Suvarna, M., Tehrani, O. · ArXiv.org · 2025
GigaAPI is a user-space API that simplifies multi-GPU programming, bridging the gap between the capabilities of parallel GPU systems and the ability of developers to harness their full potential. The API offers a comprehensive set of functionalities, including fundamental GPU operations, image processing, and complex GPU tasks, abstracting away the intricacies of low-level CUDA and C++ programming. GigaAPI's modular design aims to inspire future NVIDIA researchers to create a generalized, dynamic, extensible, and cross-GPU architecture-compatible API. Through experiments and simulations, we demonstrate the general efficiency gains achieved by leveraging GigaAPI's simplified multi-GPU programming model and showcase our learning experience through setup and other aspects, as we were interested in learning complex CUDA programming and parallelism. We hope that this contributes to the democratization of parallel GPU computing, enabling researchers and practitioners to unlock new possibilities across diverse domains.
Cited by 0Open access
Patrik Valkovič · Digital Repository (National Repository of Grey Literature) · 2021
Graphical Processing Units stand for the success of Artificial Neural Networks over the past decade and their broader application in the industry. Another promising field of Artificial Intelligence is Evolutionary Algorithms. Their parallelization ability is well known and has been successfully applied in practice. However, these attempts focused on multi-core and multi-machine parallelization rather than on the GPU. This work explores the possibilities of Evolutionary Algorithms parallelization on GPU. I propose implementation in PyTorch library, allowing to execute EA on both CPU and GPU. The proposed implementation provides the most common evolutionary operators for Genetic Algorithms, Real-Coded Evolutionary Algorithms, and Particle Swarm Op- timization Algorithms. Finally, I show the performance is an order of magnitude faster on GPU for medium and big-sized problems and populations. 1
Cited by 0
Richard C. Lawson, Linda Clare, Panagiotis Tsiotras · arXiv (Cornell University) · 2020
Sampling-based planning has become a de facto standard for complex robots given its superior ability to rapidly explore high-dimensional configuration spaces. Most existing optimal sampling-based planning algorithms are sequential in nature and cannot take advantage of wide parallelism available on modern computer hardware. Further, tight synchronization of exploration and exploitation phases in these algorithms limits sample throughput and planner performance. Policy Iteration RRT# (PI-RRT#) exposes fine-grained parallelism during the exploitation phase, but this parallelism has not yet been evaluated using a concrete implementation. We first present a novel GPU implementation of PI-RRT#'s exploitation phase and discuss data structure considerations to maximize parallel performance. Our implementation achieves 3-4× speedup over a serial PI-RRT# implementation for a 77.9% decrease in overall planning time on average. As a second contribution, we introduce the Batched-Extension RRT# algorithm, which loosens the synchronization present in PI-RRT# to realize independent 12.97× and 12.54× speedups under serial and parallel exploitation, respectively.
Cited by 0Open access
Celia Tendero Delicado (20495955), Mónica Chillarón Pérez (20495958), Josep Arnal García (20495961) et al. · Figshare · 2025
<div><p>During acquisition and reconstruction, medical images may become noisy and lose diagnostic quality. In the case of CT scans, obtaining less noisy images results in a higher radiation dose being administered to the patient. Filtering techniques can be utilized to reduce radiation without losing diagnosis capabilities. The objective in this work is to obtain an implementation of a filter capable of processing medical images in real-time. To achieve this we have developed several filter methods based on fuzzy logic, and their GPU implementations, to reduce mixed Gaussian-impulsive noise. These filters have been developed to work in attenuation coefficients so as to not lose any information from the CT scans. The testing volumes come from the Mayo clinic database and consist of CT volumes at full and at simulated low dose. The GPU parallelizations reach speedups of over 2700 and take less than 0.1 seconds to filter more than 300 slices. In terms of quality the filter is competitive with other state of the art algorithmic and AI filters. The proposed method obtains good performance in terms of quality and the parallelization results in real-time filtering.</p></div>
Cited by 0Open access
Sven Šćekić · FER Repository · 2022
Inverzna i direktna kinematika temeljni su problemi programiranja robota. Cilj ovog rada bio je ubrzati rješavanje ovih problema uz pomoć paralelizacije na grafičkoj kartici. Implementirali smo algoritme u biblioteci PyTorch te nakon toga proveli testiranje preciznosti i brzine samih algoritama. Uspješno smo značajno smanjili potrebno vrijeme računanja te omogućili korisnicima laku implementaciju ovih algoritama u već postojeće programe.
