Improving GPU Throughput of Reservoir Simulations using NVIDIA MPS & MIG

Bg resource
Synopsis

A short presentation explaining how ECHELON software, using NVIDIA's multi-processing service and multi-GPU, improves the performance of reservoir simulations.

This presentation was originally presented at the Fifth EAGE Workshop on High Performance Computing for Upstream which took place online from 6-8 September 2021. the technical paper can be downloaded from EarthDoc.

Additional resources

Graphics Processing Unit–Based, Industrial Grade Compositional Reservoir Simulator →
An Optimization-Based Facility Network Solver For Well Allocation In Reservoir Simulation →
Improving GPU Throughput of Reservoir Simulations using NVIDIA MPS & MIG →
ECHELON: SPE10 Case Study →
A Revolution in Performance using a GPU Based Simulator →
Full-GPU Reservoir Simulation Delivers on its Promise for Giant Carbonate Fields →
GPU Acceleration of Equation of State Calculations in Compositional Reservoir Simulation →
Unconventional Reservoir Model Predictions Using Massively-Parallel Flow-Simulation →
A High Performance Reservoir Simulator on GPU →
Unconventional Reservoir Model Predictions Using Massively-Parallel Flow-Simulation: Bakken/Three Forks Reservoir Development Cases and SRV Testing →
Recent Trends in High Performance Computing: Implications for Reservoir Simulation →
Unconventional Reservoir Model Predictions Using Massively-Parallel GPU Flow-Simulation →
Accelerating Tight Reservoir Simulation Workflows With GPUs →
Realizing the Potential of GPUs for Reservoir Simulation →
A GPU Accelerated Aggregation Algebraic Multigrid Method →
GAMPACK (GPU Accelerated Algebraic Multigrid Package) →
A Novel GPGPU Approach to Kirchhoff Time Migration →
Survey of Two-Point Seismic Travel Time Methods →
Accelerating Reservoir Simulation with GPUs →
Recent Trends in High Performance Computing: Implications for Reservoir Simulation →
ECHELON 2.0 Software Overview And Case Studies Webinar →