When will Jacket be merged into Parallel Computing Toolbox?

I read that MathWorks has acquired the competing Jacket product from Accelereyes:

http://blog.accelereyes.com/blog/2012/12/12/exciting-updates-from-accelereyes/

We were considering using Jacket for the GPU interface instead of Parallel Computing Toolbox since multiple articles indicate that Jacket is faster and supports more MATLAB functions than PCT. But now that Jacket is part of the MathWorks family I would expect its functionality to get merged into PCT.

Does anyone know if this is the case? I don't see anything on the MathWorks site mentioning the acquisition or any timeline for updating PCT with Jacket features.

BTW I am currently using R2012b. Do I perhaps need to move up to R2013a?

Thanks for any info.

[Walter Roberson]

The article does not indicate that MathWorks has acquired Jacket; it indicates that the two companies are "working together" (in some unspecified way.)

[Terry McKiernan]

Walter, thanks for your reply. Maybe we are interpreting what it says in the article differently.

In the 2nd paragraph of the article it states:

"AccelerEyes has sold Jacket as a 3rd-party add-on to the MathWorks MATLAB® product."

I read this to say "Accelereyes sold Jacket to MathWorks", since immediately after that it says they are no longer selling new Jacket licenses. But I guess it could instead be interpreted as simply a statement of historical fact, i.e. "in the past, Jacket was sold (to end users) as an add-on to MATLAB".

Either way, have you heard anything about future plans for MATLAB involving Jacket features?

Thanks

[Walter Roberson]

I have not heard anything.

I think the statement has to be interpreted historically, as you cannot sell a product to a product, only to a company.

[Ben Tordoff]

Hi Terry,

to me the most important question is whether PCT already supports the functions that you need. If you could describe which functions you need on the GPU or what MATLAB functionality you typically use in your work, we can work out whether PCT will do the job for you.

Cheers

Ben

[Stefan]

Hi Ben,

I pointed out what I need from MATLABs GPU computing solution here:

http://stackoverflow.com/questions/16943670/when-is-the-accelereyes-jacket-functionality-implemented-into-matlab

Basically, MATLABs built in GPU functions are implemented very poorly and are slow - I am ending up reimplementing most of them myself. Also sparse matrices are not supported.

Jacket was a much better product. However, it looks like Mathworks has sued Accelereyes (using some ridicuoulous accusations) and forced them to discontinue there product.

Otherwise I really like MATLAB - but this was a bad move and costs me a lot of time.

Regards, Stefan

[Michal Kvasnicka]

A little bit more info: here and here

[Christopher]

Are there any updates on this situation?

[Nick Chng]

Still waiting for an update. I miss Jacket's gfor and interp3 implementations, which remain faster, and with more functionality, in R2012b than they are in R2015b.

Answer 1

[Gareth Thomas]

One way would be to look at the Parallel computing toolbox release notes since R2012b.

http://www.mathworks.nl/help/distcomp/release-notes.html

As R2013a, R2013b, R2014a are now out.

The number of functions supported grows every release.

Answer 2

[Rob Campbell]

It seems they're now at http://arrayfire.com/ and are open source.

[Walter Roberson]

The phone numbers do match.