Engineering topics
Publications and source records attributed to Dube, Evi.
A range of programming models (and the languages that constitute the practical implementation of these models) exist to address parallel programming challenges. At present, however, one model dominates LLNL application codes: SPMD message-passing using MPI for internode communication, and occasionally, OpenMP for intranode parallelism.
As supercomputing moves toward exascale, node architectures will change significantly. CPU core counts on nodes will increase by an order of magnitude or more. Heterogeneous architectures will become more commonplace, with GPUs or FPGAs providing additional computational power. Novel programming models may make better use of on-node parallelism in these new architectures than do current models. In this paper we examine several of these novel models – UPC, CUDA, and OpenCL –to determine their suitability to LLNL scientific application codes. Our study consisted of several phases: We conducted interviews with code teams and selected two codes to port; We learned how to program in the new models and ported the codes; We debugged and tuned the ported applications; We measured results, and documented our findings. We conclude that UPC is a challenge for porting code, Berkeley UPC is not very robust, and UPC is not suitable as a general alternative to OpenMP for a number of reasons. CUDA is well supported and robust but is a proprietary NVIDIA standard, while OpenCL is an open standard. Both are well suited to a specific set of application problems that can be run on GPUs, but some problems are not suited to GPUs. Further study of the landscape of novel models is recommended.
A programming model is a set of software technologies that support the expression of algorithms and provide applications with an abstract representation of the capabilities of the underlying hardware architecture. The primary goals are productivity, portability and performance.
What follows is a response to the questions posed in the May 21, 2010 document "Assigned Commissioner's Ruling and Scoping Memo, Phase 1", pertaining to Rulemaking 09-11-014, also referred to as the "Scoping Memo". The questions refer to the specific components of the CPUC Energy Efficiency (EE) Evaluation, Measurement, and Verification (EM&V) program, currently in its 2010-2012 cycle. In the Scoping Memo, there was reference to the need to update the policy and methodological framework of EM&V, due to changes in law (AB32), policy (California Long Term Energy Efficiency Strategic Plan), the availability of new data, and challenges to previous EM&V cycles.
At the request of Commissioner Dian Grueneich, Lawrence Livermore National Laboratory (LLNL) has reviewed the Evaluation, Measurement and Validation (EM&V) practices of the California Public Utility Commission (CPUC) Energy Efficiency (EE) Program. LLNL's examination was performed through reviews of documents written by the CPUC, namely decisions, rulings, and Energy Division staff reports, as well as other reports written by experts in energy efficiency. This document outlines our general findings from our preliminary investigation. We have identified and consolidated the intrinsic problems and issues that currently reside in the EM&V and RRIM mechanisms. The next step in our involvement would be to assist the CPUC in restructuring the program to enhance the overall value to the State.