dcuccia

Go 'Eaters! Zot zot!  

Thanks Ron! I was particularly impressed with all the automatic logging at the end of Exercise 1 - neat.

bump - any thoughts?

Very exciting, guys. Thanks for your increasing commitment to the academic and open source communities. Hopefully, we'll see an MS-PL license for this work someday, too.

 

I do have (an admittedly biased) comment on the recent C9 "many-core" discussions. I really appreciate your encouraging discussion on many-core, Charles, in this interview and others. There seems, however, to be a supposition in the interviews that "one day in the future" we'll have many-core at our disposal. We obviously already have extremely powerful many-core processors in our everyday gaming rigs - NVIDIA and AMD have revolutionized a large set of scientific computing problems, and a cross-platform stream computing language has been ratified and adopted my many of the major players. Stream computing seems to be almost intentionally absent from the discussion. It's especially poignant to me in these two videos when discussing Dryad's capabilities for re-structuring expression tree nodes or pipelining computations based on the skew of the data - couldn't these strategies also help when targeting architectures such as the GPU (or Cell...I guess)?

 

I know you've covered the Microsoft Accelerator project in the distant past, and I know Dryad was built to target systems without shared memory, and I know parts of PFX are built specifically with CLR architectures in mind, and I knowcurrent GPU architectures have PCI-Express bandwidth limitations...but Map-Reduce has been shown to map to the GPU quite readily for some applications, Brahma and C$ languages have shown you can efficiently map high-level lamdas in managed code to GPU code, and DX Compute (though unmanaged) is here..

 

So, in the end, I guess my question is: Can we have a C9 discussion about how I might soon write something as "simple" as:

 

myVeryLargeArray.AsParallel().OnGPU().Select( x => a*x +b); ?

 

Thanks for listening - keep up the great work!!

Is www.microsoftpdc.com down?

Edit - nm, back up now.

[6:17]: "Mallocate"...I like it.

FOR PART 2:

1) Grid computing has the same challenges as multi-core computing, adding in extra overhead and remoting/serialization requirements. Windows HPC Server 2008 will provide a job manager that can distribute jobs over a network with WCF. Digipede Networks offers a similar solution. What is the feasibility of doing the same thing with the TPL and what potential hurdles must be overcome?

2) What about TPL + Silverlight 2.0?

3) What about Volta + Silverlight + TPL for #1?

I'm not much of a conspiracy theorist but did anyone else notice the edit at 26:37, between "Silverlight, and ..." "some other things"?

And, what MS?

Links broken??

It's funny, National Instruments' LabVIEW ($$$) has used a very similar graphical, multi-threaded framework for years. I actually moved away from LabVIEW to C# so that I could have a "real" runtime debugger, and managed code, robust interop, and all the rest. I love that MSRS will allow me to have the best of both worlds, and for free! Forget robots...I'm taking some data with this bad boy.