Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×
Silicon Graphics

SGI Announces New Strategy and Alliance 70

the_demiurge writes "SGI has three press releases on their site showing the new Positions for the Future. They include making separate divisions for the Cray line and the Visual Workstations and also more details on the 'strategic alliance' with NVIDIA." SGI is also getting into Open Source in a big way. Check this page for all sorts of cool stuff.
This discussion has been archived. No new comments can be posted.

SGI Announces New Strategy and Alliance

Comments Filter:
  • Just newfs it, and then restore from a fresh backup. You do have a backup, right?

    make sure you preserve permissions and things when using tar.

    cp -av works nicely too (assuming gnu cp)

  • > Why for the sake of GOD, does sgi not distribute
    > a compiler for free with its OS? This is a very big issue...

    Bzzt! 6.5 comes with gcc *and* necessary libs - you don't need to buy the IDO to get them any more.
  • Well, it is sad that sgi is firing people... I am sure there is some fat in the firings, but I imagine there are alot of people that have work/coded/concepted endlessly and are now being given the boot... sgi is a great reference, I cannot believe that those with skills who are being booted cannot grab another job easily...


    Why for the sake of GOD, does sgi not distribute a compiler for free with its OS? This is a very big issue that everyone I know who codes on sgi's(self include) gripes about(gcc wont do it, libs,.so's,etc)... That is such a destructive attitude to have on a system that has such a small base and software base... I remember getting one of the first indys and having almost no software...

    Why did sgi what until 1999 to go to the desktop graphics card market with Nvida... why did sgi have to wait until its engineers left to go form there OWN companies(3dfx etc)... sour grapes.. I hope something good happens now tho...

    Alias|Wavefront is the jewel performer of the company IMHO... port those (full versions) products (Maya,PowerAnimator|Studio) to every OS that is cost effective to port to...

    Adobe integration, If your OS doesnt run the Adobe products, or run them WELL, the desktop base will suffer(no gimp cries here please... I like gimp, I use gimp while in linux)... Adobe is the bread and butter... what person has NEVER used photoshop and what production house DOES NOT have photoshop?

    I like sgi, I think they deserved a much better market share and growth than they have gotten...

    but it seems, sadly, its not about quality, but quantity....
  • And in fact InsightII has also been on IBM AIX for many years. Not that I'd ever want run MSI software on AIX myself, but it's been there all along.

    If you look closely, MSI like many other ISV's has other plans, like shifting to a web-centric way of providing apps.
  • I am a computational scientist (as opposed to a computer scientist), who doesnt know much about kernel type stuff; however, I see the 2GiB (?!?) memory limit is currently a real limit for ia32 linux for large problems.

    Intel has a pag e [intel.com] that says describes its Extended Server Memory Architecture (ESMA) that extends the limit to 64GB, and says they are "working with the community" to get this into the linux kernel.
    I have no idea what this means.

    SGI's bigmem patch seems to be a little step in the right direction.

    My question to anyone who knows: Is large memory support coming anytime soon??
  • Not exactly. Kernel changes involving changes to existing modules is required by GPL.

    Choosing to not compile a module and replacing it with a closed-source binary-only version is allowed, and a way I'd guess some companies may choose to go in the future to provide value-add.

    And I don't necessarily think that's a *bad* thing.
  • I agree with you wholeheartedly that this is the right thing for SGI to do, and I would like to think that we will indeed see SGI once more a leader in graphics innovation. One point which I think you should've brought up: SGI's alignment with nVidia. The advances in PC graphics technology are nothing to sneeze at, as Silicon Graphics Inc. found out. nVidia, unarguably the manufacturer of the highest-powered (popular) chipsets in the PC market, could have continued to be a rival to SGI, and might have stumped their return to profitability. Hopefully they'll offer us PC users some more eyecandy in the future.

    As Lucas noted in a reply to my original post, SGI has been (and may still be, arguably) the de facto leader in the graphics world. Without them, we wouldn't have OpenGL or a whole whackload of stuff that I'm sure exists, but I'm not aware of. It's nice to have dedicated leaders in market niches (like SGI was), if only for the standards they introduce (GL). Monolithic players (M$) tend to use standards more as a way to compete with their rivals' standards for marketshare (AOL-IM).

    Besides, it -is- the last gasp of a giant... The new SGI is going to be smaller, leaner, and meaner, right? As long as IRIX fans don't storm their HQ.

  • SGI wouldn't switch to Linux if they had a large IRIX marketshare to make a profit -from-... Switching to Linux is a very wise move on their part, maybe the move that'll save their bacon, but it -is- a desperate one.

