Efficient implementation of linprog

I am using linprog to solve a linear optimization problem with thousands of constraints. I have noticed that for constraints more than a few thousand the matlab implementation of linprog performs very slow. Are there any GPU(CUDA) or parallel implementations of linprog available within Matlab or elsewhere ?

[Matt J]

What are the dimensions of your problem? And which algorithm have you selected? The default interior point algorithm?

I don't know of any parallel implementations. For that matter, I can't immediately see whether the computations in these algorithms have much parallel structure to begin with.

[John D'Errico]

People get spoiled. They have this immensely powerful computer on their lap, and they want it to be infinitely fast on any problem they throw at it. And if it runs fine with only 1000 variables, then they want it to run just as quickly with 10000 or 1000000 variables. If not, then there must be some easy fix.

Too many movies and TV shows where the computers do everything immediately.

Computers have a finite capability. And no matter how large that finite capability is, there will always be someone who decides they need to make their problem just slightly larger. After all, if the problem was trivial, then how could you call it research?

I think Matt has a good point. I'm not sure this is a code that would be simply or usefully parallelized.

[Shiv Tewari]

https://parallellp.wordpress.com/

[William Handler]

Not sure why Enrico is so down on your question. I would also be interested in a parallel version of linprog, dod you get anywhere with it?