The company is Silicon Graphics Incorporated, or SGI, which once was famous for its high-powered graphics and 3-D workstations but has fallen on hard times of late. SGI now focuses on supercomputers, but there’s a tiny coterie of fans dedicated to keeping the company’s aging but high-powered workstations alive. On a similar note, the every-three-months maintaince release of Irix is 3 months late.
Linux is what happened to sgi. People stopped paying for dedicated hardware and “attepmted” to make due with commodity x86 hardware and linux.
The article makes comparisons against Apple and the ideals behind a spiffy UNIX workstation.
Their stock price gives evidence of the turn of events since they stopped being on the forefront of the UNIX workstation market, and indeed, the decline of that market as we knew it.
http://finance.yahoo.com/q/bc?t=my&s=SGI&l=on&z=m&q=l&c=aapl
Compared to Apple, whose darkest times were during Silicon Graphics’ strongest, you can see the trend dipping downward as interest in expensive, proprietary UNIX workstation waned as Linux came on the scene, and SGI began to sell Wintel-based boxes. Apple went up where they went down.
I have to say that I really like SGI hardware, and it’s a shame that they didn’t target the consumer market, or at least work at getting that technology directly to end users. It would be nice to have them around as a peer to Apple.
doesn’t really tell us anything about current designs or the decision making process or have interviews with key sgi people.
Least it mention Nekochan which is the biggest IRIX community site out there (the forums have over 700members)
http://www.nekochan.net/
http://forums.nekochan.net/
6.5.26 shipped a few weeks ago. I understand that work on 6.5.27 is almost finished.
–ralpht
Dunno where SGI is going these days, but I had a chance a few years ago to spend the better part of a day visiting and touring their primary campus south of SF. Great stuff and lots of fun. Anyone who has a chance should do it. Lots of big honking hardware, a big totally simulated theater/game arena that sat about 50 people (the prized seat was the one with the joystick at front row center), and just a bunch of weird stuff and great toys.
Rick Belluzo.
This guy steered the company straight onto the rocks – He decided Windows NT and x86 was the future for SGI, and started building x86 workstations with integrated graphics that were class-leading in the 6 months after they were released but were doomed to fall behind the performance curve incredibly quickly.
A special SGI version of NT4 was required which meant Win2K/XP don’t work properly, and basically the whole range were lemons.
Instead of capitalizing on the strength of their MIPS-based workstations (Try moving uncompressed HDTV on anything but an Octane2 or better), they basically put their MIPS and IRIX development on the backburner, triggering a massive exodus of talented engineers.
The famous SGI logo was changed to what I regard as a truly retarded, ugly design, and this is indicative of the mindset that architected SGIs doom.
Rick Belluzo then left the company to go work at Microsoft, leaving SGI with a seriously outdated workstation product line, a massive gap to make up in MIPS development and an industry full of customers who figured if SGI was going to do x86 beige-boxes, they could get them cheaper from the guy down the road so thats what they used.
I own an SGI O2, which is a true multimedia computer – with video and audio I/O, 3D acceleration and UNIX OS that made the whole package a rock-solid pleasure to work with.
Sadly, I can’t see SGI returning to it’s workstation roots anytime soon, and they will simply have to make what money they can from their Itanium-based designs.
I, among many others, are very sad to see SGI fall, and will always remember them fondly.
The world has spoken by standaridizing on powerful, easy to use and internet-enabled Windows XP. Linux will be the next causalty of this great business standardization and move forward. I welcome new internet technologies possible by increased collaboration and communciation between standard software.
Explain to me how the “World has Spoken?” Did all the world leaders get together and vow to deter the growth of Linux, as it is the fastest growing OS in the world. And explain to me how WinXP is standardization…unless of course your referring to Microsoft’s standards…then you are entirely right. Internet Explorer, a major part of WinXP, is a prime example of Microsoft setting its own standards and saying to hell with everyone else. But…if you want Windows to come out in the end the only viable choice (and if this is the case a very expensive choice at that when there is no competition….hope you have lots of money). If you were intelligent you would want Linux to give MS a run for its money…as prices will fall. Competition is a good thing, I don’t know why people can’t seem to comprehend that one simple little thing.
