Slashdot Mirror

samsung's website claims the 240T has a dot pitch of 0.27 mm. that's about the same pitch as a 14.1" XGA (1024x768) screen. so resolution-wise, it's not that great. compare this to sgi's 17.3" 1600sw flat panel (1600x1024 at 0.23 mm dot pitch), or the screen's on those 15" uxga a21p ibm thinkpads (1600x1200 at 0.19 mm dot pitch). for viewing text or graphics up close, smaller dot pitch is better (all else being equal:)

FSN is not fake, it actually looks just like what you saw in the movie. I think the Jurassic Park people added the sound effects, but the real FSN actually let you fly around a graphical representation of your filesystem, fly into subdirs by clicking on them, launch apps, etc.

Well, my project is using OpenGL and OpenAL as the 3D library and audio library, respectively. For a 2D library, OpenPTC is always nice, but you can also try faking some 2D under OpenGL. That is the first step on getting your software cross-platform.

Input is not as problematic as you'd think and is relitively easy to port across platforms. Especially for joystick, mouse, and keys.

Right now I'm using GLUT to handle input and windowing for the actual game executable and wxWindows for the other tools. I'm tempted to switch entirely over to wxWindows, although it doesn't have an up-to-date Mac port. The problem with GLUT is that it isn't fast and isn't powerful, but it's great for getting things up and running quickly.

But as long as you just have to rewrite the program that popps up a window and sets up the OpenGL/OpenAL contexts, it's not as big as starting in DirectX and porting to GL. Just carefully architect the basic framework and there won't be any problems.

Also note that Mozilla's C++ Portability guide may prove to be useful. The goal is to think about portability from the beginning.

dmSDK is SGI's Digital Media SDK. It was recently released for Linux. The advantage IMHO is that dmSDK is already being deployed on Irix and NT, in other words, it has been field tested and is very well documented. And because it's desktop enviroment independant (i.e. it doesn't have a G or K prefix :) its acceptance might happen more frictionless. It is not (yet?) released under GPL but under SGI's OSS license. But the most important "advantage" is has is perhaps the fact that it is included in the Khronos group's standard which makes it just as standard as OpenGL. I encourage you to go check it out...

-adnans

dmSDK is SGI's Digital Media SDK. It was recently released for Linux. The advantage IMHO is that dmSDK is already being deployed on Irix and NT, in other words, it has been field tested and is very well documented. And because it's desktop enviroment independant (i.e. it doesn't have a G or K prefix :) its acceptance might happen more frictionless. It is not (yet?) released under GPL but under SGI's OSS license. But the most important "advantage" is has is perhaps the fact that it is included in the Khronos group's standard which makes it just as standard as OpenGL. I encourage you to go check it out...

-adnans

Mandrake has allowed RieserFS installs since at least version 7.1 in the ``expert'' install mode. [My laptop is currently drake 7.2 with ``RieserFS everwhere'' except for a 20M /boot.]

SGI provides a Modified Red Hat 7.0 Installer that allows installing linux on top of their XFS filesystem. I haven't tried the XFS under Linux, but I've been impressed with XFS under IRIX. [One of the larger machines I currently admin has several terabyte-sized XFS filesystems.]

Mandrake has allowed RieserFS installs since at least version 7.1 in the ``expert'' install mode. [My laptop is currently drake 7.2 with ``RieserFS everwhere'' except for a 20M /boot.]

SGI provides a Modified Red Hat 7.0 Installer that allows installing linux on top of their XFS filesystem. I haven't tried the XFS under Linux, but I've been impressed with XFS under IRIX. [One of the larger machines I currently admin has several terabyte-sized XFS filesystems.]

Mandrake has allowed RieserFS installs since at least version 7.1 in the ``expert'' install mode. [My laptop is currently drake 7.2 with ``RieserFS everwhere'' except for a 20M /boot.]

SGI provides a Modified Red Hat 7.0 Installer that allows installing linux on top of their XFS filesystem. I haven't tried the XFS under Linux, but I've been impressed with XFS under IRIX. [One of the larger machines I currently admin has several terabyte-sized XFS filesystems.]