Cited by 0
Celia Tendero Delicado (20495955), Mónica Chillarón Pérez (20495958), Josep Arnal García (20495961) et al. · Figshare · 2025
<div><p>During acquisition and reconstruction, medical images may become noisy and lose diagnostic quality. In the case of CT scans, obtaining less noisy images results in a higher radiation dose being administered to the patient. Filtering techniques can be utilized to reduce radiation without losing diagnosis capabilities. The objective in this work is to obtain an implementation of a filter capable of processing medical images in real-time. To achieve this we have developed several filter methods based on fuzzy logic, and their GPU implementations, to reduce mixed Gaussian-impulsive noise. These filters have been developed to work in attenuation coefficients so as to not lose any information from the CT scans. The testing volumes come from the Mayo clinic database and consist of CT volumes at full and at simulated low dose. The GPU parallelizations reach speedups of over 2700 and take less than 0.1 seconds to filter more than 300 slices. In terms of quality the filter is competitive with other state of the art algorithmic and AI filters. The proposed method obtains good performance in terms of quality and the parallelization results in real-time filtering.</p></div>
Cited by 0Open access
Dan Zhang, Rongcai Zhao, Lin Han et al. · 2010
Using GPU for general computing has become an important research direction in high performance computing technology. However, this is not a lossless optimization method. Due to the impact of device initialization cost, data transmission delay, specific characteristics of programs, and other factors, the general computing on GPU may not always achieve the desired speedup, and sometimes results in program execution performance degradation. On the basis of in-depth analysis of GPU internal processing mechanisms, the main factors affecting GPU implementation performance are pointed out, and a parallel cost model for GPU based on static program analysis is proposed to provide judgement basis for using GPU in general computing.
Cited by 1
Arneving, Gustav, Wilhelmsson, Hugo · KTH Publication Database DiVA (KTH Royal Institute of Technology) · 2024
Astronomical image subtraction is a method for generating a difference image from two images, which covers the same area but taken at different times, in order to see changes over time. Due to the images being taken at different times, one of the images has to be convolved, to match the atmospheric conditions ofthe other image. HOTPANTS is an open source software used for astronomical image subtraction. The problem is that HOTPANTS is written in serial C and therefore does not scale well with growing image sizes. There have been previous efforts to parallelize HOTPANTS, which include P-HOTPANTS and GBAISP. However, these projects are outdated or unavailable, respectively. The latest effort, BACH, is a reimplementation of HOTPANTS in C++, where the convolution and subtraction parts have been parallelized on a GPU using OpenCL. This thesis project is a continuation of BACH, called X-BACH, which aims to parallelize the remaining parts of the HOTPANTS algorithm using OpenCL. The results show that some parts of the HOTPANTS algorithm, excluding convolution and subtraction, are highly suitable for the GPU while other parts arenot suitable for the GPU. It is believed that some parts which are not suitable forthe GPU are highly suitable for CPU parallelization. Overall, running on an external GPU, X-BACH achieves a relative speed of 1 to 2 compared to BACH, and a relative of 0.8 to 2.5 compared to HOTPANTS. When running on an integrated GPU, X-BACH achieves a relative speed of 0.5 to 1.2 compared to BACH, and a relative speed of 0.3 to 2 compared to HOTPANTS. Some parts of the algorithm achieves a speedup of up to 10 times when parallelized on a GPU. In terms of accuracy, X-BACH generally obtains a maximum relative error in order of magnitude ranging from 10−7 to 10−1. However, on certain test images, the algorithm has been observed to be unstable.
