• The Journal of the Korea Contents Association
• /
• v.15 no.7
• /
• pp.1-11
• /
• 2015

In recent high performance computing system, GPGPU has been widely used to process general-purpose applications as well as graphics applications, since GPU can provide optimized computational resources for massive parallel processing. Unfortunately, GPGPU doesn't exploit computational resources on GPU in executing general-purpose applications fully, because the applications cannot be optimized to GPU architecture. Therefore, we provide GPU research guideline to improve the performance of computing systems using GPGPU. To accomplish this, we analyze the negative factors on GPU performance. In this paper, in order to clearly classify the cause of the negative factors on GPU performance, GPU core status are defined into 5 status: fully active status, partial active status, idle status, memory stall status and GPU core stall status. All status except fully active status cause performance degradation. We evaluate the ratio of each GPU core status depending on the characteristics of benchmarks to find specific reasons which degrade the performance of GPU. According to our simulation results, partial active status, idle status, memory stall status and GPU core stall status are induced by computational resource underutilization problem, low parallelism, high memory requests, and structural hazard, respectively.

• The Journal of the Korea Contents Association
• /
• v.22 no.8
• /
• pp.81-93
• /
• 2022

Recently, research on processing a large-capacity graph using GPUs has been conducting. In order to process a large-capacity graph in a GPU with limited memory, the graph must be divided into subgraphs and then processed by scheduling subgraphs. In this paper, we propose an efficient graph algorithm processing scheme in GPU environments with limited memory and performance evaluation. The proposed scheme consists of a graph differential subgraph scheduling method and a graph segmentation method. The bulk graph segmentation method determines how a large-capacity graph can be segmented into subgraphs so that it can be processed efficiently by the GPU. The differential subgraph scheduling method schedule subgraphs processed by GPUs to reduce redundant transmission of the repeatedly used data between HOST-GPUs. It shows the superiority of the proposed scheme by performing various performance evaluations.

• KIPS Transactions on Computer and Communication Systems
• /
• v.7 no.4
• /
• pp.99-102
• /
• 2018

Caffe is a neural net learning software which is widely used in academic researches. The GPU memory capacity is one of the most important aspects of designing neural net architectures. For example, many object detection systems require to use less than 12GB to fit a single GPU. In this paper, we extended Caffe to allow to use more than 12GB GPU memory. To verify the effectiveness of the extended software, we executed some training experiments to determine the learning efficiency of the object detection neural net software using a PC with three GPUs.

• KNOM Review
• /
• v.23 no.1
• /
• pp.43-50
• /
• 2020

As General Purpose Graphics Processing Unit (GPGPU) recently plays an essential role in high-performance computing, several cloud service providers offer GPU service. Most cluster orchestration platforms in a cloud environment using containers allocate the integer number of GPU to jobs and do not allow a node shared with other jobs. In this case, resource utilization of a GPU node might be low if a job does not intensively require either many cores or large size of memory in GPU. GPU virtualization brings opportunities to realize kernel concurrency and share resources. However, performance may vary depending on characteristics of applications running concurrently and interference among them due to resource contention on a node. This paper proposes GPU container co-execution framework with multiple server creation and execution based on Kubernetes, container orchestration platform for measuring interference which may be occurred by sharing GPU resources. Performance changes according to scheduling policies were investigated by executing several jobs on GPU. The result shows that optimal scheduling is not possible only considering GPU memory and computing resource usage. Interference caused by co-execution among applications is measured using the framework.

• Journal of Positioning, Navigation, and Timing
• /
• v.12 no.1
• /
• pp.75-81
• /
• 2023

In this study, we design an optimized Graphics Processing Unit (GPU)-based GNSS signal processing technique with the goal of designing and implementing a GNSS Software Defined Receiver (SDR) that can operate in real time all-in-view mode under multi-constellation and multi-frequency signal environment. In the proposed structure the correlators of the existing GNSS SDR are processed by the GPU. We designed a memory structure and processing method that can minimize memory access bottlenecks and optimize the GPU memory resource distribution. The designed GNSS SDR can select and operate only the desired GNSS or desired satellite signals by user input. Also, parameters such as the number of quantization bits, sampling rate, and number of signal tracking arms can be selected. The computing capability of the designed GPU-based GNSS SDR was evaluated and it was confirmed that up to 2400 channels can be processed in real time. As a result, the GPU-based GNSS SDR has sufficient performance to operate in real-time all-in-view mode. In future studies, it will be used for more diverse GNSS signal processing and will be applied to multipath effect analysis using more tracking arms.