    This move is both a smart move and desperate move to stay alive. I'm personally not counting SGI out, not by a long shot.

  • Oh well, so much for being able to be used in other operating systems. I guess they're not terribly friendly to "open source" after all..
  • There is an overlapping area between Linux and IRIX, and that's where I think most people have problems. Specifically, the sub $10,000 systems with less than 4 CPU's. Linux does OK on a dual Xeon, and the low end on MIPS has a very high price tag. Then, on the low end of MIPS add a IRIX support contract, and at that point Linux has an edge.

    Take the something from the high end at VAResearch, and compared it to a low end SGI (maybe an O2). But don't forget the hidden costs (SGI's wonderful support contracts, fees, licences for this and that, etc...).

    That's the area where I have seen people pay for the SGI name, but would have been better served with a Linux box. But, heh... we still have a Personal Iris in use here, because it works better as a Xtermial than a Linux box (OpenGL support for one reason, graphics in general, although a good 486 will blow it out of the water for CPU preformance).

  • Register article [theregister.co.uk]
    news.com article [news.com]
    AP Business article on Yahoo [yahoo.com]
  • I wish I could do a
    grep -v 'sentences containing synonyms of "focus on our core business"'

    It sure would make it quicker to read.
  • I submitted this earlier today, but it didn't get posted.

    A message on the XFS Open Source Site [sgi.com] has some new information. XFS is going to be released under the GPL.

    See their PDF Doc [sgi.com] on what this release will cover. It looks like they are holding back the real-time multimedia features of XFS, but that's about it.

    Some source is already available for download, the rest will be up as soon as they clean up the source code for GPL'ing.

  • Can anyone give a ETA for the XFS being added to the kernel tarball?

  • Current features include:

    • 9 exabytes file size limitation
    • Near-raw I/O performance. An SGI Origin 2000 was able to reach 4 GB/s (rw) on 704 disks to one multithreaded process.
    • Quick Recovery, less than a second after an unexpected interruption regardless of the number of files it is managing. (no more fsck'ing fsck :)
    • File system size limitation of 18 exabytes
    • Hierarchial storage (HSMs, DMF)
    • Table structures implemented in b-trees for fast searches and space allocation
    • Block sizes are 512 bytes to 64 KB for normal data, up to 1 MB for real-time data.

    I hope that clears it up.

    As far as a converter, well... probably not for a while.

  • Is anyone thinking of a Ext2 to XFS filesystem converter? OK, I can see instant problems with this; how do I change the root filesystem without killing everything?

    Could anyone tell us what benefits using XFS would bring to the typical Linux desktop user? Increased Speed?

    OK, better reliability and faster fsck's, but apart from that?

  • I really like the direction that SGI has taken lately(aside from the foolish name change). Their focus on open source has been great. Now, as a Cray T3e programmer, I have a request: Please update the MPI libraries to MPI2 compliance. For those of you who aren't familiar with parallel computing, MPI(Message Passing Interface) is the industry standad for parallel computing. It can be used on a single processor PC all the way up to a 2048 processor T3e with little porting overhead. The problem is that nearly every implementation is of the older 1.2 implementation. MPI2 allows for trmendous speedup. If both Cray and MPICH, a free implementation, go to MPI2 compliance, then Beowulf clusters could fit more easily into many research environments. As it currently stands, the code bases for Supercomputers and clusters remain seperat entities in many cases. MPI2 would solve this. Oh well. Just my obscure wish, though.

  • not sure if it will all be included but...

    faster, very nice repair, faster format, xfs_growfs(for if you add another 8 gig, making the existing filesystem grow, w/o disturbing the original data), logical volumes...

  • How long do you think it will take for XFS to become a standard in the Linux community? Anyone?

    Dunno, but as long as there aren't any funky restrictions then I would imagine that XFS will become a standard very quickly.

    Personally, I'd reformat my Debian Potato Install for this. Anyone (in the know) care to comment on when we'll see XFS in a Kernel tarball?

  • Poor SGI, how the mighty have fallen. It seems that many companies which have come on hard times turn to Linux for salvation. Will it help them recover? We shall see. Does their code contributions help Linux??? HELL YES. A journaling filesystem is a huge step for Linux. As a rabid Be advocate, I preach about the BeFS a lot. Now, (about time), Linux can enjoy all the advantages of journaling. It's really impressive how much of a beating these filesystems can take before errors crop up.

    How long do you think it will take for XFS to become a standard in the Linux community? Anyone?