>6.5.26 shipped a few weeks ago. I understand that work on 6.5.27 is almost finished.
Well, they should update their irix page then. That’s how I get the news, checking every three months. If they don’t update their pages or release press releases, they should not expect news sites to get the news otherwise.
I love SGI, but imagine if Apple bought them, that would be sweet
Some NetBSD hooligans are keeping old SGI kit alive with recent ports to the Indigo R4k, R3k and Personal Iris machines.
They’re nowhere near IRIX in the a/v department, and likely will never be for lack of documentation, but it’s interesting to run a very modern OS on a 15-year old SGI box.
>The famous SGI logo was changed to what I regard as a truly retarded,
This one was what I recognized as starting point of SGI going down. Apart from bad business decisions, I really can’t figure out why thay changed what happened to be one of the most interesting and bautifully designed logo in IT industry to something really retarded and mediocre. Unfortunately, I don’t see SGI coming back to (more affordable) workstation market anytime soon.
I don’t really think SGI is going down because of Linux. I think the performance increase in PC consumer and workstation graphics cards from ATI and nVIDIA played a much bigger role here.
I think Itanium is what’s going to finally kill SGI — there isn’t a whole lot of life left in Itanic and by Intel’s own words it is already relegated to only very high-end systems — there is no way for Itanium advancement on the low end, which means Intel won’t be receiving enough revenue from this product line to justify moving forward. I’ll give Intel may be a couple of years before they entirely pull the plug on Itanium. If there are any smart people left at SGI, they should be working on alternative designs based on other processor architectures — Opteron or Power would be pretty good bets.
if Nvidia bought out SGI! then they could FINALLY release their framebuffer code, at least 80% of it is based on SGI code… NVIDIA was founded by ex sgi employess so it’d be a kind of reunion, keeping it in the same family. I bet NVIDIA could do some amazing things with that sgi tech too…
Bull. 3dFX, NVidia and ATI are what happened to SGI. The advent of 3d hardware accelleration for PCs made 3d gaming possible- not to mention modelling and rendering. Quake was the first tolling of the Bells O Doom for SGI workstations- the port of Maya to WindowsNT didn’t exactly help matters, either- A windows box can do the job almost as fast for a fraction of the price.
Freenix didn’t eat the SGI mindshare. Commodity hardware caught up and shot past (they’re using ATI chips in some of their machines now), and they’ve done absolutely nothing to keep the edge they had in the early to mid nineties.
Linux will kill IRIX when I can run it on my Indigo2 with full video support, thank you.
keeps irix at least on life support, we’ll be okay. I love Irix and use it as my second OS (after the Mac). I second the wonders of the Indigo2, it’s hands down one of the best machines I’ve ever used.
”
Linux will kill IRIX when I can run it on my Indigo2 with full video support, thank you.”
claims like this never make sense in the real industry. there is NEVER a single feature that keeps one product competitive. so it can be just that one indigo2 video support keeping people in irix or whatever.
its usually a good set of important features or legacy. so if irix is alive then its means it has more than just this stupid indigo thingy going for it. got that?
The SGI machines are elegant to look at and to use. My primary machine is a dual G5 and sitting right next to it is an Indigo2 R10000 IMPACT. Linux doesn’t do it for me… nothing unique about it.
I used to use SGI Octanes at work all day until they were replaced by anonymous (and much faster) IBMs running Redhat 9. I’ve never missed the Irix GUI much I must say, but they had a certain solidity which I do miss. We still use them every now and then for the odd app here and there and for the really good quicktime tools they provided.
Not knowing much about the whole situation, I’ve always wondered why they didn’t make any efforts to bring the Irix interface into the 21st century, since the underpinnings seem really solid.
But jesus, crawl under a desk with one and you don’t half get pounded with hot air – I’m sure my old Octane would take off if it had wheels and wings.
And can anyone tell me, is it pronounced ‘Ir-rix’ or ‘Eye-rix’?
