During SC14 (SuperComputing Conference 2014), OpenCL is again all over New Orleans. Just like last year, I’ve composed an overview based on info from the Khronos website and the SC2014 website.
Finally I’m attending SC14 myself, and will give two talks for you. Tuesday I’ll be part of a 90 minute session of Khronos, where I’ll talk a bit about GROMACS and selecting the right accelerator for your software. Wednesday I’ll be sharing our experiences from our port of GROMACS to OpenCL. If you meet me, then I can hand you over a leaflet with the decision chart to help select the best device for the job.
Schedule
Sunday, 16 November | |||||
8:30 – 17:00 | Tutorial | From “Hello World” to Exascale Using x86, GPUs and Intel Xeon Phi Coprocessors (probably no or little OpenCL) | Robert M. Farber, Greg Scantlen, Timothy Thomas, Andree Jacobson | 396 | |
Monday, 17 November | |||||
8:30 – 17:00 | Tutorial | OpenCL: A Hands-On Introduction | Tim Mattson, Alice Koniges, Simon McIntosh-Smith | 395 | |
16:45 – 17:30 | Workshop Talk | Implementing the OpenCL SYCL Shared Source C++ Programming Model using Clang/LLVM (“The LLVM Compiler Infrastructure in HPC” workshop) | Gordon Brown | ?? | |
Tuesday, 18 November | |||||
11:30 – 12:00 | Exhibitor Forums | KALRAY TURBOCARD2 Scalable Accelerator: An Efficient Supercomputer Cluster on a Board | Benoît Ganne | 292 | |
11:30 – 12:00 | Papers | Managing DRAM Latency Divergence in Irregular GPGPU Applications | Niladrish Chatterjee, Mike O’Connor, Gabriel H. Loh, Nuwan Jayasena, Rajeev Balasubramonian | 391-92 | |
13:30 – 14:00 | Papers | Practical Symbolic Race Checking of GPU Programs | Peng Li, Guodong Li, Ganesh Gopalakrishnan | 388-89-90 | |
14:00 – 14:30 | Papers | Scalable Kernel Fusion for Memory-Bound GPU Applications | Mohamed Wahib, Naoya Maruyama | 388-89-90 | |
14:00 – 14:30 | Papers | Pipelining Computational Stages of the Tomographic Reconstructor for Multi-Object Adaptive Optics on a Multi-GPU System | Ali Charara, Hatem Ltaief, Damien Gratadour, David Keyes, Arnaud Sevin, Ahmad Abdelfattah, Eric Gendron, Carine Morel, Fabrice Vidal | 391-92 | |
17:15 – 19:00 | Poster | GPU Acceleration of Small Dense Matrix Computation of the One-Sided Factorizations | Tingxing Dong, Mark Gates, Azzam Haidar, Piotr Luszczek, Stanimire Tomov | New Orleans Theater Lobby | |
17:15 – 19:00 | Poster | Hardware Accelerated Linear Programming: Parallelizing the Simplex Method with OpenCL | Bradley de Vlugt, Maysam Mirahmadi, Serguei L. Primak, Abdallah Shami | New Orleans Theater Lobby | |
17:30 – 19:00 | OpenCL BOF | OpenCL 2.0 and beyond | Tim Mattson, Simon McIntosh-Smith, Vincent Hindriksen | 275-76-77 | |
Wednesday, 19 November | |||||
14:30 – 14:45 | Talk | Using the SYCL for OpenCL open standard to accelerate C++ code | Andrew Richards | Booth 839 (AMD) | |
15:15 – 15:45 | Talk | Porting Gromacs from CUDA to OpenCL. Lessons learned | Vincent Hindriksen | Booth 839 (AMD) |
There are other interesting SC14-events around OpenCL, so be sure to check the schedule carefully – especially the posters!
Booths
Below links bring you to special SC14 landing-pages.
- Altera -FPGAs
- Booth #3821. Altera will demo of rapid prototyping using OpenCL, and a demo of optical flow on an embedded SoC.
- Booth #3921: Bittware will demo the available tools necessary to begin developing applications for the Altera Stratix V using OpenCL.
- Booth #1332: Nallatech has lots and lots of OpenCL. Ask for demo’s and read the big poster.
- AMD – FirePro GPUs.
- booth #839: Demo of Gromacs. Ping me for a personal demo.
- Small theatre at booth #839 will have several OpenCL-related talks (see also agenda above).
- ARM – low-power HPC
- Booth 3458: Not sure, if they have OpenCL demo’s, but good chance there are.
- Intel – CPUs, Accelerators
- Booth #1215, #1315. Most times don’t demo OpenCL at HPC events, but they can tell about it.
- NVIDIA -Tesla GPUs
- Booth 1727: will not present OpenCL, but every time they’re funny when asked what they do about OpenCL. So go visit them and ask about i.e. OpenCL 2.0.
- Texas Instruments – DSP
- Booth 3745: Not sure if they’ll demo their OpenCL solution, but hope so. Will update, once I know.
- Xilinx – FPGAs
- Booth 3903, 4003. Will present “the industry leading OpenCL solution”. Be sure to check out and compare with Altera’s solution.
- Kalray – multi-core processor
- Booth 2815. Will demo a new processor and have OpenCL-demo’s (alpha product).
- ArrayFire – HPC library
- Booth #2725. will demo their library with backends for both OpenCL and CUDA.
- Mont-Blanc – ARM-based HPC
- Booth #1039. Will show literally cool supercomputers.
The floorplan with accents on the Khronos members can be found at Khronos’s site, as PDF.
At SC14 and saw great OpenCL demos or news?
Share this info and photos in the comments, for others to pick up.
OpenCL related SC14 photos on Flickr: https://www.flickr.com/photos/khronos/sets/72157649337728696/
Do you have links to videos of any these talks and/or your own presentation?
I’m not aware of videos – there is an overview on text: http://www.hpcwire.com/2014/11/22/supercomputing-wrap-top-stories-sc14/
To be honest, my talk at the OpenCL BOF did not go that well – so luckily it’s not on video. 🙂
The points I made during my talk at ther AMD booth, I’ll be sharing on the blog together with the results of the benchmarks.
I overheard many conversations at SC ’14 about the dislike for CUDA and the preference of OpenCL. Not only has NVidia not supported OpenCL since 2010, it only supports version 1.1 with C bindings. Doesn’t support 2.0, C++, or SPIR, all of which contain important enhancements. Clearly a strong-arm tactic. You can purchase a, AMD FirePro GPU, equivalent to an NVidia Kepler, for 1/3 the price!
Heard the same things, but thought it was because most people know me for supporting OpenCL. Nice to hear, you heard it too!
The success of OpenCL on AMD FirePro is crucial for OpenCL in general. Then we make sure the competition is on OpenCL-capable hardware, not between CUDA and OpenCL. I’m therefore very happy FirePro won the Green500! If AMD succeeds in getting a few more clusters in the Top500 or Green500, then we’re there!