Nice article ...
http://existentialtype.wordpress.com/20 ... ncurrency/
parallelism is not concurrency
-
- XCore Addict
- Posts: 228
- Joined: Thu Jan 27, 2011 3:30 pm
- Location: Argenteuil - France
-
- XCore Addict
- Posts: 216
- Joined: Wed Feb 10, 2010 10:26 am
I think there is a lot to learn from this!
I realy like the abstraction of dependent calculations. It sounds quite familar with that was XC does with it's channels - in a much simpler way.
Would be very interesting to implement a automated scheduler based on this. Unfortunately got currently not enough time at hand.
I think such an implementation would be one (very) specific solution for the channel protocol & parallelism problem of XC. Evn though it would not make everything possible that is possible with XC it would make a lot of things easier:
- There are calculations which need data and produce data
- A schedule tracks which calcuation has enough data to to start and triggers it.
- The 'data' would be an abstraction for a communication protocol over channels.
In your calculation you do not have to deal with channels & protocols. Sounds like an easy abstraction to me. But I do not want to be the person implementing the abstraction. A horrible amount of work ;)
I realy like the abstraction of dependent calculations. It sounds quite familar with that was XC does with it's channels - in a much simpler way.
Would be very interesting to implement a automated scheduler based on this. Unfortunately got currently not enough time at hand.
I think such an implementation would be one (very) specific solution for the channel protocol & parallelism problem of XC. Evn though it would not make everything possible that is possible with XC it would make a lot of things easier:
- There are calculations which need data and produce data
- A schedule tracks which calcuation has enough data to to start and triggers it.
- The 'data' would be an abstraction for a communication protocol over channels.
In your calculation you do not have to deal with channels & protocols. Sounds like an easy abstraction to me. But I do not want to be the person implementing the abstraction. A horrible amount of work ;)