1 x 1Ghz MIPS processor w/ 1MB Level 2 Cache
4 x PCI-Express Slots
1 x ATI FireGL Graphics Card
1 x Serial ATA 160gig drive
1 x DVD R/RW (+/-) Drive
512MB DDR 400Mhz memory
Throw the whole thing in a nice pizza box, along the lines of either an Indy or even a O2 case, and throw a price tag of around $1200 on it, oh, and for christ sake, don’t assemble it in some expensive location in Europe; outsource the damn production to the some company that does the assembling of IBMs Think Centre PC’s.
Talk to the old SGI software providers, sit down, and write these companies a cheque; PAY these companies to get software on SGI machines.
As for IRIX, bundle it with the desktops, free of charge with a standard 5 years free software support (updates and Upgrades), then after 5 years, charge a support fee.
If they did all the above, I would be *more* than happy to purchase an SGI machine; the problem is, they think that the high end will be their saviour; it won’t. What will save them is a strong ISV and IHV network, which can only be created by producing volume systems. SUN got that message, and they’ve dropped their system prices, and thus, their volume has increase. They’re heading back into profitability.
SGI need to do the same thing if they wish to stay in for the long haul.
I had a Indy sitting on my desk with 256meg in 1995. This was the most incredible desktop of it’s time. The funny thing is that I uses my Mac IIci more than the indy. I had no need for that power. I had no need for the memory. I had no need for the video camera that came with it. The desktop environment was ok but I used a terminal app on my Mac to edit and compile my Ada programs. Now I wish I had that old indy……
SGI’s remaining strengths are their excellent shared memory architecture and a fast interconnect. Linux clusters don’t do shared memory and have high latency interconnects. But Pathscale may put a dent in this market. They are also made up of ex-SGI engineers, have embraced Linux, Opteron (amd64), and are marketing a high speed commodity interconnect.
The bad:
– Spun MIPS off into own business.. Failed to capotlized on the CPU built specifically for multi-cpu functions.. spun off other companies that were uncool, when you defragment like that it’s dangerous.
– Sold off most of it’s valueable IP to companies like Microsoft
– SiliconGraphics took most of its attention away from graphics and workstations and visualizations and moved toward super computers—completely insane if you ask me.
– Horrible Marketing
I think SGI will either get gobbled up by someone like sun microsystems or will just die. Nobody really wants SGI that I could think of.
Why? It costs a lot of money to maintain a userbase and integrate a totally different hardware archetecture into your product line. HP had a difficult time doing it and we saw what happend to the Alpha.
Yeah, Itanium was a horrible decision for SGI. IRIX, like many unix platforms were tied to specific hardware too. Should have ported it to x86 and others like sun did.. oh look sun’s stock is up by liek 10%.. hmm..
“On a similar note, the every-three-months maintaince release of Irix is 3 months late.”
First of all, starting from 6.5.22 therre is no such thing as a maintenance or feature release anymore. There’s one release (and .22 is now downloadable to anyone with a free Supportfolio, too). Besides, the freeloaders who only want a maintenance release are not worth much to SGI (hard, but realistic). IRIX 6.5.26 was released in novembre 2004 as notified in Supportfolio as also noticed on Nekochan.net (on wed 10 nov 2004). Press statements don’t really matter much.
“This one was what I recognized as starting point of SGI going down.”
They changed their name (‘Silicon Graphics’ to ‘SGI’) and their logo because 3D was no longer their core business.
I see here mostly the Sun fans saying they should have never ditched MIPS and such. What they forget, IMO, is that
1) In general, customers don’t really care what kind of CPU or OS is under the hood. They care for the overal price / performance, features and service. Unless those customers are geeks…
2) There are all kind of migrations tools for Linux available to get a lot of features IRIX had. Deamons such as FAM, Linux patches such as CSA, they’re all available on the Altix and Prism. Linux saves SGI development and maintenance costs of their own UNIX. Not only that, they also don’t have to invest much in their own architecture or graphic cards.
3) The MIPS workstations were only stopped recently in 2004 with the Tezro being the latest one. I guess the fact they stopped with MIPS workstations is because they weren’t profitable anymore.
