Review of LLVM Compiler Architecture Enhancements for CUDA
Keywords:
Graphical Processing Unitt, Compute Unified Device Architetcure, LLVM CompilerAbstract
Heterogeneous platforms are now becoming increasing omnipresent due to the availability of multicores at commodity prices. In order to benefit from the immense parallel capability of these multicores, more and more applications are now being developed as well as ported using the CUDA framework that programs these cores. Hence it becomes imperative to devise new compiler paradigms that cater varied languages and provide easy and flexible multicore programming. LLVM compilers have traditionally been the bedrock for such endeavors. New runtime processes are being researched to make CUDA platform more amenable for supporting a maximum set of language architectures. An analysis is made to understand the advantages and the accompanying pitfalls of various types of linking techniques.
References
Chris Lattner, Vikram Adve, LLVM: A Compilation Framework for Lifelong Program Analysis & Transformation, University of Illinois at Urbana-Champaign.
Yuan Lin, Building GPU Compilers with libNVVM, GPU Technology Conference.
The LLVM Compiler Infrastructure, http://llvm.org.
Xiang Zhang, Aaron Brewbaker, How to design a language integrated compiler with LLVM, GTC 2014, San Jose, California, March 27, 2014.
Dmitry Mikushin, Nikolay Likhogrud, Eddy Z. Zhang, KERNELGEN – The design and implementation of a next generation compiler platform for accelerating numerical models on GPUs, Parallel & Distributed Processing Symposium Workshops (IPDPSW), 2014 IEEE International.
Adam DeConinck, HPC Systems Engineer, NVIDIA, Introduction to CUDA Toolkit for Building Applications.
Downloads
Issue
Section
License
- Papers must be submitted on the understanding that they have not been published elsewhere (except in the form of an abstract or as part of a published lecture, review, or thesis) and are not currently under consideration by another journal published by any other publisher.
- It is also the authors responsibility to ensure that the articles emanating from a particular source are submitted with the necessary approval.
- The authors warrant that the paper is original and that he/she is the author of the paper, except for material that is clearly identified as to its original source, with permission notices from the copyright owners where required.
- The authors ensure that all the references carefully and they are accurate in the text as well as in the list of references (and vice versa).
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Attribution-NonCommercial 4.0 International that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
- The journal/publisher is not responsible for subsequent uses of the work. It is the author's responsibility to bring an infringement action if so desired by the author.
