GPU Predictor-Corrector Interior Point Method for Large-Scale Linear Programming
This master’s thesis concerns the implementation of a GPUaccelerated version of Mehrotra’s predictor-corrector interior point algorithm for large-scale linear programming (LP). The implementations are tested on LP problems arising in the financial industry, where there is high demand for faster LP s...
Main Author: | |
---|---|
Format: | Others |
Language: | English |
Published: |
KTH, Numerisk analys, NA
2015
|
Online Access: | http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-168672 |
id |
ndltd-UPSALLA1-oai-DiVA.org-kth-168672 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-UPSALLA1-oai-DiVA.org-kth-1686722015-06-09T04:55:02ZGPU Predictor-Corrector Interior Point Method for Large-Scale Linear ProgrammingengGPU-accelererad inrepunktsmetod för storskalig linjärprogrammeringRydberg, DavidKTH, Numerisk analys, NA2015This master’s thesis concerns the implementation of a GPUaccelerated version of Mehrotra’s predictor-corrector interior point algorithm for large-scale linear programming (LP). The implementations are tested on LP problems arising in the financial industry, where there is high demand for faster LP solvers. The algorithm was implemented in C++, MATLAB and CUDA, using double precision for numerical stability. A performance comparison showed that the algorithm can be accelerated from 2x to 6x using an Nvidia GTX Titan Black GPU compared to using only an Intel Xeon E5-2630v2 CPU. The amount of memory on the GPU restricts the size of problems that can be solved, but all tested problems that are small enough to fit on the GPU could be accelerated. Detta masterexamensarbete behandlar implementeringen av en grafikkortsaccelererad inrepunktsmetod av predictor-corrector-typ för storskalig linjärprogrammering (LP). Implementeringarna testas på LP-problem som uppkommer i finansbranschen, där det finns ett stort behov av allt snabbare LP-lösare. Algoritmen implementeras i C++, MATLAB och CUDA, och dubbelprecision används för numerisk stabilitet. En prestandajämförelse visade att algoritmen kan accelereras 2x till 6x genom att använda ett Nvidia GTX Titan Black jämfört med att bara använda en Intel Xeon E5-2630v2. Mängden minne på grafikkortet begränsar problemstorleken, men alla testade problem som får plats i grafikkortsminnet kunde accelereras. Student thesisinfo:eu-repo/semantics/bachelorThesistexthttp://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-168672TRITA-MAT-E ; 2015:37application/pdfinfo:eu-repo/semantics/openAccess |
collection |
NDLTD |
language |
English |
format |
Others
|
sources |
NDLTD |
description |
This master’s thesis concerns the implementation of a GPUaccelerated version of Mehrotra’s predictor-corrector interior point algorithm for large-scale linear programming (LP). The implementations are tested on LP problems arising in the financial industry, where there is high demand for faster LP solvers. The algorithm was implemented in C++, MATLAB and CUDA, using double precision for numerical stability. A performance comparison showed that the algorithm can be accelerated from 2x to 6x using an Nvidia GTX Titan Black GPU compared to using only an Intel Xeon E5-2630v2 CPU. The amount of memory on the GPU restricts the size of problems that can be solved, but all tested problems that are small enough to fit on the GPU could be accelerated. === Detta masterexamensarbete behandlar implementeringen av en grafikkortsaccelererad inrepunktsmetod av predictor-corrector-typ för storskalig linjärprogrammering (LP). Implementeringarna testas på LP-problem som uppkommer i finansbranschen, där det finns ett stort behov av allt snabbare LP-lösare. Algoritmen implementeras i C++, MATLAB och CUDA, och dubbelprecision används för numerisk stabilitet. En prestandajämförelse visade att algoritmen kan accelereras 2x till 6x genom att använda ett Nvidia GTX Titan Black jämfört med att bara använda en Intel Xeon E5-2630v2. Mängden minne på grafikkortet begränsar problemstorleken, men alla testade problem som får plats i grafikkortsminnet kunde accelereras. |
author |
Rydberg, David |
spellingShingle |
Rydberg, David GPU Predictor-Corrector Interior Point Method for Large-Scale Linear Programming |
author_facet |
Rydberg, David |
author_sort |
Rydberg, David |
title |
GPU Predictor-Corrector Interior Point Method for Large-Scale Linear Programming |
title_short |
GPU Predictor-Corrector Interior Point Method for Large-Scale Linear Programming |
title_full |
GPU Predictor-Corrector Interior Point Method for Large-Scale Linear Programming |
title_fullStr |
GPU Predictor-Corrector Interior Point Method for Large-Scale Linear Programming |
title_full_unstemmed |
GPU Predictor-Corrector Interior Point Method for Large-Scale Linear Programming |
title_sort |
gpu predictor-corrector interior point method for large-scale linear programming |
publisher |
KTH, Numerisk analys, NA |
publishDate |
2015 |
url |
http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-168672 |
work_keys_str_mv |
AT rydbergdavid gpupredictorcorrectorinteriorpointmethodforlargescalelinearprogramming AT rydbergdavid gpuaccelereradinrepunktsmetodforstorskaliglinjarprogrammering |
_version_ |
1716805101557907456 |