Tuesday Mar 11, 2008

How to get the most out of your CPU with Sun Studio 12

There used to be a time when hardware (especially CPUs) had a much longer release cycle compared to software. Not anymore. The silicon industry has joined the "frenetic squirrel" squad and cranks up updates of architectures at a pretty alarming pace. Of course, for C, C++ and Fortran developers it means only one thing -- your compilers have to stay ahead of the game, otherwise you might as well program that shiny new CPU in an assembly language.

The good news for Sun Studio 12 users is that the only thing that separates you from the kick ass performance on the latest contraptions from AMD and Intel is this set of patches. So, if you are planning to deploy on Intel's Woodcrest and Clovertown or AMD's Barcelona chips, just download the patches, install them and don't forget to tell the compiler what you want:
  1. if your development and deployment boxes are the same:-xtarget=native
  2. same as above, but forcing 64bit code generation:-xtarget=native64
  3. if your target is Intel's Woodcrest and Clovertown in 32bit mode:-xtarget=woodcrest -m32
  4. if your target is Intel's Woodcrest and Clovertown in 64bit mode:-xtarget=woodcrest -m64
  5. if your target is AMD's Barcelona in 32bit mode:-xtarget=barcelona -m32
  6. if your target is AMD's Barcelona in 64bit mode:-xtarget=barcelona -m64

Wednesday Feb 13, 2008

Sun Studio for Linux on the eve of Oscars

It is no secret (in fact, at this point I probably bored everyone to death with this) that Sun Studio for Linux was a very personal project for me. For a movie freak, this was the closest experience to subverting a major studio into making an independent movie. Of course, seeing Sun Studio nominated by Linux Journal for a "favorite Linux software development tool" seems like the closest I can ever get to being part of Oscars. Unlike Oscars, though, there's no Academy Members who vote, it is you. And since today is the last day for casting your vote my pledge is very simple -- if you like Sun Studio on Linux, don't be shy to scroll to the question #22, if you don't like Sun Studio on Linux don't be shy to shoot me an email and above all if you haven't even tried Sun Studio on Linux now would be the best time.

Monday Jul 23, 2007

Sun Studio and SXDE Live Chat

I would like to invite all of the developers out there to join a bunch of my colleagues and myself in the Live Chat event that we're putting together to discuss anything that is related to the software development on Solaris. We would be answering your questions on Solaris Express Developer Edition, Sun Studio, HPC and talking about all things related to the development process starting from how come we still have /usr/sfw/bin all the way to whether or not iPod support is part of the proper developer's desktop. The event will run from 24th of July all the way to 27th of July every day from 9am-11am PST. All you have to do is just click on the Live Chat link and you should be good to go. The whole thing looks strikingly similar to a "public" IM and is quite simple to use.

P.S. Come on! Bring your embarrassing questions with you, 'cause you know what -- quoting snippets of this upcoming live chat might just be the best way to convince PHBs around here. After all you are the customers and you know what that means.

Wednesday Jun 20, 2007

Git: yet another SCM system or a revolutionary model of software development?

