RAPTOR: Practical Numerical Profiling of Scientific Applications
Faveo Hoerold, Ivan R. Ivanov, Akash Dhruv, William S. Moses, Anshu Dubey, Mohamed Wahib, Jens Domke
摘要
Figure 1: Qualitative visualization of bubble deformation and splitting over time (𝑡) for the Bubble test case at Reynolds number Re = 3500, performed using incompressible multiphase fluid dynamics solvers in Flash-X. The contours depict the zero level of the level-set function, 𝜙 = 0, which defines the air-water interface (𝜙 > 0 in the air phase and 𝜙 < 0 in the water phase). Adaptive mesh refinement (AMR) levels Ω 𝑀 , Ω 𝑀 -1 , and Ω 𝑀 -2 are shown and labeled in the left panel. AMR dynamically refines the mesh near the interface to accurately capture interface dynamics. We use RAPTOR to truncate FP operations in the advection and diffusion modules of the Navier-Stokes solver following two different strategies. In the first strategy (Trunc. Everywhere), truncation is applied throughout the entire mesh. The second strategy selectively applies truncation in each mesh cell using a cut-off at the 𝑀 -𝑙 AMR level (where 𝑙 = 1, 2, . . .). Insets show zoomed-in views of the liquid-gas interface for simulations using 4 bit and 12 bit mantissas, highlighting the effect of reduced precision on the final simulation outcome. Further details in §6.2.
问问这篇 Paper
智能体会读完全文。
Lune 把这篇 Paper 索引到了每一个公式,引用它的顶会 Paper 也一样。你提问,回答直接引用原文。
它引用的顶会 Paper5
- Instead of Rewriting Foreign Code for Machine Learning, Automatically Synthesize Fast GradientsWilliam S. Moses, Valentin ChuravyNeurIPS 2020 · 被引用 144 次
- Spying on the Floating Point Behavior of Existing, Unmodified Scientific ApplicationsPeter A. Dinda, Alex Bernat, Conor HetlandHPDC 2020 · 被引用 20 次
- Design and Evaluation of GPU-FPX: A Low-Overhead tool for Floating-Point Exception Detection in NVIDIA GPUsXinyi Li, Ignacio Laguna, Bo Fang, Katarzyna Swirydowicz 等HPDC 2023 · 被引用 12 次
- Approximate Computing Through the Lens of Uncertainty QuantificationKonstantinos Parasyris, James Diffenderfer, Harshitha Menon, Ignacio Laguna 等SC 2022 · 被引用 5 次
- FPVM: Towards a Floating Point Virtual MachinePeter A. Dinda, Nick Wanninger, Jiacheng Ma, Alex Bernat 等HPDC 2022 · 被引用 4 次
相关 Paper
- AMR-Transformer: Enabling Efficient Long-range Interaction for Complex Neural Fluid SimulationZeyi Xu, Jinfan Liu, Kuangxu Chen, Ye Chen 等CVPR 2025
- Constraint bubbles and affine regions: reduced fluid models for efficient immersed bubbles and flexible spatial coarseningRyan Goldade, Mridul Aanjaneya, Christopher BattySIGGRAPH 2020 · 被引用 17 次
- A practical octree liquid simulator with adaptive surface resolutionRyoichi Ando, Christopher BattySIGGRAPH 2020 · 被引用 29 次
- Ray Tracing Structured AMR Data Using ExaBricksIngo Wald, Stefan Zellmann, Will Usher, Nate Morrical 等IEEE VIS 2020 · 被引用 18 次
- A clebsch method for free-surface vortical flow simulationShiying Xiong, Zhecheng Wang, Mengdi Wang, Bo ZhuSIGGRAPH 2022 · 被引用 33 次