Add this one to the list:
A Garbage collector for C and C++

Don't bother with manual memory management unless you're writing OS kernels or device drivers. A good gc *will* do it faster than you, especially in a multithreaded program. If you *are* a memory management guru who knows how to optimize it better than any gc, you've probably written your own gc anyway and don't need this link. Otherwise, just get it, use it, and #define free(x) from now on.
--

Yeah, it's called fsn, and it lives here. Note: only works with IRIX versions 5.3 and below.

Speaking of true random number generators, check out SGI's Lavarand .

"Lavarand... harnessing the power of Lava Lite® lamps to generate truly random numbers since 1996"

fun! ;)

i'm assuming that someone would eventually insist that i backed up my statements with some sources.

the sources i can provide with 5 minutes of research are, sadly, weak, but here goes:

Brocade, a very highly respected manufacturer of FC switching products, has a discussion about this very topic here.

also, as someone else already mentioned in another post under this article, a counterpoint as researched by SGI is here.
keep in mind that this is still a research project and probably can't be considered ready for prime-time yet, but it shows tremendous promise and validates the counterpoints made almost 10 years ago now quite well.

whether you agree with these sources or not, the prevailing opinions for years have been both what brocade *AND* sgi state.
half the camp said "FC is designed for high-bandwidth streaming, ethernet is too laden with baggage", while the other half said "but if we are smart (maybe even tricky) about the way we implement a,b and c, we should be able to make it a moot point."

so, be your own judge :)

Peter

i'm assuming that someone would eventually insist that i backed up my statements with some sources.

the sources i can provide with 5 minutes of research are, sadly, weak, but here goes:

Brocade, a very highly respected manufacturer of FC switching products, has a discussion about this very topic here.

also, as someone else already mentioned in another post under this article, a counterpoint as researched by SGI is here.
keep in mind that this is still a research project and probably can't be considered ready for prime-time yet, but it shows tremendous promise and validates the counterpoints made almost 10 years ago now quite well.

whether you agree with these sources or not, the prevailing opinions for years have been both what brocade *AND* sgi state.
half the camp said "FC is designed for high-bandwidth streaming, ethernet is too laden with baggage", while the other half said "but if we are smart (maybe even tricky) about the way we implement a,b and c, we should be able to make it a moot point."

so, be your own judge :)

Peter

All the big post-production houses in the world (i.e. the companies who add FX to and edit movies) and companies like SGI and Discreet were responsible for pushing SCSI to it's limits. They couldn't make go faster so that's where they left it.

All these companies use fiber channel disk arrays now. It already is pretty mature and as stable as it needs to be. If it wasn't, they wouldn't be using it.

----------------------------

STP also works over gigabit ethernet hardware (but only at gigabit speeds). It will probably work over 10gigabit ethernet, when that is available in quantity.

Why use special disk interface hardware, if network hardware has better bandwidth, latency, and is cheaper?

There was some confusion a ways up in this thread about Netscape for SGI IRIX. Here are three useful links:

SGI's build of 4.75 (4.76 should be there soon):
http://www.sgi.com/products/evaluation/

Netscape's build of 4.76:
ftp://ftp.netscape.com/pub/communicator/english/4. 76/unix/supported/irix65/

Mozilla, etc, for SGI IRIX:
http://reality.sgi.com/rhess_engr/mozilla/irix/

There was some confusion a ways up in this thread about Netscape for SGI IRIX. Here are three useful links:

SGI's build of 4.75 (4.76 should be there soon):
http://www.sgi.com/products/evaluation/

Netscape's build of 4.76:
ftp://ftp.netscape.com/pub/communicator/english/4. 76/unix/supported/irix65/

Mozilla, etc, for SGI IRIX:
http://reality.sgi.com/rhess_engr/mozilla/irix/

Better yet, leave one machine of each model you configure running in your showroom all the time with the tool running on it.

Even if you sell machines preinstalled with Windows, let your customers know that they're getting machines that are of more robust hardware quality - if they're pass these tests thrashing the system under Linux, well you can figure it'll be pretty reliable running Word under Windows 98.

Also have a bin near the door full of "Free - take one" floppies duplicated with Memtest86 - and test the memory you install with Memtest86 before giving it to a customer (to some extent it should validate CPU's and motherboards as well).

Ask your customers to take memtest86 home and try it out on the machines they may already have. (You boot off the floppy to run the test; on Linux systems you can install it in LILO or Grub).