I have recently come across the video recording of Linus Torvalds himself arguing a case for Git as the ultimate source code management system. The presentation is done in the usual trademark Linus style (which I happen to like immensely since it seems to be the only way to wake up the audience sufficiently enough to be engaged in a conversation) but the issues he tackles are very poignant and go far beyond the merits of a particular SCM system into a realm of what is the best tool for automating highly decentralized peer-reviewed software development model:
  1. Distributed vs. centralized development.
    Bottom line: centralized SCMs run against the proverbial Bazaar.
  2. Using networks of trust as the key means of conquering complexity of the modern software projects and dealing with compartmentalization of key developers/experts.
    Bottom line: SCM should reflect how humans are wired.
  3. How to keep the pace of development activity at the highest possible level (by making it cheap and easy to experiment and not care about breaking other stuff) yet allow the easiest transition path for the changes that seem to be beneficial for the project.
    Bottom line: branch early, branch often.
  4. Developers vs. gate keepers and why the problems they face are fundamentally different.
    Bottom line: even the best branching is useless without merging
I find his arguments about why CVS is the most braindead SCM ever invented and why Subversion simply denies itself the right to exist by proclaiming that they are the CVS done right quite convincing. Especially so when they are coming from a guy who has a project of ~22000 files to maintain and does about 25 merges per day. His main pain points with CVS/Subversion hit very close to home:
  1. There is no data coherency model to write home about. Which might not be a problem for most projects but certainly is for something as security sensitive as an OS kernel. Basically with CVS (and I think even Subversion) the only way you know that your data got corrupted is when its too late.
  2. Branching is waaay too painful because of things like global namespace for branches (read: constant collisions and things like test_12345 branches), all branches being pushed down every developer's throat, etc.
  3. Subversion patches certain things as far as branches go, but fails miserably as far as merging is concerned.
  4. The tools you use are supposed to make you more productive. Period.
Now, I must admit that as much as he was successful at positioning Git to look as the best tool for the job I am still not convinced on two accounts:
  1. Does git really offer a nice way of structuring complex projects like KDE and such? His suggestion of Super projects (workspaced with pointers to individual Git repositories) might be an interesting one, but it certainly requires some practice and experience to be evaluated properly. [2008 Update: It seems that this is no longer an issue. Git got the infrastructure for supporting Submodules and from what it documented on their TWiki looks like they got it right]
  2. Is the approach Git takes with making it easier to work with projects as a whole at an expense of treating projects as collection of files the right one? See the problem is -- I'm much more of a lone developer than a gatekeeper. So Git might be optimizing for the role I rarely find myself in (and Linus finds all the time)
But regardless of these concerns I highly recommend you watch the presentation yourself -- it is well worth it. Just keep in mind one thing: just before the BitKeeper (something Linus seems to have very fond memories of) Lary used to develop this little project called TeamWare here at Sun. What was TeamWare? Well it was "a distributed source code revision control system... which BitKeeper seems to share a number of design concepts with".

Friday Jun 15, 2007

A sandbox for playing with the coolest CPU around.

For quite sometime now I've been giving these presentations on how the layman's understanding of the Moore's Law is dead and how the world is embracing the multicore (CMT) model. The usual question I get whenever I convince people that this is, in fact, the case seems to be: "But how can I jumpstart my foray into this magical world of parallel computing?" My usual answer to that has been: get your hands on the only 8 core CPU available today in the market -- UltraSPARC T1 by applying for the Try'n'Buy program.

Well, now I've got an even better answer for all of the OpenSource developers out there: you can ssh into one of these things: ssh.sundemo.ru!

That's right. Apply for the account on the Sun Studio Demo Server and hack away!

P.S. Even thought the form is in Russian, here's what the individual fields mean:
  1. First name
  2. Last name
  3. E-Mail
  4. Address
  5. Phone
The last one is a pretty generic EULA check box.

Sunday Jun 10, 2007

Roboinvasion @LinuxTAG 2007

Ever since Sun Studio compilers and tools started to support Linux we've been present at various Linux shows and LinuxTAG in Germany has always ranked pretty high on the list. This year, however, we had a very special reason to make it to LinuxTAG: it just so happened that the release date for Sun Studio 12 (our first release ever to officially support Linux at the same level Solaris has always been supported) coincided perfectly with the LinuxTAG 2007 in Berlin. Add to it the fact that Solaris Express Developer Edition 6/07 was supposed to have its release around the same timeframe and it doesn't take a genius to figure out that we had to do something extraordinary for all those hardworking C, C++ and Fortran developers out there. The real developers. We had a couple of ideas like having a contest similar to the Google's Summer of Code or just bringing a couple of kegs of beer on site but they all either were too heavyweight to pull off in just a couple of weeks or downright illegal. Finally we've asked ourselves a very basic question: since we all are, in fact developers, what would we, personally, get excited about? I hate to say it but it turned out that the BSD community had beaten us to a punch with the idea that was unanimously deemed to be the most exciting one, so we had to opt out for programmable robots. After all, it is much easier to control a robot, you know.