  • As a T3E programmer, what do you want in MPI2? It doesn't really seem to me that there's much in there that will make much difference in terms of speed on MPP machines - maybe the IO and onesided stuff, but the dynamic process creation is pointless on a T3E

    No one has done a full MPI2 implementation yet, except for Fujitsui
  • OK, better reliability and faster fsck's, but apart from that?

    Well, XFS on SGIs does 64-bit files ( > 2GB), guaranteed rate I/O and better I/O performance in general. The user can also specify the block size (within a certain range).

    It also has journalling so you can recover after a system crash - fsck is not required on XFS filesystems.
  • Why do people seem to expect that XFS will automatically be added to the main kernel tarball? Nobody outside of SGI has even seen the code yet; it's far too early to speculate on whether or not it'll be added to the mainline kernel, let alone when.

    Sure, maybe it'll be a great feature that everybody wants. Or maybe it'll be 100k lines of heavyweight filesystem code that very few people want, and others will whine about how much it bloats the kernel. There's no way to tell yet. Give it time, okay?
  • And all slashdoter's are worried about is
    the IRIX source code. You people are sad.

    Oh yeah:

    "this logo's for you!"
  • I agree as well. SGI could really cut down on overhead by adding all their cool IRIX stuff to the Linux kernel and then abandoning IRIX altogether. That way they can focus on hardware and they have a cheap OS that is maintained for them. Maybe the prices of their workstations will drop.

    Plus, Linux on an SGI box would make it a helluva lot easier for me to get that Indy up and running at work :-)

  • I've been looking around the SGI website and it isn't immediately obvious to me who to write to at SGI to thank for all these wonderful Open Source donations. Anybody have email addresses that we could slashdot with thank-yous?

  • What makes you think that a SGI Linux support contract is going to be any cheaper than an Irix Support Contract?

    We keep hearing that the money in Open Source is in services, and if you're going to spend the dough on an SGI Linux box, you probably are going to want SGI to support it.
  • I really like the direction that SGI has taken lately(aside from the foolish name change). Their focus on open source has been great. Now, as a Cray T3e programmer, I have a request: Please update the MPI libraries to MPI2 compliance.

    Nobody (that I know of, anyway) has a complete MPI-2 implementation. There are a few MPI parallel IO implementations out there (PVFS, ROMIO, and one from IBM whose name escapes me), and one of the free MPI implementations (LAM) has the new dynamic process allocation mechanism. Nobody's implemented the event model and some of the other stuff though, because a lot of it is hard to do and there hasn't been that much user demand. It's hard enough to convince users to port to MPI-1.

    If anybody knows of an MPI implementation that implements all of MPI-2, I'd love to see it.

  • One may ask the rather pertinant question for SGI, where are the profits in the computing industry?

    If you compare it with other industries, two models usually work, high volume & thin margins or a specialist niche. Big players like IBM have discovered that the PC market is basically a money drain within their corporate structure. I think SGI are discovering the same phenomena with their smaller volume of sales. The big problem is that their traditional primary market (graphics) has suddenly shifted into the mainstream PC workstation range and until they can define/create a new market for their high-end wares, are effectively losing customers at the margin which is hurting them as they have longer product cycles than the current disposable PC (when you get a 486 chip in mobile phones/PDAs with shelf life of 12-18 months, you know something is crazy).

    Personally I see them releasing people wholescale as a bad move as their core strength is in the knowledge that their staff carry around in optimising software for high-end kickass machines. SGI could provide a valuable service by hiring their people out to other companies in order to tune specific applications to go like the hammers of hell and provide a cost-effective hardware platform and software APIs to support them. In other words, loosen the reins before they jump ship. Without the skills and detailed know-how of coding to the wire, we will otherwise be left with bloatware relying on Moore's law to progress.

    Economists have concluded there are several ways of making a guarenteed profit, rent-seeking monopoly (no guesses here as to the most successful culprits), efficiency differential between low and high cost providers, permanent gains due to unique technological breakthroughs and inelastic demand. As SGI is in no position to be a monopoly and their cost structure is inappropriate for a high-volume box shuffler, IMHO their long-term hope is to develop/morph new technologies, and spin them and associated staff off into as many subsidiary companies as quickly as possible to gain the capital appreciation. Given the rather dynamic, coopetitive and fluid nature of the computing market, I suspect it is easier to ally with siblings than with potential competitors.

    Well, if not then the fate of DEC is illustrative of one path that SGI could face.