They may get a suprise - I have a PC-133 DIMM that failed memtest86 when the machine was brand-new from a small PC shop like yours; I doubt they tested their memory with more than the BIOS test - my BIOS test still doesn't show any errors, but memtest86 consistently shows the same error on this DIMM.

Michael D. Crawford
GoingWare Inc

If E is too sluggish, I'd suggest trying sawfish instead. I'm using that (also installed from the tardists at freeware.sgi.com) on my O2 and find it to be far better than 4DWM. But then, ultimately that's just my personal taste.

You can buy them in a package with a PCI Number 9 revolution IV and the whole kit can be found for less than $1400(US).

Almost all computer have a couple PCI ports open, and unless you're doing fast texture writes it's really not much slower than AGP.

We're really thinking about rolling these out as the standard dual-head setup at my workplace to cut down on eye strain and electricity costs.

What I'm saying is, that since you have two caches, there is a better chance for data that get into a cache to stay there. For a more detailed explanation see item 4.2.2 in SGI's multiprocessor programming manual.

Threads are paths of execution through the same process context. It's very difficult to distribute them without a single system image, which is what Irix can do. In general, if you want to distribute a task across multiple machines, each part of the task must be able to run self-contained, and be incorporated within the overall result of the computation. This would very well for, say, rendering, because you can simply give each node a frame to do, then assemble all the frames into a movie. But it would be very difficult to break a scene apart into objects, render each one of those on a different machine and then incorporate them into a new image, because the different machines wouldn't be able to compute the effects of a shadow of one object on another.

I have read a lot of material from Microsoft that is directed at Linux. Various Microsoft employees have said to the effect, Linux is not ready for the enterprise, it doesnt scale, major players don't support it, It's not really free, etc... Well, I did some research and while I see some of Microsofts points, the majority of their rhetoric is either pure FUD or libelous marketing because that's the only thing Microsoft can do now. Microsoft can't buy Linux, can't "embrace and extend", can't buy a company and put it out of business, and basically can't do anything. I will now list a series of excerpts from various articles suggesting that linux is ready for prime time. I have also put in the links if you want to read the whole article. Here are some strong backers of Linux and various contributions and/or excerpts:


IDC
has predicted that Linux will hold 38 percent of the market by 2004. Interestingly enough, Microsofts group products manager, Doug Miller, claimed that recently released numbers from IDC System Software Research show that "Linux growth in server OS share has been flat for two quarters, and Unix and Novell continue to fall." Even more interesting is that IDC manager, Al Gillen, would not confirm Miller's analysis. Wired News

IBM
Big Blue committed to spending $300 million on Linux services over the next three years. IBM has already committed to investing $1 billion in Linux over the next 12 months. President and COO, Sam Palmisano, said "IBM has made our choice....we put a significant amount of IBM's future prosperity behind Linux. We don't invest a billion dollars casually. Lou [Gerstner] and I don't write those checks without, shall I say, some engaging meetings." Big Blue also unveiled Linux-based network processor software development tools and services for ISPs and networking equipment vendors, including:

Domino Workflow on Linux -- software which enables customers to build, modify and improve business processes like employee hiring and CRM by streamlining and automating interactions

Plans to expand Linux support for Tivoli Systems management software

IBM Director for advanced systems management software available on Linux for the IBM eServer xSeries product line, including a "self healing" feature to predict server failures

Availability of the NetVista Thin Client, the N22001, running Linux

Linux-certified IntelliStation Z Pro workstations based on Intel's new 64-bit Itanium processor.

Citing such real-world Linux customers as Weather.com, Shell Oil, and National Center For Supercomputing Applications at the University of Illinois, Palmisano said people who doubt that the operating system can scale to the biggest of applications are just wrong. Weather.com, one of the Web's most popular sites, supports anywhere from 5 million to 27.5 million page views per day running Linux and can scale even higher to 40 million per day, according to the company's CTO Mark Ryan. Techweb or eltoday

Oracle
Has ported Oracle 8i already to Linux. They recently released "Oracle Internet File System" and "Oracle Parallel Server" for Linux. If this isnt a major move by a major company then I don't know what is. Databases need to scale and thus if Linux can scale then Microsoft is full of it... Read on. "Oracle Parallel Server is the most mature and trusted high-availability database technology available for the Linux platform. It provides sub-minute failover capability, allowing Linux environments to achieve significantly improved levels of application and data availability. Oracle Parallel Server allows applications running on any server in a cluster instant access to all data in a database, and will support up to a 4-node, 8-way cluster." Hello Microsoft do you see this?

