1 of 22

Viskores Update

April 29, 2025 | DOECGF

Kenneth Moreland

Oak Ridge National Laboratory

ORNL IS MANAGED BY UT-BATTELLE LLC �FOR THE US DEPARTMENT OF ENERGY

1

2 of 22

Acknowledgements

This work was supported by the U.S. Department of Energy (DOE) RAPIDS SciDAC project under contract number DE-AC05-00OR22725.

This research was supported by the Exascale Computing Project (17-SC-20-SC), a collaborative effort of two U.S. Department of Energy organizations (Office of Science and the National Nuclear Security Administration) responsible for the planning and preparation of a capable exascale ecosystem, including software, applications, hardware, advanced system engineering, and early testbed platforms, in support of the nation’s exascale computing imperative.

This material is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Advanced Scientific Computing Research, under Award Numbers 10-014707, 12-015215, and 14-017566.

Thanks to many, many partners in labs, universities, and industry.

2

3 of 22

Why VTK-m

Viskores

3

3

4 of 22

Distributed

Parallelism

4

5 of 22

Distributed

Parallelism

5

6 of 22

Contour

Streams

Clip

Render

CUDA

Xeon Phi

CPU

AMD ROCm

Intel GPU

Surface

Normals

Ghost Cells

Warp

…

6

7 of 22

7

8 of 22

8

9 of 22

What is Going on Now?

Transition from VTK-m

9

9

10 of 22

Transition From VTK-m to Viskores

Version 2.3.0

Version 1.0.0

viskores/Utilities/Scripts/vtkm_to_viskores_project.bash

10

11 of 22

New Development will Happen in Viskores

11

12 of 22

Why the Change in Name?

Name Change was Necessary to Join the High Performance Software Foundation (HPSF)

Provides a Community of HPC Software

  • Helps join users with software products
  • Joined the ranks of Spack, Kokkos, E4S, AMReX, WarpX, Trilinos, HPCToolkit, and more

Consolidates Resources for Software Support

  • Hosts community events (such as the HPSFCon next week)
  • Provides resources for CI build clusters

12

13 of 22

What is Going on Now?

13

13

14 of 22

Overhead costs for parallel particle advection

  • Parallel particle advection is notoriously difficult
  • Performed a rank- and particle-centric analysis
  • Rank-centric: two new metrics, rank participation and aggregated rank participation
  • Particle-centric: non-communication overheads are significant
  • Overheads associated with communication (but not the communication) are significant
    • Particle batch size
    • “Ping pong” particles

​

Rank participation for each dataset

Accumulated batch size for the slowest particle. Since particles are processed in groups, large groups result in higher overheads

An example lf “ping pong” particles circulate along block boundaries

Zhe Wang, Kenneth Moreland, Matthew Larsen, James Kress, Hank Childs, David Pugmire Parallelize Over Data Particle Advection: Participation, Ping Pong Particles and Overhead. To appear in IEEE Transactions on Visualization and Computer Graphics

14

15 of 22

Correlation-Aware Uncertainty Visualization

T. M. Athawale, K. Moreland, D. Pugmire, C. R. Johnson, P. Rosen, A. Georgiadou, M. Norman, A. Entezari

MAGIC: Marching Cubes Isosurface Uncertainty for Gaussian Uncertain Data with Correlation, Submitted to IEEE TVCG, 2025.

15

16 of 22

Uncertainty Visualization: Integration with Production Visualization Software

  • Wang et al., FunMC2: A Filter for Marching Cubes Uncertainty on Many-Core Devices, EGPGV, 2023;
  • Athawale et al., Uncertainty Visualization of Critical Points of 2D Scalar Fields for Parametric and Nonparametric Probabilistic Models, IEEE TVCG, 2025

Near-real-time uncertainty visualization in ParaView using the VTK-m library [Moreland et al., 2016]

16

17 of 22

Planned ANARI Integration

17

17

18 of 22

Raycasting

Device

18

19 of 22

Raycasting

Device

19

20 of 22

Raycasting

Data Objects

Device

Interop

20

21 of 22

Raycasting

Data Objects

Device

Interop

Helide

RadeonProRender

OSPRay

VisRTX

21

22 of 22

Questions?

22

22