  • Well, currentlly, to get the newer versions of IRIX you need a support contract. Sure, they will probably have one for Linux. How are they going to not be in violation of GPL and prevent you from upgrading Linux on your own without paying them?
  • People assume it will be added to the main kernel tarball because it is a filesystem that a lot of people want. It only bloats the kernel if you compile it in, you know.
  • Photoshop does run on SGI, but it's a horrible, horrible port of version 3.0. (the Mac versions are up to 5.5 by now).

  • No, I don't think so. You only have to transfer the copyright to the FSF if you want it added to the mainstream distribution of a GNU program - like gcc, etc.

    Linux is NOT a GNU program. AFAIK, everyone who submits code to Linux still has complete control of it and must agree to license changes.

  • Building on changes that started 18 months ago, SGI today announces a series of initiatives and partnerships that will enhance the company's position as the leader in high-performance and visual computing. These actions will accelerate SGI's transition from a vertically integrated company to one committed to broadening its market reach by leveraging its technology through new partnerships with industry leaders.

    I realize that's the opening paragraph of the press release, so we'd expect it to be vague. But if you replace the "SGI" with another name, it could be used to describe the re-org of any high-tech company. I'm in awe of the ability of marketroids to come up with completely meaningless fluff like this.

  • Lots of people seem to be slating SGI recently, both for their recent performance and for the radical steps they're taking to turn themselves around and get back to being profitable.

    Nobody can deny that SGI's performance over the last few years hasn't been terrific. Their once-profitable principle strategy of selling high-performance graphics workstations became stale as PC technology moved forward, with 3D chipsets, etc. which could compete with SGI products like the O2.

    Their servers are now regarded as being relatively over-priced and not worth the extra performance (e.g. memory bandwidth, etc.) that they feature.

    SGI was originally a company which developed cool technology. They parlayed that into a range of products which, initially, found a receptive market. Since then, however, the market has moved forward, but SGI have failed to move fast enough to keep up with it.

    The plan accelerates SGI's transition from a vertically integrated company to one committed to broadening its market reach by leveraging its technology through strategic partnerships with industry leaders.

    What this says to me is that they're refocusing on technology - i.e. coming up with cool stuff, basically - and moving away from building their own products towards selling the technologies they develop to other companies, in the same way as companies like Creative developed technologies like Sound Blaster and marketed them in partnership with other companies.

    They're taking radical moves, shedding divisions which aren't profitable enough or which don't form part of this strategy and embracing new developments, such as Linux. The Cray and Visual Workstation lines are being seperated as a precursor to possible sale to a third-party who will continue to work with SGI in developing the product lines. This implies that SGI will develop the technology, leaving someone else to worry about the actual box-shifting.

    All of this is pretty much a textbook business strategy for conducting a turnaround of an unprofitable company which had begun to lose touch with the market.

    MrEd just described this announcement as the "last gasp of a giant" and whilst I would agree that we are witnessing the end of the old Silicon Graphics, Inc., I'd say we're about to witness the rebirth of a new SGI.

    I think that we're going to see a leaner, fitter, better company emerge from this reorganisation, slimmed down and better able to move quickly and take advantage of developments in an industry which is probably the fastest-moving in the world. I can see SGI becoming a significant developer of technology and a significant player in the future development of Linux, and I think that in a few years, we'll be recognising that this was, in fact, the right thing for SGI to do.

    ..is for Devious.

  • Just because SGI has decided to embrace Linux dosent mean they are dying. In many respects, it is a very cost effective measure. Even IBM is turning to linux in a big way. Simply put, it is cheaper for SGI to port the few high end features from Irix to Linux for there IA-32/IA-64 servers then it is for them to port the whole of Irix over. in addition to this, there customers get the benefit of patches for problems in under 12 hours from when it becomes public knoledge. also, they will get quite a few good features from peaple outside of SGI. Simply put it is cheeper for them to use Linux then Irix. There are probably very few peaple that buy an SGI machine just because it runs Irix, from what I have heard, Irix can be rather buggy to work with. Dont count SGI out yet. I personaly think that this is a smart move by SGI, not a desperate move to stay alive...
  • Your thanks is best appreciated by buying the
    products and services, enabled by the embracing
    of OpenSource. The "Thanks-you's" will go directly
    to all the people being layed off to make it all

    Thank you!
  • How did you manage to figure that out from ~23 comments?

  • Oh, willyoupleaseshutup!?

    This would you prefer? That they laid off half their staff or carried on as they had been up until recently, and go under in a year or two, with everyone losing their jobs.

    From what I've heard of the US jobs market in the computer industry (and personal experience of the number of Americans who've asked for a copy of my resume), 1,500 ex-employees of SGI would get snapped up pretty damned quick.

    ..is for dastardly.

