1 / 12

Key Issues

Key Issues. Colin Cunningham Chair, Opticon Key Technology Network. Are we repeating mistakes of history?. CCD controllers?. Hardware-independent software – is it achievable?. Is OpenCL the answer? Dubious Not yet ready? But worth looking at

raziya
Download Presentation

Key Issues

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Key Issues Colin Cunningham Chair, Opticon Key Technology Network

  2. Are we repeating mistakes of history? • CCD controllers?

  3. Hardware-independent software – is it achievable? • Is OpenCL the answer? • Dubious • Not yet ready? • But worth looking at • Other suppliers are following in line (AMD etc) • 10-20% performance hit • Need to wait for maturity • What will happen in the next few years

  4. Balance CPU/FPGA/GPU/NBT • NBT – next big thing • Someone has to come up with something to fit most peoples needs • Communication • FPGAs and CPUs on same die • Socket fillers • Computation • Software bloat? • Optimising algorithms for parallelisation • Finding bottlenecks • Compiler optimisation – Assembly?

  5. Can we build a community? • Develop or use common tools • DARC? • SPARTA? • Set standards • Negotiate with industry • Share best (and worst!) practice • E.g. CCD world email list? • CfAO fall retreat

  6. What common tools and standards do we need? • Bench marks • Common benchmark e.g. Whetstone? • 2 types: • Has to perform from AO point of view (equal to MVM MAP) • Better computationally • Different systems will have different metrics • Library of algorithms • Wiki AO • Definitions – e.g. Latency (includes settling time? Camera readout?) • Turbulence Test Data • Well defined datasets for benchmarking with • IP cores for FPGAs • Make open source? • No warrenty • OpenCores, openFPGA communities • ?

  7. Can we work with other communities? • Radio astronomy • ALMA • SKA • Invite to future workshops • Games • Scope here • Other with similar requirements • Low-latency/high frequency trading? • HPC • Cyber security • Image processing • Machine vision • Continental power transmission • Fusion community • Navigation systems? • ?

  8. Who can coordinate? • In Europe • ESO • OPTICON • Worldwide • SPIE$$$ • ? • Students? • A committee of people... ACTION

  9. Ideas for future workshops • Algorithms • Deformable mirrors and associated technology • What hardware (FPGA) are people using? • Demonstrations • Lessons learned • Current AO systems • Benchmark problem – and then a (competitive?) session based around this? • If it doesn’t work, don’t turn up? • People showing Hermine how on-sky results have been used to verify models • One button operation • From proof of concept to operational availability • Systems engineering • ?

  10. Gaps in Technology? • Low latency communication everywhere • With high throughput • High memory bandwidths • 100GHz CPUs • Quantum computing • Linux GPU tools • FPGA openCL • High level programming tools for FPGAs that give reasonable performance • As good as hand coded... • Magic mapping tools • C -> FPGA • Direct access to GPUs • As a community... • Parallel competent high level languages

  11. Or should we just wait to see what comes next? Especially for ELTs? NBT is now in the lab somewhere? Analogue computing?

  12. Specific Actions What Who

More Related