Change search
ReferencesLink to record
Permanent link

Direct link
Acceleration of Distance-to-Default with GPU
KTH, School of Information and Communication Technology (ICT).
2012 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

Distance-to-Default(DTD), which is used to describe the default risk of a rm, acts an important role in credit research. Nowadays, since we can access a large amount of historical data, we can get a more accurate DTD result. However, this directly increases the computation time as well as the computation power. Meanwhile, Graphic Processing Unit(GPU), with its tremendous capability of parallel computing, becomes more and more popular in High Performance Computing. In addition, CUDA, a parallel computing platform and programming model that invented by Nvidia, makes GPU programming much more easier and faster. So GPU can be a proper candidate to accelerate the DTD processing if we can ooad the most computation intensive part to the GPU.

In this thesis project, the author explore the existing DTD program(provided by RMI, NUS), analyze the algorithm and ooad the most computation intensive part to GPU using CUDA. The platform used for testing consists of an 2.4GHz Intel i5 CPU and a Nvidia GeForce GTX460 GPU. The GPU is used as a co-processor, which is connected to the mother board with PCIe2 bus.

Place, publisher, year, edition, pages
2012. , 34 p.
Trita-ICT-EX, 2012:189
National Category
Engineering and Technology
URN: urn:nbn:se:kth:diva-102877OAI: diva2:557179
Educational program
Master of Science - System-on-Chip Design
Available from: 2012-10-22 Created: 2012-09-27 Last updated: 2012-10-22Bibliographically approved

Open Access in DiVA

fulltext(659 kB)919 downloads
File information
File name FULLTEXT01.pdfFile size 659 kBChecksum SHA-512
Type fulltextMimetype application/pdf

By organisation
School of Information and Communication Technology (ICT)
Engineering and Technology

Search outside of DiVA

GoogleGoogle Scholar
Total: 919 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Total: 68 hits
ReferencesLink to record
Permanent link

Direct link