Cited by 0Open access
Celia Tendero Delicado (20495955), Mónica Chillarón Pérez (20495958), Josep Arnal García (20495961) et al. · Figshare · 2025
<div><p>During acquisition and reconstruction, medical images may become noisy and lose diagnostic quality. In the case of CT scans, obtaining less noisy images results in a higher radiation dose being administered to the patient. Filtering techniques can be utilized to reduce radiation without losing diagnosis capabilities. The objective in this work is to obtain an implementation of a filter capable of processing medical images in real-time. To achieve this we have developed several filter methods based on fuzzy logic, and their GPU implementations, to reduce mixed Gaussian-impulsive noise. These filters have been developed to work in attenuation coefficients so as to not lose any information from the CT scans. The testing volumes come from the Mayo clinic database and consist of CT volumes at full and at simulated low dose. The GPU parallelizations reach speedups of over 2700 and take less than 0.1 seconds to filter more than 300 slices. In terms of quality the filter is competitive with other state of the art algorithmic and AI filters. The proposed method obtains good performance in terms of quality and the parallelization results in real-time filtering.</p></div>
Cited by 0Open access
Ana Flávia Peixoto de Camargos, Viviane Cristine Silva · 2014
Cited by 1
C. Feichtinger, J. Habich, H. Köstler et al. · 2010
Sustaining a large fraction of single GPU performance in parallel computations is considered to be the major problem of GPU-based clusters. We address this issue in the context of a lattice Boltzmann flow solver that is integrated in the WaLBerla software framework. Our multi-GPU implementation uses a block-structured MPI parallelization and is suitable for load balancing and heterogeneous computations on CPUs and GPUs. The overhead required for multi-GPU simulations is discussed in detail. It is demonstrated that a large fraction of the kernel performance can be sustained for weak scaling on InfiniBand clusters, leading to excellent parallel efficiency. However, in strong scaling scenarios using multiple GPUs is much less efficient than running CPU-only simulations on IBM BG/P and x86-based clusters. Hence, a cost analysis must determine the best course of action for a particular simulation task and hardware configuration. Finally we present weak scaling results of heterogeneous simulations conducted on CPUs and GPUs simultaneously, using clusters equipped with varying node configurations. 1
Cited by 0
Biao Wang, Diego F. de Souza, Mauricio Álvarez-Mesa et al. · International Journal of Parallel Programming · 2017
Cited by 6
Wang, Biao, de Souza, Diego F., Álvarez-Mesa, Mauricio et al. · DepositOnce · 2017
In the High Efficiency Video Coding (HEVC) standard, multiple decoding modules have been designed to take advantage of parallel processing. In particular, the HEVC in-loop filters (i.e., the deblocking filter and sample adaptive offset) were conceived to be exploited by parallel architectures. However, the type of the offered parallelism mostly suits the capabilities of multi-core CPUs, thus making a real challenge to efficiently exploit massively parallel architectures such as Graphic Processing Units (GPUs), mainly due to the existing data dependencies between the HEVC decoding procedures. In accordance, this paper presents a novel strategy to increase the amount of parallelism and the resulting performance of the HEVC in-loop filters on GPU devices. For this purpose, the proposed algorithm performs the HEVC filtering at frame-level and employs intrinsic GPU vector instructions. When compared to the state-of-the-art HEVC in-loop filter implementations, the proposed approach also reduces the amount of required memory transfers, thus further boosting the performance. Experimental results show that the proposed GPU in-loop filters deliver a significant improvement in decoding performance. For example, average frame rates of 76 frames per second (FPS) and 125 FPS for Ultra HD 4K are achieved on an embedded NVIDIA GPU for All Intra and Random Access configurations, respectively.