• Smart Media Journal
• /
• v.8 no.2
• /
• pp.9-15
• /
• 2019

Since the computational complexity for hologram generation increases exponentially with respect to the size of the point cloud, parallel processing using CUDA and/or OpenCL library based on multiple GPUs has recently become popular. The CUDA kernel for parallelization needs to consist of threads, blocks, and grids properly in accordance with the number of cores and the memory size in the GPU. In addition, in case of multiple GPU environments, the distribution in grid-by-grid, in block-by-block, or in thread-by-thread is needed according to the number of GPUs. In order to evaluate the performance of CGH generation, we compared the computational speed in CPU, in single GPU, and in multi-GPU environments by gradually increasing the number of points in a point cloud from 10 to 1,000,000. We also present a memory structure design and a calculation method required in the CUDA-based parallel processing to accelerate the CGH (Computer Generated Hologram) generation operation in multiple GPU environments.

• Smart Media Journal
• /
• v.3 no.1
• /
• pp.33-38
• /
• 2014

Even though the performance of microprocessor is improved continuously, the performance improvement of computing system becomes hard to increase, in order to some drawbacks including increased power consumption. To solve the problem, general-purpose computing on graphics processing units(GPGPUs), which execute general-purpose applications by using specialized parallel-processing device representing graphics processing units(GPUs), have been focused. However, the characteristics of applications related with graphics is substantially different from the characteristics of general-purpose applications. Therefore, GPUs cannot exploit the outstanding computational resources sufficiently due to various constraints, when they execute general-purpose applications. When designing GPUs for GPGPU, memory system is important to effectively exploit the GPUs since typically general-purpose applications requires more memory accesses than graphics applications. Especially, external memory access requiring long latency impose a big overhead on the performance of GPUs. Therefore, the GPU performance must be improved if hierarchical memory architecture which can reduce the number of external memory access is applied. For this reason, we will investigate the analysis of GPU performance according to hierarchical cache architectures in executing various benchmarks.

• Journal of Korea Game Society
• /
• v.11 no.1
• /
• pp.147-157
• /
• 2011

Webcam doesn't perform memory-alignment in order to reduce the transmission time of image data. Memory-unaligned image data is unsuitable for the processing on GPU. Accordingly, we convert it to available color format for optimization in high speed image processing. In this paper, we propose a technique that accelerates webcam's color format conversion by using NVDIA CUDA. We propose an optimization which is about memory accesses and thread composition, also evaluate memory and computing performance for verifying a hypothesis which is the performance of the proposed architecture and optimizing degree on low-performance GPU. Following the optimization technique, we show performance improvements over maximum 68 percent.

• Geophysics and Geophysical Exploration
• /
• v.14 no.1
• /
• pp.98-104
• /
• 2011

Numerical simulation in exploration geophysics provides important insights into subsurface wave propagation phenomena. Although elastic wave simulations take longer to compute than acoustic simulations, an elastic simulator can construct more realistic wavefields including shear components. Therefore, it is suitable for exploration of the responses of elastic bodies. To overcome the long duration of the calculations, we use a Graphic Processing Unit (GPU) to accelerate the elastic wave simulation. Because a GPU has many processors and a wide memory bandwidth, we can use it in a parallelised computing architecture. The GPU board used in this study is an NVIDIA Tesla C1060, which has 240 processors and a 102 GB/s memory bandwidth. Despite the availability of a parallel computing architecture (CUDA), developed by NVIDIA, we must optimise the usage of the different types of memory on the GPU device, and the sequence of calculations, to obtain a significant speedup of the computation. In this study, we simulate two- (2D) and threedimensional (3D) elastic wave propagation using the Finite-Difference Time-Domain (FDTD) method on GPUs. In the wave propagation simulation, we adopt the staggered-grid method, which is one of the conventional FD schemes, since this method can achieve sufficient accuracy for use in numerical modelling in geophysics. Our simulator optimises the usage of memory on the GPU device to reduce data access times, and uses faster memory as much as possible. This is a key factor in GPU computing. By using one GPU device and optimising its memory usage, we improved the computation time by more than 14 times in the 2D simulation, and over six times in the 3D simulation, compared with one CPU. Furthermore, by using three GPUs, we succeeded in accelerating the 3D simulation 10 times.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.9
• /
• pp.3-11
• /
• 2013

Recently, a plenty of researches have been conducted using the massively parallel processing of GPU for the implementation of communication system. In this paper, we tried to reduce software simulation time applying GPU with sliding block method to Viterbi decoder in DVB-T system which is one of European DTV standards. First of all, we implement DVB-T system by CPU and estimate cost time whereby the system processes one OFDM symbol. Secondly, we implement Viterbi decoder by software using NVIDIA's massive GPU processor. In our work, stream process method is applied to reduce the overhead for data transfer between CPU and GPU, as well as coalescing method to lower the global memory access time. In addition, data structure design method is used to maximize the shared memory usage. Consequently, our proposed method is approximately 11 times faster in 2K mode and 60 times faster in 8K mode for the process in Viterbi decoder.