"Oracle has announced all of its major Internet Platform software products on Linux, including Oracle8i(TM) Release 3, the latest version of its database; Oracle9i(TM) Application Server; and Oracle JDeveloper with Business Components for Java and Oracle Forms, two popular Oracle application development tools. In August 2000, Oracle announced an industry first with the shipment of the first enterprise-edition application server on Linux. Oracle adds to its firsts with Linux with the addition of Oracle Parallel Server and Oracle Internet File System." So much for the myth of no vendor backing. eltoday

SGI
Is looking at linux as the future. Much of SGI's work is underground and less advertised. Much of it is kernel level enhancements, such as scalability, NUMA, big memory support, etc... SGI has released several of it's graphical products for linux such as, Open Inventor, Open GL Performer, and many other high end development tools. In the filesystem arena, XFS is in stable beta and is very promising for mass storage management and reliability. Open Source at SGI

Dell
"Dell Computer and Oracle agreed Wednesday to establish a Linux center in Austin, Texas... Dell will use the facility, which is scheduled to open in the spring, to test and tune Oracle databases running on Intel-based systems running Linux. Oracle also agreed to use Dell's servers and storage products for building the Oracle 9i database on Linux, the companies said." CNet News

Not enough corporate backers? Think again. Here are some other companies who have started partnerships with linux companies, cooperated, released specs, or released products for linux: Informix
Compaq
HP
Sun
Cisco
AMD
Intel
IDG
Adaptec
O'reilly and Associates
Nokia
Tivo
NeTraverse Inc.
3dfx
Nvidia
Creative
this list goes on and on......

I believe XFS and JFS are both journaled in this way.

And based on what do you believe that? Here's an excerpt from an IBM document:

First, JFS only logs operations on meta-data, so replaying the log only restores the consistency of structural relationships and resource allocation states within the file system.

I can't find anything quite as authoritative for XFS, but there is this, from an SGI document:

The log section (or area, if it is internal to the data section) is used to store changes to filesystem metadata while the filesystem is running

If the log section contained data, one would certainly expect them to mention it. There's also this, from a copy of the Fileystems HOWTO:

File systems update their structural information (called metadata) by synchronous writes...A journaling file system uses a separate area called a log or journal. Before metadata changes are actually performed, they are logged to this separate area

Again, the reference is to metadata but not data. As a professional filesystems developer, I know that even the overhead of journaling metadata is a big problem for journaled filesystem implementors trying to provide performance competitive with non-journaled filesystems. Journaling the data as well would be absolutely horrendous, and I certainly think I would have heard if there had been some breakthrough allowing this to be done efficiently.

There are some filesystem technologies that do provide safety for data in addition to metadata. Log-structured filesystems can include this feature, though they're somewhat out of vogue right now. The only current effort I know of in this direction is the atomic update ("phase tree") methodology in Tux2.

Not just IBM, but SGI, CRAY, Compaq, VA and Penguin are all hardware vendors putting serious resources to making linux better.

Bear in mind that there have been at least four entities called "Cray". All but one would be perfectly at home building clustered micros.

1. The original Cray, pioneer of vector supercomputers. Probably the company you're thinking of.
2. The same company after it branched out into microprocessor-based high-performance computing, system integration, and consulting. This entity not only made MIPS, SPARC, and Alpha-based systems, they resold Sun and SGI workstations.
3. A loosely-defined entity never completely assimilated by SGI. The only parts of Cray SGI really wanted was Craylink (as much to keep it away from Sun as for themselves) and maybe the compiler software. The rest they more or less ran into the ground.
4. Tera Computer, which bought the Cray name from SGI, together with the Alpha-based and vector supercomputer lines. SGI had already sold the SPARC-based line to Sun and kept the MIPS-based line for itself.

__________________

Bear in mind that there have been at least four entities called "Cray". All but one would be perfectly at home building clustered micros.

