Carmine's Kernel

Developing Manycore Applications with Concurrency Auto-Tuners

Continuing on with our coverage of ICSE 2011, meet Dr. Victor Pankratius. Dr. Pankratius heads the Multicore Software Engineering investigator group at the Karlsruhe Institute of Technology, Germany. He also serves as the elected chairman of the Software Engineering for parallel Systems (SEPARS) international working group. Dr. Pankratius' current research concentrates on how to make parallel programming easier. His work on multicore software engineering covers a range of research topics including empirical studies, auto-tuning, language design, and debugging.

In this video, Wolfram Schulte joins Victor to discuss the challenges of making concurrency easier for developers. One of the really interesting approaches that Victor and team are investigating is concurrency auto-tuning, and the example discussed here involve adding OS kernel-level support for auto-tuning user mode applications for manycore processor architectures. This is very fascinating research with great potential. Concurrency auto-tuner in an OS kernel? Concurrency-enlightened operating systems? Why not? Always great to meet young innovators with no fear of failure. I hope to see this type of thing materialize. Very interesting research and real world problem. Go Victor. Go!

Thanks to Wolfram and Victor for another great conversation. Tune in.

Back on Earth: Going Deep on Channel 9

It's been a while since my last post, but in my opinion that's fine because I haven't had much to say. That said, I still don't that have too much to say, but I am going to say it anyway.

This year, I will be doing many more interviews for Channel 9 (and a lot less coding, which is good, but also a little strange, but I need to grow...). Going Deep will continue strong this year and I will be focusing primarily on Vista (but not exclusively by any means).

Yesterday, we sat down and talked with 4 Windows kernel architects about the state of Windows architecture today, what the Windows Architecture team is (and why), the challenges of functional dependencies and the overall impact they have on shipping frequency, the future of NT, Security and other fun OS topics. Windows is a wonderfully complex OS that is on track to get better and better, from a core architecture standpoint, with each iteration. The obvious benefit of a cohesive architecture with well-understood functional dependency boundaries, smart thread management (to name a few) is that Windows users will have a more reliable, stable, performant and safe operating system. Great discussion. You'll see it on Channel 9 in the near future as part of the Going Deep series.

I also recently interviewed the Singularity research team (again) and had a great (and deep) discussion with some super smart folks in MSR (Microsoft Research). The interviewees were The Usual Suspects, Galen Hall and Jim Larus, plus two new faces, Manuel Fahndrich and David Tarditi. They are doing some really great work over in MSRLand and it was fun to talk to four of the scientists working on the Singularity project. Singularity, it seems, is evolving nicely... Also, in this episode, you will learn how Singularity is designed and why design decisions were made (like whey they decided to stick with Threads when, theoretically, they didn't need to). I really enjoy talking to people who are orders of magnitude more educated than myself in the computer science domain, not mention much smarter! Wow. That was a great interview. Stay tuned to Channel 9...

There's plenty more upcoming from C9's Going Deep so please stay tuned. Also, I would love to hear from you about what technologies you'd like to learn more about, from a programmatic and architectural point of view. Also, let me know what types of content you'd like to see on Channel 9. How about having a dev talk about a chunk of code she's writing, explaining design decisions, gotchas, tricks, etc? Hmm. Wonder what our legal department would say about that one.

I plan on blogging more this year. But, I won't blog just to blog. Read my first blog post to see why.

(Hey, TypePad team, what's up with the formatting on my archived posts?)

Cheers,

C

Levels of Organization in General Intelligence

In order for Artificial Intelligence (AI) to be realized, a comprehensive understanding of general intelligence (human intelligence) must be achieved, but not necessarily replicated. What is intelligence, anyway? Eliezer S. Yudkowsky states: In humans, intelligence is a brain with a hundred billion neurons and a hundred trillion synapses; a brain in which the cerebral cortex alone is organized into 52 cytoarchitecturally distinct areas per hemisphere.  Intelligence is not the complex expression of a simple principle; intelligence is the complex expression of a complex set of principles.  Intelligence is a supersystem composed of many mutually interdependent subsystems - subsystems specialized not only for particular environmental skills but for particular internal functions.

He postulates that there are no simple set of rules we can use to define the complex process of thinking abstractly and further argues that unlike physics, the field of AI cannot succeed by condensing complexity into relatively simple expressions.

Read Yudkowsky's Levels of Organization in General Intelligence to learn more about the application of general intelligence in AI. Overall, it's an interesting read.

Thoughts?

Fractal branching ultra-dexterous robots

If you're into AI and robots, then you no doubt know who Hans Moravec is. He pretty much has set out to make better-than-human machines. I'm not really a robot guy, so I honestly didn't know too much about Hans and his work. Thanks to Jim Gray for tuning me into this. Really brilliant stuff. ( Maybe Hans will help me design TOS: Thinking Operating System :) )

Here's a quote from one of his papers on bush robots that nicely generalizes the notion of fractal hands:

"Once upon a time, the most complex animal was a worm. The stick-like shape was poorly adapted for manipulation and even locomotion. Then these stick-like animals grew smaller sticks, called legs, and locomotion was much improved, although they were still poor at manipulating. Then the smaller sticks grew yet smaller sticks, and hands, with manipulating fingers were invented and precise manipulation of the environment became possible.

Generalize the concept. Visualize a robot that looks like a tree, with a big stem, repeatedly branching into thinner, shorter and more numerous twigs, finally ending up in vast numbers of microscopic cilia. Each intermediate branch would have several degrees of freedom of sensed and controlled motion. Though each branch would be a rigid mechanical object, the overall structure would have an organic flexibility because of the huge numbers of degrees of freedom. At the outer extremes, the machine would have an enormous number of individually positionable and naturally swift manipulators, coordinated for simultaneous execution of otherwise unimaginable tasks by signals and power from the central regions."

Yes, I see.

Check out more of these papers on Bush robots. Enjoy.