MPP LS-DYNA Intermittent Profiling Information

During MPP LS-DYNA execution, often there is a need to know how each of the “threads” are doing in terms of overal CPU utilization. Until now, one has to wait till the job is fully finished so we can review the load-balance information from the main D3HSP file. From LS-DYNA 971 r7 onwards, there is…

September 26, 2012 | by

Hybrid LS-DYNA

With the recent growth of multi-core chips, scalibility of pure MPP LS-DYNA beyond 128 cores has shown degration due to several factors including latency and network communications. LSTC has recently developed a new code named “Hybrid LS-DYNA” that provides sustained scalibility for large number of cores and also yields digit-2-digit repeatibility when changing the number…

October 21, 2010 | by

RCB Decomposition and LS-DYNA Built-In Intelligence

Domain decomposition is central to MPP-LSDYNA as it determines the final load balance on each of the compute nodes. The default (and most widely used) decomposition method is the Recursive Coordinate Bisection (RCB) which recursively halves the model using the longest dimension of the input model. army assessment risk dodge car accessory pipe water ochelari…

February 13, 2008 | by

MPP and Deformable to Rigid Switching

You may have read the previous post on deformable to rigid switching which works by turning a predefined set of deformable components into a rigidbody at a user-defined time to save on element processing costs. This may not necessarily be true when running using any of the MPP-LSDYNA executables since the domain decomposition routines do…

February 19, 2007 | by

Load Balance in FMVSS 207/210 Simulations for MPP

Analyses involving automotive seat designs for loadcases such as FMVSS 207/210 has shown some poor load balance on compute nodes when using the default decomposition in MPP LS-DYNA. In FMVSS 207/210 type simulations, the computationally expensive portion of the model lies at the seat and its immediate viscinity while the rest of the model involves…

February 16, 2007 | by

Simulation Model Decomposition Using Recursive Coordinate Bisection (RCB) Method

In the area of distributed computing using Massively Parrallel Processing (MPP) LS-DYNA, finite element model decomposition is performed after initial processing of the input deck to “distribute” the model content to compute nodes. There are two primary goals for model decomposition. First goal is of of course to “break-down” the given problem into smaller pieces…

September 18, 2006 | by