4) Wether SGI stops with MIPS or not doesn’t mean the architecture is dead. The architecture ain’t dead, the company behind that is well alive and MIPS continues to live on in various incarnations (such as in consoles, satellite receivers, among others).
5) The workstations, including Onyx, Octane (and later workstations) are still used because they have features current workstations don’t have.
6) Whatever runs on your Indigo2, wether you like 4DWM is completely *not* relevant (anymore). Versions later than IRIX 6.5.22 don’t support the Indigo2 anymore (its unsupported) which means no security updates which means its now nothing more than a toy or somerthing which needs a Fort Knox around it. 4DWM is good enough to get an application such as Discreet Flame, Maya or a CAD application started and used. The WM was never meant for general usage with MSN, and all those other bells and whistles casual users care about these days. Neither was CDE (which you may also chose to run on IRIX). So don’t threat them like they are.
“Some NetBSD hooligans are keeping old SGI kit alive with recent ports to the Indigo R4k, R3k and Personal Iris machines.
They’re nowhere near IRIX in the a/v department, and likely will never be for lack of documentation, but it’s interesting to run a very modern OS on a 15-year old SGI box.”
So does Debian GNU/Linux MIPS port and Linux in general. Here’s a list of supported hardware.
http://www.netbsd.org/Ports/sgimips/
http://www.linux-mips.org/wiki/index.php/Systems
http://www.debian.org/ports/mips/
From what i gathered a while ago, i found out both OSes don’t support the exact same hardware and so it could be needed one has to run one or another to get it running. Its not only A/V hardware which doesn’t work, its a lot more than that. For example, the Phobos ISA NICs don’t work either.
SGI builds their Altix line around Itanium today. But what stops them porting to other CPUs?
Linux is ported to most major platforms ever made. SGI has allready done a lot of work in Linux for scalability to large systems. This work will stay with them if they changed to other CPUs.
Itanium might not be a big problem for them. Itanium works just fine in high end and they would have little problem to offer other alternatives for low end. Their main CPU suplier has enough choises.
Everyone is talking like SGI is already dead, but the truth is that their itaniuem/linux strategy have played out really well. They are no longer in any immediate financial danger thanks to the success of the Altix line.
And the new Prisms are really cool too, and that technology could be used for graphics workstations if they wanted to. And with stuff like montecito and tukwila coming in the future, i don’t see why they would waste any time porting to x86?
Why does everyone say that Itanium is evil? It’s a new technology, and it’s not that bad. Of course, Joe Public won’t go for Itanium system because of its high price, but neither he needs the 6 terabyte of memory installed. Not until some Sierra title states this as a minimal required hardware… But it’s a shame they forget MIPS! And the way new prisms are loaded with Linux, that is just very sad… This is just sooo mainstream and lacks imagination.
Irix should be pronounced with a long I, like “iris” or “wire”.
The rule is that when a vowel is followed by a consonant, and then by an I or an E, the vowel is long. If you want to cancel that effect, you either use a double consonant or put a U in the sequence.
Examples: “tiger” – long I
“Tigger” – short I
“mice” – long I
“Mickey” – short I
“rage” – long A
“ragged” – short A
> Why does everyone say that Itanium is evil? It’s a new technology, and it’s not that bad.
Itanic is evil and it is pretty bad as it was too overpromised and phenomenally underdelivered, Intel managed to fool the market into believing that Itanic is going to be the greatest thing since sliced bread, but turned out to be an absolute dud — Itanic is the biggest failure in the history of semiconductors.
Because, as has been noted, it’s a dud. It’s just not very fast and it’s hideously expensive; the price / performance ratio of Itanic is beaten by some other processor in every single segment (usually the Opteron). Intel has already tacitly admitted this; they started out suggesting everyone (with the possible exception of Joe Desktop) was going to be using Itanic, now they’re saying it’s ‘for high-end servers only’. Which as someone above has noted doesn’t really make any sense either; Intel isn’t a firm that works by providing different architectures for different market segments, its aim is to have a single architecture it can tweak for each segment, which is what it managed with x86 for so long.