  • SGI has a lot more to contribute to the computing world than a few extra functions for linux. They're kind of like Apple in that respect. Even if you don't use their computers, their innovations are felt around the industry. Witness OpenGL, for one.

    If SGI trully does die off, who'll take over the leading edge of graphics? I may be wrong, but it doesn't seem at all that anyone besides SGI (OpenGL) and Apple (QuickTime) have really contributed such radical thoughts to this market segment. Other companies just reimplement their idea's in less graceful ways (Direct3D, ActiveMovie) or simply boost the clock rate...

    Same as if Apple had died, if SGI dies, we'll all feel the pain, eventually.

  • SGI is going to throw a lot of thier long term high dollar customers into total shock. I know a couple people personally who just bought new SGI's within the last year and don't want anything to do with Linux. :-) Heh...

    Oh, this will be fun... Finally I will get to say "I told you so" to some people who thought SGI was the only way to go, and Linux wasn't even something they could consider.

    What I wanna know is how the SGI ISV's are going to handle this news. Like MSI [msi.com] who have always been "we only port to IRIX" kind of people.

  • Thanks, I have looked at the kernel. But thanks for being condescending anyway! I appreciated it.

    Where did I say that there's no chance XFS will go into the kernel? All I said was that it's not guaranteed to go in and that it's too early to give time estimates.

    XFS is a lot of code; some time back, one of the SGI folks working on it estimated that it was around a hundred thousand lines. That's a lot of code. Assuming that Linus et al do want it in the kernel, that's still a lot of code to check through and test. Filesystems are hard, especially filesystems as complex as XFS. It's not like dropping in another driver for a random piece of hardware; there's a _lot_ of work that needs to be done to get XFS to the point where it could even be considered for inclusion in a mainline kernel.
  • True, but it makes the kernel tarball bigger whether or not you compile it in. Current kernels are already about 13-14MB compressed; I wouldn't be surprised if people complain about including large chunks of code that not everyone will want.

    Note that I'm not claiming it _won't_ be added or that no one will want it; all I'm claiming is that it's not necessarily just a matter of "cool! when will it be in the kernel!". Nothing is guaranteed.
  • Linux has needed something like this for a long time. I can't wait to dig my teeth into this one. Even more lovely will be when it is finally integrated into the mainline distributions (Debian, etc.)

    Thank you SGI!

    The choice of the GPL license is also another smart move. The GPL is a great license for this kind of technology, as it will ensure the freedom of the code for everyone and make it resistant to proprietary forking. I may not be an expert of license reconciliation but I don't see with the *BSD operating systems won't be able to use XFS? Is it simply impossible to include a GPL'ed filesystem with a BSD OS?

  • I'll bet that 18 months from now we'll all go: "How did we ever survive with out it [XFS]?".
    I personally will use it the second it comes into the kernel tarball.

    LINUX stands for: Linux Inux Nux Ux X
  • Great choice of license, truly 'giving' back to the community.

  • >Oh, willyoupleaseshutup!?
    gosh that got my attention.
    >This would you prefer?
    I would prefer SGI did what I liked best,
    not buy into ubiquitious pc technology.
    >From what I've heard of the US jobs market in the >computer industry
    Well, after 20 years in the computer industry I
    can tell you "pretty damn quick" aint quick
    enough sometimes.
  • Actually, I am refering specificly to thier Insight package that they aquired when they bought out BioSym, and which has always only been an IRIX product. MSI has ported some of thier products to other OS's, so that wasn't a totally fair comment.
  • And I can tell you, after 2 years living with XFS, I'd never go back. These features rock!

  • gosh that got my attention.

    *smug look*

    I would prefer SGI did what I liked best,

    Well, what's that?

    not buy into ubiquitious pc technology.

    Well, they are spinning off their Visual Workstation division, with a view to finding a partner to take over operation of the new company...

    ..is for deranged.

  • Get off the lay off issue already. You already made your point earlier. SGI is stumbling. I wouldn't say falling yet, but definitely stumbling. The way things were wasn't working for them. They had to lay off people. They are trying to change their direction based on the way they see the direction that computers are heading. They aren't laying off people because they changed their business strategy, they are changing their business strategy because they had to lay off people. And as was previously mentioned, its not like it would be hard to get a job somewhere else after working at SGI anyway.

    Just my 2 cents.
  • All of our work on the Linux kernel front is released under the GPL (GNU General Public License.)

    Sharing kernel changes is legally required by the GPL, but I wish SGI volunteerly will support and give some credit to the GNU project, if they will use GNU components in their distribution.

Why won't sharks eat lawyers? Professional courtesy.