GPU computing from Python - PyOpenCL - wednesday instead of Friday this week Event
- Time:
- 16:00 - 17:00
- Date:
- 29 June 2011
- Venue:
- Building 85, Room 2207
Event details
Increased computational power is currently achieved by parallel architectures such as multi-core CPUs and multi-CPU computers -- not an increase of clock rate as in the recent past. Probably the most recent notable trend is the use of graphical processing units (GPUs) to carry out numerical calculations. GPUs have many (of the order of 100) computational cores are can carry out specialised operations in parallel at rates that are 100 faster than the computer's main CPU. The GPU hardware, targeted at gamers and home consumers tends to be relatively cheap. In this talk, I demonstrate how code can be executed on GPUs using OpenCL from Python. OpenCL a language which enables programmers to run the same code on different hardwares and is thus manufacturer independent. It is an open standard and helps writing portable code. I will give a short introduction to PyOpenCL, showing how very easy it is to write programs, that run on GPUs or CPUs, from Python.
http://cmg.soton.ac.uk/events/event-349/
Complex Systems Simulation Seminar Series (CS^4)
from the Institute for Complex Systems Simulation , the Complexity in Real-World Contexts USRG , and the Computational Modelling Group .
PLEASE NOTE THIS WEEK'S SEMINAR IS ON WEDNESDAY INSTEAD OF FRIDAY.
Abstract
Increased computational power is currently achieved by parallel architectures such as multi-core CPUs and multi-CPU computers -- not an increase of clock rate as in the recent past.
Probably the most recent notable trend is the use of graphical processing units (GPUs) to carry out numerical calculations. GPUs have many (of the order of 100) computational cores are can carry out specialised operations in parallel at rates that are 100 faster than the computer's main CPU. The GPU hardware, targeted at gamers and home consumers tends to be relatively cheap.
In this talk, I demonstrate how code can be executed on GPUs using OpenCL from Python.
OpenCL a language which enables programmers to run the same code on different hardwares and is thus manufacturer independent. It is an open standard and helps writing portable code.
I will give a short introduction to PyOpenCL, showing how very easy it is to write programs, that run on GPUs or CPUs, from Python.
Speaker
CHANGE OF LOCATION
PLEASE NOTE: THIS WEEK WE ARE IN BUILDING 85 / ROOM 2207.
Refreshments
Available from 3:30pm, lecture starts at 4pm.
Complex Systems Simulation Seminar Series
For the complete CS^4 schedule please click here: http://www.interdisciplinary.soton.ac.uk/cs4.html