HRF-relaxed: Adapting HRF to the complexities of industrial heterogeneous memory models
- Submitting institution
-
University of the West of England, Bristol
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 836451
- Type
- D - Journal article
- DOI
-
10.1145/2701618
- Title of journal
- ACM Transactions on Architecture and Code Optimization
- Article number
- -
- First page
- 1
- Volume
- 12
- Issue
- 1
- ISSN
- 1544-3566
- Open access status
- Out of scope for open access requirements
- Month of publication
- April
- Year of publication
- 2015
- URL
-
http://dx.doi.org/10.1145/2701618
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
2
- Research group(s)
-
-
- Citation count
- 7
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- This paper formalises weak memory models of heterogeneous architectures, in particular, GPUs. Building on the Data-Race Free memory model of languages such as Java and C++, this work bridges the gap between CPU-centric assumptions in them to the realities of CPUs, GPUs, DSPs, and more. At the time the Khronos standards group was developing OpenCL 2.0 that initially lacked a formal definition of the proposed memory model. This work provided that formalization. Additionally, it exposed a number of shortcomings with the original OpenCL 2.0 model, with proposed changes that were adopted in a revision of the standard.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