SGI simply cannot afford to employ a huge staff of engineers for R&D anymore. What you’re seeing is that SGI will coopt off-the-shelf technologies as much as possible and apply their key IP to that. That is how we get Altix. They can’t afford to continue MIPS development, so they’ve adapted to Itanium. They can’t afford to really continue Irix development, so you’re going to see a shift to Linux. I think there is probably bad blood at the management level between SGI and NVidia, so they’ll work with someone like ATI to use their graphics chips. The profit margin on high end machines is a lot bigger also, so don’t expect a lot on the workstation end. I think SGI will be around for as long as their interconnect has value. I also wouldn’t be surprised if they rolled out an AMD version of the Altix platform–it can’t be too hard.
If I bought a machine, say, off of Ebay, how should I obtain Irix? Once I’ve got Irix, what about security patches?
–Jeff
“Itanic is evil and it is pretty bad as it was too overpromised and phenomenally underdelivered, Intel managed to fool the market into believing that Itanic is going to be the greatest thing since sliced bread, but turned out to be an absolute dud — Itanic is the biggest failure in the history of semiconductors.”
Nonsense. If “Itanium” is a failure then so is POWER – they’re targeted at the same markets. The 2nd most powerful supercomputer at the moment is an Itanium2 Altix from SGI.
Calling it “Itanic” is just as childish as “M$”.
http://www.futuretech.blinkenlights.nl/sgidepot/cds.html
The article mentions this site: http://www.13w3.com/ They sell O2 for cheap, 250 pounds only, and even buy IRIX there… That’s a great way to learn more about SGI hardware and software. Maybe I’ll even buy one for myself…
“If I bought a machine, say, off of Ebay, how should I obtain Irix? Once I’ve got Irix, what about security patches?”
Some machines come with IRIX pre-installed, some with CDs, some nothing. If not or if unsure, ask for that. IRIX is also sold via eBay and via various individuals, corporations. Price depends on how new the version is. Lower than 6.5 are usually cheap. What version you need or will work depends on the SGI machine so do yourself a favor and search what machine you actually want, see what’ll run on it, or arrange you get IRIX with the machine (excellent website about this: http://www.futuretech.blinkenlights.nl and click on SGI). You can download newest 6.5 overlay maintenance releases from support.sgi.com, thats free. Maintenance basically means security + reliability only, not features. Only maintenance releases of 1+ year old are available.
Consider buying a CDROM drive with it (if applicable) it’ll be a tad easier to install IRIX with that than with TFTP / BOOTP. Consider buying a support contract if this is all very important to you.
Also see above mentioned site (UK though), 13w3.com, reputable.com, Usenet sgi archives. Given you’re not from near Germany i ommit naming those sites.
Once a platform has the stink of death on it, YOU CANNOT GIVE IT AWAY. Examples: PA-RISC, Sparc, Alpha.
MIPS is just another in a long line of dead platforms.
Note I am not saying these platforms are without merit…just for one reason or another the market dropped them.
> Linux saves SGI development and maintenance costs of
> their own UNIX.
Anyone who has to deal with the Linux community and vendors like RedHat and SuSE will tell you that jumping on the Linux bandwagon does not save you development and maintenance costs.
The lack of backwards compatibility is frustrating for everyone.
Instead of fixing bugs and working on new features, you spend most of your time arguing to have a simple bug fix accepted.
Linux is still behind Solaris, IRIX and other proprietory operating systems in many areas, its going to take a long time to catch up at this rate, giving M$ every chance.
> Once a platform has the stink of death on it, YOU CANNOT GIVE IT AWAY. Examples: PA-RISC, Sparc, Alpha.
Sparc doesn’t belong on that list, Sparc is very much live and kicking. Fujitsu brought some very interesting well performing technology (SPARC64) to the table recently and so did Sun with throughput computing. Throughput computing will most definitely turn the table in the server space (32 cores running at just 50 watts anyone?) pretty soon. Sun still outships all yearly Itanium shipments in less than two weeks with UltraSparc. If there is a platform stinks of death, that has to be Itanic. RIP Itanium.