And that's how controlling a LEGO Mindstorm NXT robot using Solaris Express Developer Edition and Sun Studio has become THE whole point of the contest:

Here's a funny bit of trivia...[Read More]

Tuesday Jun 05, 2007

What do Sun Studio 12, Project D-Light and SXDE have in common.

Today feels good. In fact today feels exactly like the day after the finals when I can finally stop running like a chicken with its head cut off and focus on import stuff (like writing this very blog entry for example ;-)). Because you see, today is the day when two of the projects I consider to be my personal ones have reached important milestones: Sun Studio 12 is now officially released as a full fledged fully supported suite of compilers and tools for Solaris and Linux. And at the same time Project D-Light has made its first public appearance as part of the Sun Studio Express - June 2007 Build. As I said, these two are very personal (which is not to say that I was the only one who worked on them) yet very different projects. The first one (Sun Studio 12 compilers on Linux) started out as a true skunk works and by the time we got any public exposure we actually had a pretty decent implementation already so the only logical question to ask was: when should we ship it? Project D-Light on the other hand got exposed right after the idea for it ripped through the management chain Alien-style at the time where we didn't really have much functionality to write home about. So what do these two projects and Solaris Express Developer Edition (which is supposed to have its next release available any minute now) have in common? The answer is quite simple -- you. Yes -- YOU, our users, early adopters and loyal customers. Without your support nothing of what makes me proud today would have been possible. Not Sun Studio 12, nor Project D-Light. I wish to thank you and once again ask for the only thing that matters: your feedback. Sun Studio 12 and compilers for Linux ended up a success in large because you clearly showed your interest at our Sun Studio for Linux Forum. I truly hope that Project D-Light has a chance of following that road to success. The Sun Studio Tools Forum awaits!

P.S. As for Solaris Express Developer Edition I keep entertaining the idea of hijacking it and turning into my third personal project. Although it feels like I might be biting off way more than I can chew ;-)

Sunday Apr 29, 2007

DTrace probes -- assert() on steroids?

Quite recently I've had a conversation with a good friend of mine who shares my interest in DTrace and compilers and it seems that we've come up with a pretty neat use for DTrace probes. The idea is to start using DTrace probes for assert(3). You know how, during the development cycle the assertions are kept in the code and when the final build is supposed to happen most of them get deleted by turning off a magic define? Well, with DTrace probes used instead of assert() the beauty is -- they are all turned off by default (no funky defines required) but the dynamic nature of DTrace lets you turn any subset of them on when and only when you really need them. Like at a customer site (with the production code running) to diagnose a particular problem. Neat, isn't it?

Friday Mar 30, 2007

Interested in what's up with Sun Studio?

Kuldip has the latest scoop! And for even more details make sure to login
SDN Channel

Wednesday Mar 21, 2007

A bit of C code that costs $200 per line.

If you're the type of a person who played a fair share of arcade games in the past AND at least once the code you wrote drove your coworkers to the brink of insanity you should get a kick out of the thingy a bunch of us here at Sun put together just to show how much fun Sun Studio and Solaris can be. So there's fun, fame and $5000 fortune for anybody who dares to enter

Temple of the Sun!

Monday Jan 08, 2007

The Landscape of Parallel Computing Research: A View From Berkeley

As a somebody who believe in parallelism as the only way out of our dependency on Moore's Law I was very pleased to discover this Wiki put together by Berkeley research team. Now that I have this URL I don't have to keep maintaining my collection of links on del.icio.us ;-)

Monday Oct 02, 2006

How to become a compiler wizard in 1 hour

You know how when you work in a particular field long enough you slowly become a resident guru or a go-to guy at which point you promise to yourself that one of those days you would definitely write an FAQ if only not to repeat the same thing over and over again ? You never do, of course.