1. The original Cray, pioneer of vector supercomputers. Probably the company you're thinking of.
2. The same company after it branched out into microprocessor-based high-performance computing, system integration, and consulting. This entity not only made MIPS, SPARC, and Alpha-based systems, they resold Sun and SGI workstations.
3. A loosely-defined entity never completely assimilated by SGI. The only parts of Cray SGI really wanted was Craylink (as much to keep it away from Sun as for themselves) and maybe the compiler software. The rest they more or less ran into the ground.
4. Tera Computer, which bought the Cray name from SGI, together with the Alpha-based and vector supercomputer lines. SGI had already sold the SPARC-based line to Sun and kept the MIPS-based line for itself.

__________________

True, Sun does not actively provide support to all of the other platforms -- but the people who write those platforms generally do (and who better?)... For example, the BlackDown distro of the JVM supports Debian, RedHat, Slackware, and SuSe for x86, PowerPC, Alpha, and SPARC. Besides, Sun support for that PLATFORM would only be necessary if your are writing the VM. The java binary is the same regardless of platform.

this project was called lavarand. i spoke to the guy who started this about 3 years ago, smart guy.

heres the url for more info:
http://lavarand.sgi.com

.brad

Nice to see a full open source release though rather than just an IRIX binary.

--LP

• Stick a lava lamp in front of it, and you`re sorted!

lavarand!

walking past a fishbowl at SGI and seeing a bunch of "heat lamps" [sic] was pretty trippy...

--

How clueless can this guy be ? If someone went to such great lengths to defeat the 2gb limit, then I'm pretty sure it's because it's been a problem for a while. Uncompressed video comes to mind [...]

You don't even have to reach that far. Compressed video easilly grows to larger than 2 gb for any non-trivial project. For example, I used dvgrab to capture multiple small video clips[1] from my ieee1394 sony trv-900 camcorder and media converter (sony), then edited them together into a 25 minute home video. This is all using compressed DV format, which is small enough that captures work perfectly fine in realtime to ATA33 IDE drives (unlike traditional analog captures which demand much faster drives because the quantity of data is so much greater).

25 minutes of video, even in 4:1:1 or 4:2:0 compressed DV format, is way bigger than 2 GB.

My solution was to upgrade to kernel 2.4.0 (which is easy to do with Mandrake 7.2, as long as you do not compile in devfs support) with the ieee1394 fixes. I opted to use SGI's XFS filesystem (which rocks) but to get around the 2GB limit upgrading to 2.4.0 was sufficient (ext2 and reiser both worked fine for test files of about 5.5 GB in size).

[1]This is a limiting bug in dvgrab which segfaults at around 900MB, but works fine in "looping" mode with filesizes 900MB.

-dB

SGI's Linux FailSafe is the premiere next generation High-Availability (HA) solution for business critical applications. It provides a simple way to make your applications HA without having to resort to rewriting or recompiling or the need to invest in expensive hardware solutions. Linux FailSafe provides a robust clustering environment with resilience from any single point of failure.

Linux FailSafe is architected to scale up to 16 nodes in a cluster with the cluster members sharing storage. Linux FailSafe and shared storage allows multiple servers to assume control of data in the event of a failure. At the point of failure, applications are resumed on the remaining system{s} and filesystems are automatically made available to their applications.

---

Thanks for the Link. Actually it appears that XFS fares the best of all the filesystems. I attended a Linux Road Tour from SGI about a year ago. I attended a session on XFS and I was truly blown away by it's capabilities. XFS on IRIX is truly amazing and for SGI to opening up the specs for the Linux community is truly admirable. XFS is SGI's crown jewels and they are giving it to the community. For more on XFS, read this. Alternatively, you can watch a streaming video about XFS here. XFS will be the industrial strength filesystem that will push Linux into the High Availability server arena. For desktops, ReiserFS should still be sufficient.

Thanks for the Link. Actually it appears that XFS fares the best of all the filesystems. I attended a Linux Road Tour from SGI about a year ago. I attended a session on XFS and I was truly blown away by it's capabilities. XFS on IRIX is truly amazing and for SGI to opening up the specs for the Linux community is truly admirable. XFS is SGI's crown jewels and they are giving it to the community. For more on XFS, read this. Alternatively, you can watch a streaming video about XFS here. XFS will be the industrial strength filesystem that will push Linux into the High Availability server arena. For desktops, ReiserFS should still be sufficient.

