Saturday, January 28, 2017 3:14:30 PM
- For the Parallel Computing Toolbox (PCT), essentially, the number of cores your CPU has locally implies the number of identical compute resources that are at hand. If yours is a two core machine then we can divide independent work between two workers who can work on identical data and hence achieve speedup.
- ‘parpool’ command starts a parallel pool locally equal to the number of cores on your machine. ‘delete(gcp)’ deletes the pool. Where basically you get an object of class parallel.Pool which has various methods and attributes that can be configured.
- There is no guarantee of sequence between the workers themselves. The parpool object will have properties that can be set such as adding listeners, attaching files, setting the timeout property etc. More info on the ‘parfor’ can be found here.
- The ‘batch’ command can be used to offload work to another matlab session (worker) that will run in the background. Note that this command is non blocking and you can continue to do other work with your current matlab session and wait for the offloaded work to come back with the results.
- If you want to keep working in matlab but want some result to be obtained by offloading the work to a worker / matlab session in the background and just use the result once its done then ‘batch’ is the right command for this. Note that batch does not require a parallel pool to be opened.
- ‘batch’ can be used with ‘parfor’ to run a loop in a parallel pool and offload a job. For more info look here.
Saturday, December 24, 2016 11:09:07 PM
- Matlab is essentially the language of choice for technical computing. At the core, it is really based on and optimized for linear algebra and matrices.
- Simulink on the other hand is based on ordinary and partial differential equations. Its highly used for Model Based Design and Simulations where time is the varying quantity.
- Linear algebra and differential equations are two foundational and key areas of mathematics.
- Matlab code quite literally powers the world: smartphones, satellites, airplanes, cars et al. You name the industry and Matlab code is quietly and reliably enabling those systems and industries.
- Matlab is also the language of choice for scientists and engineers worldwide. With an integrated development environment, best in class graphic visualization capabilities and toolboxes which can support almost every major computing need, Matlab can be your one stop solution.
- Some popular areas for Matlab include machine learning, signal processing, image processing, computer vision, communications, computational finance, control design, robotics et al.
- With Matlab there’s no need to re-invent the wheel. Have a problem to solve? Its likely that Matlab has the functions, algorithms and toolboxes you will need.
- With its best in class technical support, you can be assured that you will never walk alone. MathWorks will ensure you are covered for all your Matlab needs.
- Matlab code can run in parallel, scale up to clusters and production servers and the cloud while easily interfacing with other languages and software systems / libraries.
- MathWorks has about 90 products that address various industries and technical problem areas.