If Itanium is really dead (which I don’t seriously dispute) then what about HP’s long-held plans to migrate all or most of it’s OpenVMS and Tru64 customers away from Alpha (a truly dead cpu?) and supposedly onto…Itanium?
Before i enter the university, some prof. told us that there was an SG for student use and programming, but for some dumb administration issues it was dismantled. Really a pity.
Question: can the “open-sourcing” be a solution for SGI?
I think that they should start evaluating the option of at least making a cameo in the end user market, ’cause the masses are really missing ’em. Throwing a special line of workstations for the low end people, following some good agreements with maybe not the leading companies(to keep low costs; maybe some startups consortiums, and 1 big fish–ATI–) but known houses and opensourcing that thing later on the march, can at least produce $$ benefits and can make the collective conscience asking for more…So they can keep that as some kind of “special sale” to get easy money, to stand up a bit more in the industry, and starting to open their thoughts about the end-user market. They can be in the low end market, and yet still be exclusive. That will make some sort of “elite” feeling in their customers, so more and more will “want to join the i-have-a-SGI club”, with the support and innovation from the open-source card. I think that can: work for them, and work for us. What do u think?
they’re fucked, and they need a new strategy. Betting the company on Itanic is turning out to be a huge mistake. Itanic’s failure actually hurts HP a lot more than Intel – at least Intel didn’t have all its eggs in one basket, it’s still got a bunch of other cash cows. (To mix metaphors wilfully.)
> Once a platform has the stink of death on it, YOU CANNOT GIVE IT AWAY. Examples: PA-RISC, Sparc, Alpha.
MIPS is absolutely far from dead, and that you would even place it and all its embedded and research applications in the same category as abandoned workstation/server RISC architectures is laughable.
Funny how some of the posters defend SPARC as a viable product while knelling the death bell for Itanium. Truth is that besides POWER, no one can come close to Itanium’s performance at least in the high end FP arena. Which is what SGI and other supercomputing vendors are targetting…
Just because you can not go to your local retailer and build an itaninum machine to play with during those lonely nights at your parent’s basement, doesn’t mean that the platform is dead.
SPARC has had some serious issues in the past couple of years. First off, since the early nineties SPARC has been pretty much behind the performance curve. Second SUN had to scrap the US V because they just could not get out-of-order working on their own design and had to beg Fujitsu for the SPARC64, which in its latest incarnation implements OOrder and it beats the crap out of SUN’s own US designs. And lastly, SUN has relied on TI as its primary fab partner when they (TI) have some serious issues with their own process technologies and haven’t been able to push US IV to any sort of competitive plateu with respect to POWER5 or Itanium2.
Sad to see SGI MIPS dissapear (not the embedded MIPS which is a separate entity from SGI’s MIPS), but onestly all they have produced in the past decade is new implementations of the R10K core. They had some serious designs in the pipeline in the late 90’s that could have kept them in the forefront with Alpha, but Belluzo made sure his sh*t got all over the company….
> aven’t been able to push US IV to any sort of competitive plateu with respect to POWER5 or Itanium2
Processor benchmarks are a constant leapfrog game, the fact UltraSparc is now a little behind in performance compared to Power doesn’t mean much as UltraSparc IV+ is coming out fairly soon and will probably leapfrog Power in performance. Plus there is much more to processor viability than just SPECfpu or SPECint, which most people do not even consider seriously nowadays — what matters is overall system price/performance, this is exactly where SPARC platforms are quite a bit stronger than both Power and Itanic.
And oh yeah, Itanium is pretty dead — even HP steering more towards Opteron nowadays, there is no future for it anymore.
Obviously you do not seem to understand that the UltraSparc IV is well behind both POWER and Itanium with respect to any sort of significant performance indicator. IBM can push POWER5 to much more aggresive processes than SUN can, and Intel can pretty much put its financial might behind both the Itanium2 and the Xeon64. Either SUN gets something out with the SPARC64 technology in it pronto, or the future of SPARC is more than questionable. Period, no matter how much pink paint you lay over your spectacles.