I am using reiser at work (and quite like it for some things), but have recently begun experimenting with SGI's xfs filesystem as well, and must say that thus far I am very, very impressed.

So impressed that, at home, I have migrated from reiserfs (the reiser 2.4.0 patch and the XFS cvs tree wouldn't coexist, though that will probably change now that reiser is in the official tree). For the video editing I'm doing XFS works very well, and the scalability is astounding!

The only thing that worries me is that SGI has commented that they "won't support competing standards" (paraphrased) if the community chooses something other than their work. While I applaud this stance in principle, I think for filesystems it is very misguided. Linux is designed to support a choice of many filesystem types, and it would be very unfortunate indeed if XFS were not among those choices. Reiser is great, ext3, JFS, etc. are probably fine, and XFS (even in beta form) is just plain awesome.

If anyone from SGI should be reading this, I hope you will not construe the inclusion of reiserfs in the official kernel tree to mean the community "has decided" on a standard, and that even if the community had, that work on XFS will continue. Hopefully, when it comes to chocies like which filesystem one prefers, there will never be a "standard," but rather a standard set of choices which will include ext2, xfs, reiser, and perhaps in the not too distant future one or two encrypted filesystems as well.

The other repliers are all correct WRT it wasn't linux, it was IRIX, probably running on a Crimson (was the indy available then?). The program was called fsn and you can run it if you have a machine running IRIX 4.0.1 through 5.3.

It's interesting, becuase the childrens' interaction with the computer system got a lot of space in the book (proportionally way more than, what, 30 seconds or so in the movie). They actually had "screenshots" (in a sort of ncurses-ish way) in there to show what the kids were doing. Pretty interesting. (of course as a whole the book was a lot better than the movie but isn't that always the case?)

In about the middle of Summer 2000, Silicon Graphics Inc. (SGI) came out with the Onyx 3000 and Origin 3000 machines. The "O3K" can take up to 1024 CPUs and 2 TERABYTES of RAM as a single system. For those that want some flexability, it can be "clustered" or segmented into smaller virtual machines in hardware, each running a different OS or different hardware/software configuration. The standard "node boards" that contain CPUs and RAM have 4x MIPS R12000A CPUs running at 400 MHz and with 8 MB L2 cache. A single R12KA/400 compares well in floating point to a 1.0 - 1.3 GHz Pentium III or Athlon. The R14K is due soon, with twice the L2 thruput and higher clockspeeds. The whole system is very modular, made up of various 19" rackmount "bricks" linked together by NUMA-3/Craylink cables. Total system bandwidth on a 512 CPU system is 714.00 GB/sec. From what I understand (I really only work with the much older Origin 2000 series), the standard max configuration is 512 CPUs / 1 TB RAM, but can support twice that with an additional megarouter brick. Expensive machines, but total monsters... and the price per CPU and price to raw power is much better than that of Sun Enterprise. Onyx systems are made by adding one or more InfiniteReality 3 graphics subsystems to an Origin. These SGI systems run IRIX 6.5.X, a real version of unix. Lots of freeware, gcc, etc, can be found for IRIX at http://freeware.sgi.com. Freeware is updated quarterly, the same timeframe as IRIX. Full commercial developer tools (LOADS!) plus an included year of support, software updates, etc, can be had by joining the SGI developer program at the 'Venture' level for $900.

