Using a custom specification model, developers can describe transient applications as a set of atomically executed kernels with explicit data dependencies. Our optimization flow can partition data- and energy-intensive applications into multiple execution cycles with bounded energy consumption. By leveraging interkernel data dependencies, these energy-bounded execution cycles minimize the number of system activations and nonvolatile data transfers, and thus the total energy overhead. We validate our methodology with two batteryless cameras running energy-intensive machine learning applications. Results demonstrate that compared to ad hoc solutions, our method can reduce the required energy storage by over 94% while only incurring a 0.12% energy overhead.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
