Ct - Automatic Multithreaded Code!?

[Nid Hogge]

Here's something that sounds absolutely fabulous, and holds alot of promise and excitement.. Worth the read!

"The language is an extension of C/C++, which are very familiar to software developers getting applications ready to run on Intel or AMD platforms. Where the languages require developers to manually partition code to run on specific cores, Ct does it automatically. "With Ct, it's almost like you're writing to a single-core machine," Mohan Rajagopalan, a senior researcher for Intel, said. "You leave it to the compiler and runtime to parallelize."

"The Ct compiler developed by Intel chops up the code to run on separate cores based on the type of data and the operation being performed on the data, Rajagopalan said. Intel also has developed the runtime and an API for the compiler.

Most C/C++ programmers should be able to pick up Ct quickly, since less than 5% of the language will be new, according to Rajagopalan. In addition, programs compiled in Ct can scale to as many cores as are available. "Once the code is compiled, the runtime figures out the platform and adjusts accordingly," Rajagopalan said."

See following links for more info:

Intel Develops Programming Language For Multi-Core Computers
http://www.informationweek.com/news/sof ... =208403616

Intel Ct leaps ahead in multi-threaded development
http://blogs.zdnet.com/Ou/?p=554

We can transform single thread to multithread: Intel
http://apcmag.com/we_can_transform_sing ... _intel.htm

Official Intel Page
http://techresearch.intel.com/articles/ ... e/1514.htm

Could it be it? Did Intel really find the Holy Grail of modern programming & computing?

This is surely an interesting development.. Especially as we approach the multicore era, And while still the majority of programs can't utilitize more than 2 cores or don't support multithreading at all.. Should ease up many things I hope.

If it really works, It could save Chess programmers alof of time dealing with all the SMP stuff and scaling issues.

Sounds too good to be true though. I'll remain skeptic for now.. What do you think?

[Dann Corbit]

Here's something that sounds absolutely fabulous, and holds alot of promise and excitement.. Worth the read!

"The language is an extension of C/C++, which are very familiar to software developers getting applications ready to run on Intel or AMD platforms. Where the languages require developers to manually partition code to run on specific cores, Ct does it automatically. "With Ct, it's almost like you're writing to a single-core machine," Mohan Rajagopalan, a senior researcher for Intel, said. "You leave it to the compiler and runtime to parallelize."

"The Ct compiler developed by Intel chops up the code to run on separate cores based on the type of data and the operation being performed on the data, Rajagopalan said. Intel also has developed the runtime and an API for the compiler.

Most C/C++ programmers should be able to pick up Ct quickly, since less than 5% of the language will be new, according to Rajagopalan. In addition, programs compiled in Ct can scale to as many cores as are available. "Once the code is compiled, the runtime figures out the platform and adjusts accordingly," Rajagopalan said."

See following links for more info:

Intel Develops Programming Language For Multi-Core Computers
http://www.informationweek.com/news/sof ... =208403616

Intel Ct leaps ahead in multi-threaded development
http://blogs.zdnet.com/Ou/?p=554

We can transform single thread to multithread: Intel
http://apcmag.com/we_can_transform_sing ... _intel.htm

Official Intel Page
http://techresearch.intel.com/articles/ ... e/1514.htm

Could it be it? Did Intel really find the Holy Grail of modern programming & computing?

This is surely an interesting development.. Especially as we approach the multicore era, And while still the majority of programs can't utilitize more than 2 cores or don't support multithreading at all.. Should ease up many things I hope.

If it really works, It could save Chess programmers alof of time dealing with all the SMP stuff and scaling issues.

Sounds too good to be true though. I'll remain skeptic for now.. What do you think?

Software transactions are a good idea but not magic.
See:
http://groups.google.com/group/comp.lan ... ce883db9ba

[bob]

this stuff has been around for 20+ years. Kennedy and Kuch, two name only two, worked on automatic parallelism for a couple of decades. But it only works for specific types of algorithms that depend on loops, not recursion. Hopeless for chess...

[Zach Wegner]

I'm not too impressed. All of the problems of parallelizing chess are not solved by this. Alpha-beta is a tricky algorithm to parallelize, and this language, at first glance, doesn't seem to offer any way to simplify it. It can maybe simplify areas that can be parallelized with a simple map-reduce, but chess programming is nowhere near that simple. The best scaling programs are going to be explicitly parallel.

Additionally, there is probably a big slowdown just from using this language. There is a lot of implementation details of parallelism that are handled at runtime, if I understand right.

[rjgibert]

Perhaps more interesting for parallel chess programming is the language NESL, which doesn't try to take the job of parallelizing code out of the programmers hands, but instead tries to make the task by the programmer more painless.

See http://www.cs.cmu.edu/~scandal/nesl.html

[Dann Corbit]

Perhaps more interesting for parallel chess programming is the language NESL, which doesn't try to take the job of parallelizing code out of the programmers hands, but instead tries to make the task by the programmer more painless.

See http://www.cs.cmu.edu/~scandal/nesl.html

CilkChess is written in Cilk, and actually participated in some big chess contests.
http://supertech.csail.mit.edu/cilk/

[Nid Hogge]

Well good thing I was skeptic than. ty for your comments.

I still hopes it could help simplfy writing parallel code writing. Maybe for uses other than chess, where software still lags behind and hadn't caught up up with newest hardware. Chess is actually one area which is quite developed in terms of multi processor support, so perhaps it can assist different territories as well.