Cited by 0Open access
Naveen Elumalai · 2019
With the continually increasing operating frequencies high-speed effects of modules such as multiconductor interconnect structures and packages are becoming increasingly influential in determining the performance of modern electronic designs. At higher frequencies, they are often characterized by electromagnetic tools yielding tabulated scattering parameter based multiport descriptions or directly using multiport measurements. However, integrating such tabulated data models in regular SPICE like tool environment is a challenge. This was handled by the Vector Fitting (VF) technique, however, it suffers in the presence of large number of ports or poles and becomes computationally slower. To address this problem, recently, parallel vector fitting using multi CPU environment was proposed in the literature. In this thesis, VF algorithm is advanced by proposing the use of the emerging computing platform of GPUs. Several parallel strategies are explored for optimal use of resources: CPUs, GPU and memory, for arriving at better computational performance.
Cited by 2Open access
Gac, Nicolas · HAL (Le Centre pour la Communication Scientifique Directe) · 2016
Cited by 0
Altınok, Kaan Furkan, Peker, Afşin, Temizel, Alptekin · OpenMETU (Middle East Technical University) · 2020
Cited by 0
Ana Flávia, Peixoto Camargos, Viviane Cristine Silva · 2013
We present a performance analysis of a parallel implementation the conjugate gradient (CG), biconjugate gradient (BiCG) and conjugate residual (CR). These solvers using graphic processing units with CUDA parallel programming model. The solvers were optimized for a fast solution of sparse systems of equations arising from Finite Element Analysis (FEA) of electromagnetic phenomena. Results show that the speedup factor for CG method was above 2.5 when compared to the CPU algorithm. With respect to other methods, the performance of CR under GPU was also superior. 2.
Cited by 0
Sofija Tasevska, Viktor Kostadinoski, V. Zdraveski · 2025
Generation of prime numbers and factorization are very important topics in number theory and cryptography. Sieve-based methods like Sieve of Eratosthenes and Sieve of Euler have widespread usage. Pollard’s Rho algorithm, a probabilistic method for integer factorization, is also widely used in cryptographic applications. This paper explores a parallel approach alternative to these methods employing CPU and GPU acceleration to improve efficiency. Although parallelization imposes overhead on small inputs, experience suggests significant speedup for bigger prime boundaries. Our experiments show the potential of parallel computation to improve cryptographic computations, and in particular, this paper aims to demonstrate how increasing the number of CPUs in a system impacts the performance of these algorithms.
Cited by 0
López Ales E., Menchón Lara R.M., Martín Fernandez M. et al. · 2026
Dynamic cardiac MRI (cMRI) is essential for diagnosing cardiovascular diseases, demanding high resolution and image quality. However, achieving superior quality increases data volume and reconstruction time. To tackle this, we propose a solution using parallel imaging and Compressed Sensing (CS) with high-capacity computing devices (e.g., GPUs) for accelerated reconstruction of undersampled data. GPU mem- ory limitations, especially in 3D cMRI, present challenges. Our scalable approach splits the reconstruction problem and employs multiple GPUs (or multiple multi-core CPUs) to per- form multiple optimizations in parallel using the well-known NESTA algorithm, while preserving smoothness between ad- jacent frames in the temporal dimension. Preliminary results on 5D cMRI reconstruction show that our parallel proposal achieves equivalent reconstruction quality in less time, en- abling larger data processing and cost reduction with smaller, more affordable GPUs, as opposed to a single, highly expensive GPU. Moreover, the adoption of the OpenCLIPER frame- work eliminates proprietary GPU technologies. Exploration of adaptability to other sampling schemes opens new possibilities in this field.
Cited by 0Open access
Yuya Sato, Soichiro Ikuno, Atsushi Kamitani · Journal of Advanced Simulation in Science and Engineering · 2025
In this study, we compare the parallel performance of the Improved Variable-Reduction Method (iVRM) and a variable preconditioned Krylov subspace method (VP Krylov) on a single GPU.We treat Poisson equations discretized by the element-free Galerkin method as target problems. iVRM avoids QR decomposition and eliminates Lagrange multipliers, enabling efficient solving of saddle-point systems dominated by sparse matrix operations. Numerical experiments show that both methods converge quickly, with iVRM exhibiting higher CPU efficiency. However, GPU results suggest that larger-scale problems are needed to fully utilize GPU resources. These findings guide the parallel design of iVRM for large-scale saddle-point problems.
Cited by 0Open access