Well, all of the procrastinating gurus in the compiler field can rejoice since from now on all it takes to enlighten your trustworthy apprentices is to point them at the introductory article written by Abdulaziz Ghuloum An Incremental Approach to Compiler Construction so that you can happily get back to playing Nethack.

But just in case -- keep your Muchnik handy for those who might come back ;-)

Many thanks to Horsh for sharing this URL with me.

Tuesday Sep 12, 2006

A tale of three conferences

Perhaps some of you have noticed that I haven't blogged much lately (well, I certainly hope you have -- otherwise what's the point of me blogging in the first place ;-)). The reason of course is that I've been quite busy lately going to various software conferences talking about MultiCore revolution and how Software Tools such as compilers play a key role in it. So far I've been at 3 conferences: ClueCon in Chicago, LinuxWorld in San Francisco and Sun TechDAYs in Seattle and every single one of them proved to be quite interesting and exciting in its own right. ClueCon, for example completely shattered my concerns about being a VoIP conference (and mind you, I don't really have any expertise in that domain) and in fact happened to be one the most welcoming and feel-right-at-home software conferences I've ever been to. I believe that if there's anybody to be thanked for making ClueCon the way it was it would be Tony Minnesale . The brain (and soul!) behind a FreeSWITCH Project. I can only hope that ClueCon atendees felt as good about my presentation as I felt about being there, but since at least one guy said to me: "Wow! Good thing we didn't leave -- we thought it would be just another Java zealot from Sun talking about irrelevant stuff but you actually gave us a bunch of good ideas on how to make our telco apps faster on AMD64 and Niagara boxes. That's useful" I think it went ok.

LinuxWorld, of course, was bigger louder and much more commercial. Comparing it to ClueCon is sort of like comapring a nice boutique to Macy's ;-) However, the fact that I met Ron Peacetree and spent a day with Josh Abraham and the "maddog" definitely made it a very memorable experience. My only concern is that I somehow misplaced a business card that a very nice guy from O'Reilly booth gave to me and I really want to thank him for his gift and follow up on his proposal. So if anyone who reads this happens to know who was staffing the O'Reilly booth there -- please drop me a note.

Now Seattle was all about sunny and warm weather and an opportunity to talk directly to our customers -- both things which I personally appreciate very much.

As for the presentation I've made -- since several folks have expressed their interest in getting a .PDF copy of it -- here it is: Unleashing the Power of Chip multi-threading with software tools. Enjoy!

Monday Aug 14, 2006

Linux Technology preview is dead! Long live the Sun Studio Express!

Today is the first day of the LinuxWorld in San Francisco (and actually I'm writing this waiting for a speaker to arrive and give us a lecture on what's new in the field of buffer overflow exploits and related fields of fine hacking ;-)) which is quite exciting especially if you take into account that it'll be my first LinuxWorld where I participate as a speaker , giving a presentation on 'Unleashing the Power of Chip multi-threading with software tools'. On top of which our our team has just released Sun Studio Express 2, which signifies the end of the Linux Technology Preview phase and makes Linux just another platform Sun Studio supports. We should all be grateful to the Sun Studio Linux Technology Preview Program since it actually made it possible to break the physiological ice and create the real deal -- Sun Studio Express. Which is basically our way of making the development as transparent to you as possible. It is also your opportunity to tell us what's wrong with our software before its too late for us to fix.

Anyway -- it is a great time to subscribe to Sun Studio Express, start using it on regular occasion and give us your feedback.

As I said -- this particular one is totally focused on Linux and comes with the boatload of features and bug fixes so that even the pickiest ones shouldn't be disappointed ;-) And we've got Solaris developers covered as well! Its all there. Sun Studio Express gives you exact same bits we use internally. For those of you in close proximity to San Francisco it is also a great time to drop by our booth in Moscone Center (we're actually located in the AMD's pavilion -- so don't be fooled) talk to us and for the really curious ones may be even attend my presentation on Wed 08/16/2006, 10:15 AM - 11:15 AM. Hope to see you there!