In about the middle of Summer 2000, Silicon Graphics Inc. (SGI) came out with the Onyx 3000 and Origin 3000 machines. The "O3K" can take up to 1024 CPUs and 2 TERABYTES of RAM as a single system. For those that want some flexability, it can be "clustered" or segmented into smaller virtual machines in hardware, each running a different OS or different hardware/software configuration. The standard "node boards" that contain CPUs and RAM have 4x MIPS R12000A CPUs running at 400 MHz and with 8 MB L2 cache. A single R12KA/400 compares well in floating point to a 1.0 - 1.3 GHz Pentium III or Athlon. The R14K is due soon, with twice the L2 thruput and higher clockspeeds. The whole system is very modular, made up of various 19" rackmount "bricks" linked together by NUMA-3/Craylink cables. Total system bandwidth on a 512 CPU system is 714.00 GB/sec. From what I understand (I really only work with the much older Origin 2000 series), the standard max configuration is 512 CPUs / 1 TB RAM, but can support twice that with an additional megarouter brick. Expensive machines, but total monsters... and the price per CPU and price to raw power is much better than that of Sun Enterprise. Onyx systems are made by adding one or more InfiniteReality 3 graphics subsystems to an Origin. These SGI systems run IRIX 6.5.X, a real version of unix. Lots of freeware, gcc, etc, can be found for IRIX at http://freeware.sgi.com. Freeware is updated quarterly, the same timeframe as IRIX. Full commercial developer tools (LOADS!) plus an included year of support, software updates, etc, can be had by joining the SGI developer program at the 'Venture' level for $900.

In about the middle of Summer 2000, Silicon Graphics Inc. (SGI) came out with the Onyx 3000 and Origin 3000 machines. The "O3K" can take up to 1024 CPUs and 2 TERABYTES of RAM as a single system. For those that want some flexability, it can be "clustered" or segmented into smaller virtual machines in hardware, each running a different OS or different hardware/software configuration. The standard "node boards" that contain CPUs and RAM have 4x MIPS R12000A CPUs running at 400 MHz and with 8 MB L2 cache. A single R12KA/400 compares well in floating point to a 1.0 - 1.3 GHz Pentium III or Athlon. The R14K is due soon, with twice the L2 thruput and higher clockspeeds. The whole system is very modular, made up of various 19" rackmount "bricks" linked together by NUMA-3/Craylink cables. Total system bandwidth on a 512 CPU system is 714.00 GB/sec. From what I understand (I really only work with the much older Origin 2000 series), the standard max configuration is 512 CPUs / 1 TB RAM, but can support twice that with an additional megarouter brick. Expensive machines, but total monsters... and the price per CPU and price to raw power is much better than that of Sun Enterprise. Onyx systems are made by adding one or more InfiniteReality 3 graphics subsystems to an Origin. These SGI systems run IRIX 6.5.X, a real version of unix. Lots of freeware, gcc, etc, can be found for IRIX at http://freeware.sgi.com. Freeware is updated quarterly, the same timeframe as IRIX. Full commercial developer tools (LOADS!) plus an included year of support, software updates, etc, can be had by joining the SGI developer program at the 'Venture' level for $900.

For those who want to find out more about this, see the SWF info on PHP's site, or grab a copy of libswf (aka Flash Synthisesizer.)

-Waldo

If you have problems with debugging deadlocks and other thread-related problems, I strongly recommend state-threads. If you are struggling with the traditional thread libraries way of dealing with shared address space, State Threads could be the right choice for a threads library. State Threads were designed for server/client I/O intensive applications. The library is portable, scales extremely well on multi-processor machines, and locking is not necessary since it allows for private thread data natively (no mutexes and locks needed, everything is user-space also). The server.c that's shipped with this library is not your common multi-threaded application, very easy to understand and doesn't suffer from locks. It's GPLd. http://oss.sgi.com/projects/state-threads/

I'm not a programmer and I have never seen it in action, but SGI has an OSS product named Jessie that's written in Java. I don't know what the performance is like.

Anyone wanna hire me, I'm very bored at my present job. The Linux Pimp

All I want now are packets that sound like "Mein Leben" for re-sends after collisions.
--

Another mozilla landmark was reached recently, at least for those of us who use SGI's IRIX. There is finally a working IRIX build, after being broken for about a year. Check out this page for a script and the latest information, and go get an IRIX mozilla at long last.

Posted with nightly build ID 2000120521 under IRIX 6.5. Haven't tried .6 yet.

One might also use the Pro64 compiler from SGI. This compiler does implement IA-64 specific optimizations and it even generates assembler code which is easily readable. The compiler does not come with an assembler or a linker, however, so you'll have to rely on GCC to do that part of the job for you.

Trinitrons look like dog wank once you have used one of these SGI 1600SW flat panels. Best part is, you can get them direct from SGI for either $1495 or $1395 w/video card if you know how.

"Linux sux!" - me