As for HP moving over to Opteron, well… you seem to think that offering a couple of products based on the processor means that HP is somehow moving over to that platform… hello? Can you get an Opteron Superdome? Can you get HP-UX or VMS based Opteron systems? Since the answer to those questions is no, I sincerelly doubt that HP is thinking about dumping Itanium for Opteron anytime soon, if ever. They will edge their bets, no doubt… but please use common sense… HP and Intel both invested way too much on the itanium to just give it up like some of you seem to assume.
> As for HP moving over to Opteron, well… you seem to think that offering a couple of products based on the processor means that HP is somehow moving over to that platform… hello? Can you get an Opteron Superdome? Can you get HP-UX or VMS based Opteron systems
This probably explains why customers are leaving HP with HP-UX and VMS in droves — absolutely no viable platform for the future. Future of Itanium is questionable to say the least and HP-UX is really behind both Solaris and AIX featurewise. VMS as a platform is already on life support and customers are already pretty feaked about moving from Alpha to Itanic.
> Intel can pretty much put its financial might behind both the Itanium2 and the Xeon64
Itanic already cost about 2 billions to Intel, do you think they gonna keep on bleeding cash on something that is never going to take off? Give me break.
> IBM can push POWER5 to much more aggresive processes than SUN can
As I mentioned in the previous post, processor performance is a leapfrog game and Sun will most definitely catch up with upcoming US IV+ (Fujitsu already caught up with SPARC64 — 90nm process running at 2Ghz). Only Sun systems are a bargain compared to any POWER system sold by IBM, which in a typical IBM fasion comes as a forklift upgrade ($$$$).
In the mid nineties the firm I worked for were thinking of providing third party support for SGI customers. We were primarily a MAC and Windows shop. We were invited to have a look at the Indies and 02’s and I can still remember how gobsmacked we all were at the features and power of these sylish bits of kit. After working with the “revolutionary”(yawn..), new win95 and NT4 PC’s available at the time SGI’s machines seemed 20 years into the future.
A decade later I see the same software features of IRIX in Linux and BSD so maybe thats why SGI’s workstations have become rare items? Nevertheless XP still hasn’t caught up and appears to be the same tired technology I used to find on NT4..
>
I have to say that I really like SGI hardware, and it’s a shame that they didn’t target the consumer market, or at least work at getting that technology directly to end users. It would be nice to have them around as a peer to Apple.
<
Actually, SGI tried that, about 5 years ago. Seems that few people wanted a $10,000.00 PC. I don’t think the vendor lock thing helped. Also doesn’t help to no popular consumer apps.
1 x 1Ghz MIPS processor w/ 1MB Level 2 Cache
4 x PCI-Express Slots
1 x ATI FireGL Graphics Card
1 x Serial ATA 160gig drive
1 x DVD R/RW (+/-) Drive
512MB DDR 400Mhz memory
Throw the whole thing in a nice pizza box, along the lines of either an Indy or even a O2 case, and throw a price tag of around $1200 on it,
——————————
If you’re going to buy that from SGI, figure on $12,000 instead of $1200. Also, *you* may want something, but very few others would. SGI is strictly a niche platform.
They’re afraid of the white-box market because the margins when competing against Dell are so much smaller than those for workstations. So they have been wasting too much time making intermediate almost-white-boxes that are not going anywhere. They did this with almost intel/windows machines and they’re doing it again with the itanium. I’d recommend that they make commodity opteron workstations well tested with the graphics creation software that people use them for and sell based upon brand name. However that brand name is shrinking quickly and I’m not sure it would even work anymore. I expect them to keep kicking around in the supercomputer market for a while but that’s mostly a slow death.
Michael
LOL… may please direct you to the spec website and have you check out the performance numbers for the SPARC64, the Itanium and the POWER5? Maybe real hard numbers may give you the wake up call to reality you so desperatedly need?
And BTW it is ironic that SUN seems to be the ones moving to the Opteron… LOL!
Altix/itanium is a great platfrom to use, I use it myself at work. most of the above posters know nothing about whats really going on, they probably have just read few articles and saw the stocks here and there, or maybe even own an old/outdated system, and consider themselves as experts/super users. what I see above is pages of pointless replies.
this doesn’t apply to all poseters, some posts are interesting, but most of them…