T
Thomas L. Falch
Researcher at Norwegian University of Science and Technology
Publications - 11
Citations - 342
Thomas L. Falch is an academic researcher from Norwegian University of Science and Technology. The author has contributed to research in topics: Software portability & Compiler. The author has an hindex of 6, co-authored 11 publications receiving 310 citations.
Papers
More filters
Journal ArticleDOI
Medical image segmentation on GPUs – A comprehensive review
Erik Smistad,Erik Smistad,Thomas L. Falch,Mohammadmehdi Bozorgi,Anne C. Elster,Frank Lindseth,Frank Lindseth +6 more
TL;DR: It is concluded that most segmentation methods may benefit from GPU processing due to the methods' data parallel structure and high thread count, however, factors such as synchronization, branch divergence and memory usage can limit the speedup.
Proceedings ArticleDOI
Machine Learning Based Auto-Tuning for Enhanced OpenCL Performance Portability
Thomas L. Falch,Anne C. Elster +1 more
TL;DR: This paper uses machine learning-based auto-tuning to address poor performance portability in heterogeneous computing, and builds an artificial neural network based model that achieves a mean relative error as low as 6.1%, and is able to find configurations as little as 1.3% worse than the global minimum.
Journal ArticleDOI
Machine learning-based auto-tuning for enhanced performance portability of OpenCL applications
Thomas L. Falch,Anne C. Elster +1 more
TL;DR: In this article, a machine learning-based auto-tuning approach is proposed to find interesting subspaces for further search, which can then be used to find the optimal tuning subspace.
Proceedings ArticleDOI
Register Caching for Stencil Computations on GPUs
Thomas L. Falch,Anne C. Elster +1 more
TL;DR: In this article, the registers of multiple threads are combined and used as a shared, last level, manually managed cache for the contributing threads, enabling threads in the same warp to exchange data directly.
Journal ArticleDOI
GPU-Accelerated Visualization of Scattered Point Data
TL;DR: This paper presents a method for visualizing scattered point data sets based on volume ray casting, and distinguishes itself by operating directly on the unstructured samples, rather than resampling them to form voxels.