Saturday Jul 29, 2006

SDN Channel on Linux: deleted unrated scenes

As you might have noticed in my previous blog entry I really wanted to get your opinion on my interview. But aside from the usual curiosity I had a practical reason to beg for it to be reflected in the comments section of my blog. Because you see, the other part I had to do for the Linux episode of the SDN Channel resulted in the most bizarre and troubling interaction I have ever had while working for Sun.

And that other part I am talking about is making of flash demos.

As a producer of these demos I really wanted them to do one thing -- transport every viewer into the office(s) of the guys who actually wrote the software in question (one way or the other). I wanted them to be as geeky and as similar to the usual hallway conversations we all have as possible. One thing I didn't want them to be was flashy and marketing looking. I'm a developer myself and I consider it to be a bait'n'switch when somebody is trying to feed me a marketing presentation calling it a developer demo.

Two demos were produced under these assumptions and they were shown to a bunch of colleagues of ours. Both native and non-native English speakers. The response we got was almost unanimous -- audio might use some fixin' but otherwise they are nice. The coast was clear and the demos were on their way to http://developers.sun.com when all of a sudden we had a really strong pushback from a new person responsible for ok'ing everything that goes on our site. What that person demanded was a total removal of \*both\* demos (which wasn't half as bad as the actual reason given to us). Now, don't get me wrong here I would be the first one to agree that the technical merits of these demos are probably not quite at the level where I would feel comfortable submitting them for an Oscan nomination. That I was prepared to hear and discuss. But I was completely unprepared for an actual reason that was given to me: I was told that the only reason they wouldn't be published is because having demos narrated by non-native English speakers or even by somebody who doesn't happen to be a professional voice talent is below the quality bar that is acceptable for a successful image of Sun microsystems.

Just like that

I tried to argue my case. First by trying to convince my opponent that it is always better to hear an unscripted developer chatting about things which are near and dear to him than to have a voice talent doing a lip syncing job. I even referred to the governor of the state of California elected to his position regardless of the thickest middle-European accent since Kissinger. All to no avail. I think somewhere along the way it actually got personal (after all when it comes to accents I'm as touchy as the next [Slavic] guy)

At that point I stopped arguing and figured that as long as one of the demos did survive and as long as Jonathan has a soft spot in his heart for http://blogs.sun.com I can safely use this place instead of the official one (http://developers.sun.com) to talk to the sort of guys and gals I want to talk to -- you.

And so without further ado, here's a portion of the SDN portal which you were not supposed to see. And as with my previous plea I hope that you would find it possible to spend a couple of minutes adding a comment or two on what you saw and heard. I can not stress enough how important it is for me to find out whether all of the allegations made against this demo (and even the first one!) are true or false. And who's the better force to set this record straight than you -- viewers and listeners we created this demo for. Just keep in mind one thing -- I don't want to assert that quality of the audio here is good enough for a broadcast all I'm trying to say is that it is comprehensible. Should we do better next time? Absolutely! Was it necessary to remove this demo ? NO!

Sun Studio Next Generation IDE Tour
Watch the Sun Studio Next Generation IDE Tour
See what kind of the IDE experience you should be preparing yourself for once we made it available for early adopters some time later this year. This demo was created by (in alphabetical order): Andrew Krasny, Maria Tishkova, Anton Vysotsky and Anton Youdkevitch. Editing and postproduction -- Roman Shaposhnik. The team would like to thank Anya Barsky for feedback and support and Jonathan Schwartz for being a Patron Saint of http://blogs.sun.com.

» Take the IDE tour (flash)

P.S. And especially if you happen to be a non-native English speaker please let me know whether you had major difficulties understanding the demo. One of the allegations against it was exactly that it would be practically impossible for non-native speakers to understand it. However, as a non-native speaker myself I find it particularly hard to believe.



Top Tags
« February 2017