hiperc
hiperc copied to clipboard
usnistgov
High Performance Computing Strategies for Boundary Value Problems
- [ ] CUDA code looks weird due to thread allocation model - Traditional: many memory addresses per core -> looping constructs - Accelerator: one memory address per thread ->...
On Debian Stretch, `opencl-headers` was insufficient: users will also need the appropriate ICD libraries, *e.g.* for MESA: - `ocl-icd-dev` - `ocl-icd-libopencl1` - `ocl-icd-opencl-dev`
It says the page doesn't exist yet.
How about also adding C++11 threads into the mix? I've had quite good experience with them (I have a forthcoming _J. Res. NIST_ paper on that topic, and our other...
http://on-demand.gputechconf.com/gtc/2016/presentation/s6510-jeff-larkin-targeting-gpus-openmp.pdf
in gpu/cuda, I run ./diffusion ../params.txt The code appears to execute. PNGs and CSVs are generated. However, it looks like no time marching is occurring. The attached image is the...
The tiled convolution algorithm is the heart of the accelerator codes. Explain it as clearly as possible.
- grids - mask - bc - tiles - Discuss A[y][x] vs. A[y*nx + x] - Origin (0,0) is in the bottom-left for grids and the convolution mask
- [x] [readthedocs](https://readthedocs.org) - [x] [gitter](http://gitter.im) - [ ] [reviewable](http://reviewable.io)