IBM Open Sources Multi-core Smalltalk VM | Hacker News Reader

IBM Open Sources Multi-core Smalltalk VM

github.com

111 pointstialys13 comments

Open articleView on HN

Is anyone here using Smalltalk for a production application? What are your thoughts on this VM?

It's still at the research stage. It won't be production-ready for a while.

I got all excited, and then:

"Hosting an Object Heap on Manycore Hardware: An Exploration"

and

"Garbage collector is as simple as possible it is neither concurrent nor parallel performance can be problematic"

Well, they've solved the allocation part... but that's kind of not really useful without a concurrent+parallel collector.

And Squeak just went to a massive effort to get everything MIT licensed. Wonder if they will incorporate this given its license.

addendum: My mistake, there appear to be some Pharo oriented files in the release already. Happy days.

Let me be the first to say it: Wow!

That was my reaction as well. I've been learning Smalltalk recently, and this seems like a really great addition to Squeak!

I learned a bit of Smalltalk in the mid-80's. It took me eight years to decide to learn C++ after that. It never felt very object oriented (or oriented).

Today, the RoarVM supports the parallel execution of Smalltalk programs on x86 compatible multicore systems and Tilera TILE64-based manycore systems. It is tested with standard Squeak 4.1 closure-enabled images, and with a stripped down version of a MVC-based Squeak 3.9 image.

Combined with the recent efforts to tidy up and pare down Squeak's image, this is really exciting!

Now I really want a Tilera workstation.

Anyone knows if there are Tilera powered systems in production environment? (And what can you run on them?)

I happen to have a manycore fetiche :D

Edit: http://www.tilera.com/solutions/cloud_computing

I want one of these. Anyone has one of these SQ2?

It will be interesting to see if Pharo merges in this work, but.. regardless, this is an excellent development and I look forwards to playing with it.

i hope for a ruby front-end will emerge for it.

like what happened with maglev: https://secure.wikimedia.org/wikipedia/en/wiki/MagLev_%28Rub...

a language cannot have enough active implementations. yet only needs one spec.

Reply on news.ycombinator.com