Morpheus: Creating application objects efficiently for heterogeneous computing
In high performance computing systems, object deserialization can become a surprisingly
important bottleneck---in our test, a set of general-purpose, highly parallelized applications …
important bottleneck---in our test, a set of general-purpose, highly parallelized applications …
The landscape of gpu-centric communication
D Unat, I Turimbetov, MKT Issa, D Sağbili… - arXiv preprint arXiv …, 2024 - arxiv.org
In recent years, GPUs have become the preferred accelerators for HPC and ML applications
due to their parallelism and fast memory bandwidth. While GPUs boost computation, inter …
due to their parallelism and fast memory bandwidth. While GPUs boost computation, inter …
Network endpoint congestion control for fine-grained communication
Endpoint congestion in HPC networks creates tree saturation that is detrimental to
performance. Endpoint congestion can be alleviated by reducing the injection rate of traffic …
performance. Endpoint congestion can be alleviated by reducing the injection rate of traffic …
Multi-GPU Communication Schemes for Iterative Solvers: When CPUs Are Not in Charge
I Ismayilov, J Baydamirli, D Sağbili, M Wahib… - Proceedings of the 37th …, 2023 - dl.acm.org
This paper proposes a fully autonomous execution model for multi-GPU applications that
completely excludes the involvement of the CPU beyond the initial kernel launch. In a typical …
completely excludes the involvement of the CPU beyond the initial kernel launch. In a typical …
InfiniBand Verbs on GPU: a case study of controlling an InfiniBand network device from the GPU
Due to their massive parallelism and high performance per Watt, GPUs have gained high
popularity in high-performance computing and are a strong candidate for future exascale …
popularity in high-performance computing and are a strong candidate for future exascale …
dCUDA: hardware supported overlap of computation and communication
Over the last decade, CUDA and the underlying GPU hardware architecture have
continuously gained popularity in various high-performance computing application domains …
continuously gained popularity in various high-performance computing application domains …
GPU triggered networking for intra-kernel communications
M LeBeane, K Hamidouche, B Benton… - Proceedings of the …, 2017 - dl.acm.org
GPUs are widespread across clusters of compute nodes due to their attractive performance
for data parallel codes. However, communicating between GPUs across the cluster is …
for data parallel codes. However, communicating between GPUs across the cluster is …
Gpu initiated openshmem: correct and efficient intra-kernel networking for dgpus
K Hamidouche, M LeBeane - Proceedings of the 25th ACM SIGPLAN …, 2020 - dl.acm.org
Current state-of-the-art in GPU networking utilizes a host-centric, kernel-boundary
communication model that reduces performance and increases code complexity. To address …
communication model that reduces performance and increases code complexity. To address …
Exploiting GPUDirect RDMA in designing high performance OpenSHMEM for NVIDIA GPU clusters
GPUDirect RDMA (GDR) brings the high-performance communication capabilities of RDMA
networks like InfiniBand (IB) to GPUs (referred to as" Device"). It enables IB network …
networks like InfiniBand (IB) to GPUs (referred to as" Device"). It enables IB network …
Scalable communication architecture for network-attached accelerators
S Neuwirth, D Frey, M Nuessle… - 2015 IEEE 21st …, 2015 - ieeexplore.ieee.org
On the road to Exascale computing, novel communication architectures are required to
overcome the limitations of host-centric accelerators. Typically, accelerator devices require a …
overcome the limitations of host-centric accelerators. Typically, accelerator devices require a …