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.

Deep Vista Audio Stack and API

Seasoned Niner (Channel 9 member), Larry Osterman, an SDE and 20 year Microsoft veteran, and Elliot H Omiya, a Software Architect and audio guru, dig into the innerworkings of Vista's updated Audio Stack and new user mode API in the latest installment of the Going Deep series on Channel 9. Much of the guts of Windows audio have been moved up into the land of the user and this has consequences for both Windows audio developers at the API level and for Windows at the general programmability, reliability and stability levels.

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

Singularity on Channel 9

Check out the interview I did on Channel 9 with two of the MS researchers working on a prototype operating system, Singularity, written primarily in safe C#. Of course, Singularity's HAL is written in unmanaged code, but most everything else is managed. Very cool indeed.

I look forward to seeing where this project leads in terms of the evolution of consumer operating systems. As was addressed in the interview, research operating systems like Singularity will enable wide-open experimentation and implementation of crazy ideas like a threadless OS, an intelligent OS (cybernetic, homeostatic system with the ability to learn and adapt), you name it.

I look forward to the day when my OS will not only be secure, reliable and performant but also adaptable, preemptive, modular and capable of thinking. Of course, the later requirement is probably 50 to 75 years out, but the notion of a Windows NT where T stands for Thinking is quite compelling!

Enjoy the interview. Please share your thoughts either on Channel 9 or here.

What will the 64-bit world look like?

With the advent of 64 bit processors and the operating systems that virtualize them, we are going to see some amazing new technologies emerge in the next 5-10 years. I had the pleasure of interviewing some folks involved in the 64 bit revolution today and really left the interview feeling excited about the future of applications and what we will be able to do on our personal computers thanks to powerful 64 bit processors coupled with cheap memory. You'll be able to see the interview on Channel 9 in a few weeks.

With 64 bit computing, there is no longer the 32-bit memory barrier that hampers things like fast analysis of very large datasets (because on 32-bit systems you can't hold them in-memory without causing significant perf problems or a dreaded out of memory exception...). Also, there is increased performance of the system generally due to the greater working address space. When apps start being written in 64-bit mode from scratch (that is, not ported from 32-bit), the possibilities of what we will see are truly compelling, like new types of games and powerful video/audio processing applications for your 64-bit laptop.

What new applications might we expect as a result of 64-bit computing?

What happened to OS research?

Think about this for a moment: The operating systems we all use on a daily basis, regardless of the particular flavor we prefer, are based on the same exact core principles and ideas that have been around for over twenty years.

Of course, we do see many high level innovations in consumer operating systems like Windows, OSX and Linux. New technologies like the CLR (it's what enables everything .NET), for example, represent significant innovation in the areas of virtual machines and software development in general. Systems are becoming more stable and reliable due to innovations at the kernel level. Computers themsleves, however, are not becoming smarter. They continue to be input-driven machines. There is little innovation happening at the OS level from a cybernetic point of view (adaptive regulatory feedback). It's as if operating system research, for all intents and purposes, has died.

I am not criticizing what we have today. My PC, for example, hasn't crashed in ages and does pretty much what I expect it to do. I am primarily concerned with the problem of what we will have tomorrow. How homeostatic and adaptive will operating systems of the future be if we, as a community, invest so little in the science of operating systems? Without significant basic research, what will drive OS evolution?

Ironically, the application of machine learning research to consumer operating systems is almost exclusively targeted at solving problems like automatic email prioritization, automated content management, automated software analysis, and other really nifty user features, which is great, but what about a machine capable of learning about itself and using this knowledge to better deal with perturbations and to further develop the ability to anticipate potential problems?

Addendum:

I am well aware that Intelligent machines have already beaten the world's best chess players, discovered abstruse mathematical theorem proofs, etc. Artificial Intelligence is primarily applied to robots not personal computers. I'm talking about the evolution of operating systems that run our PCs, not the software that drives the complex machinery of intelligent robots capable of thinking and reasoning. While the latter technology is on pace to usher in truly smart machines within the next 40 to 50 years, the former is frozen in a state of scientific stasis.