Slashdot Mirror
SGI to sell 85% stake in MIPS
Zebulun writes "the New York Times reported today that Silicon Graphics Corp will sell its majority stake in MIPS Technologies Inc. which produces the RS5k and RS10k processors powering SGIs high-end workstations. This comes not long after SGI announced and began selling intel PII based "VisualStations". "
Andrey gave us the link
to this Associated Press report about it,
commenting "What a sad day it is..."
That makes me very sad. We've known for a while
that SGI would phase out MIPS processors, but
this... it's just too painful. If SGI continues
to produce high-quality equipment, I won't mind so
much, but that doesn't seem likely. At least they'll
keep using MIPS CPUs in their high end servers
Fsck off with the lame "first comment" crap, you troll.
On a lighter note, I see SGI moving towards Intel.. sigh. The Intel behemoth grows. I want MIPS, not Intel!
SGI still employs the team which built the R10k
and continues to develop the high-end MIPS
processor line, e.g. R12k, and on...
The MIPS, Inc. spin-off develops the low-end
and embedded MIPS processors.
Hold your horses people, the MIPS processor
line is still moving at SGI.
Have always loved SGIs outstanding hardware design, i hope they will contuine to use MIPS
processors in their future high-end servers.
(if not MIPS i hope on DEC Alpha 21264/21364)
We all know that MIPS have some problems with delivering thier future R12000 and R14000 processors, release dates slipping time at time.
I hope MIPS will get thier tumbs out thier asses
and deliver the R12000 and R14000, otherwise our lovely IRIX is going to its death. *sniff*
Hm, I think John R. Mashey which was one of MIPS' founders (?) works for SGI - and posts on comp.arch - it's rather cool to be able to read frequest postings from someone that famous. :()
I wonder how he's taking this whole Intel deal..
Imagine, I was just stepping through some MIPS assembler code today to fix a weird shared library problem..
At least he bothered to include something relevant, you tit.
MIPS processors were never top of the performance curve, and in recent times, they have been falling even more behind. I think it was wise of SGI to release that they can't compete in the processor arena with companies that spend tremendous amounts of money to make faster processors. I know everyone hates Intel here, but the PII Xeon is _very_ fast, even when compared with other chips. There is also no way they are going to even come close to the 21264.
Plural??
Nope. IRIX will be ported to Merced. Say what you want about IRIX's security, but it's ROCK SOLID stable. Someone told me his SGI crashed once. I asked him if he was making a joke.
Long live IRIX
sKroz (too lazy to log in)
A fast kludge pile on top of a 8086 if I may correct (And so will merced be if it is true that it can run x86 code). Most other architechtures at least are clean but to Intel the most important things seem to be backwards-compatibility and money.
If SGI continue to support NT and then Intel, then
I should imagine it would be a rather long time
before a true 64-bit operating system would be
available. I hope some bright spark notices a
possible niche, perhaps a 64-bit version of an
Open source operating system?
-kojak
When the R10k came out it was very near the
top of the heap - stale development has been
its worst enemy
And while the Pentium II Xeon has respectable
integer performance, it's floating point
performance is still poor compared to the R10k
and especially the Alpha (which seems to be the
king right now).
they were, if they had the same clock speed
They've been mediocre for some time, at least they are admitting it. Any bets on when SGI disappears?
I read that the fastest microprocessor is currently a PA-RISC.
Hey, there's still HP PA-RISC. Oh, wait...
There's a weird problem with the link to that page that slams it to the bottom where Sun is listing discontinued products. The link is:
But I think it should be:
It seems to make a difference here, maybe due to a bug in Netscape. In my browser (Communicator 4.5), the second version of the link takes you way up the page, where the current generation machines are shown.
Mike Greaves
Let me try that again :-)
i ons
i ons
There's a weird problem with the link to that page that slams it to the bottom where Sun is listing discontinued products. The link is:
/desktop/products/index.html#UltraSPARCWorkstat
But I think it should be:
/desktop/products/index.html#ultrasparcworkstat
It seems to make a difference here, maybe due to a bug in Netscape. In my browser (Communicator 4.5), the second version of the link takes you way up the page, where the current generation machines are shown.
Mike Greaves
I personally dont see all the cause for the bitching and moaning. If you want to see SGI succeed in the market, it has to get rid of its dead weight and get with the program.
:)
It costs SGI on the order of 7x as much to produce one MIPS 10K processor than it does Intel to produce one Pentium II with *better* speed! Who the hell wants that? I dont know about you, but i'd rather have a cheaper SGI workstation, not one that ran on a processor that was rapidly reaching the end of its evolutionary path.
Note, not once up there did I say MIPS sucks. The processors kick ass. However, Merced kicks more ass -- And yes, I know you dont want to hear it, but Merced is where the future is. Not MIPS.
IRIX is SGI's bread and butter. I really dont think youre going to see IRIX vanish anytime soon. Rather than concern yourself with MIPS, look forward to a revived SGI with a touch of Linux.
Makes a helluvalot more sense than whining about a dead-end processor.
Oh well.. the only thing that can make me regain favor is for them to port linux to their new workstations.
Ask and ye shall receive.
It is done.
bah, merced == vapourware
for now it's only living in
Intel/HP dreams (simulators).
Moving optimization entirely from
HW to the compiler is a major
obstacle. Most compilers aren't
very good at optimizing now and
the VLIW technology is going to
make it orders of magnitude more
difficult. Hence performance will
probably be waay sub-peak for
several years while the compilers
catch up.
What's up with this emotional attachment people seem to have developed for MIPS Processors? Why be sad to see it go? Is there some kind of personal bond people have with a MIPS processor?
This is the computer industry. Things get phased out, whether it's cause lack of marketing, technical ability, financing, or whatever. Move on. Retrain.
Besides, why would anyone use an SGI Irix Machine. Use Solaris or AIX instead- they're better in every possible way. There is absolutely no need for an IRIX box.
Also, I like the business model that SGI is now taking: Selling managed high performance enterprise NT workstations and charging a ton for them, stuff IT managers love. I think they're marketing these new workstations properly, and they have tremendous value added in their custom chipsets. (Compaq has been known to make their own chipsets as well.) If SGI can add a web based sales system, they might end up being the next compaq or dell. (Note to self- buy SGI Stock when they start selling on the web.)
The old SGI (IRIX/MIPS) is dead.
-bobby, who's on his fourth job in 3 years...
it's not the cpu -- it's the overall design. that's what's good about sgi. my old indy r5000 kicks ass on any wintel box i've seen -- when it's doing openGL (c:
so, the cpu is no biggy. it's the video and drivers that rock.
sgi is doomed, anyway. the wintel boxes will continue to catch up....and zap! goodbye sgi.
microsoft buttlickers, anyway.
Eras end .... This only surprises ppl who live in the past. SGI followed the ancient high margin workstation model. The world has changed. PPL demand (gasp) value for money. Its time to move on from high priced Unices. SGI milked the niche cow as long as they could... now they can't.
MS has shown the way (and done pretty well) with low-margin-high-volume pricing. Linux, beowulf et al can further improve on this to really bring_computing_costs_down and give power to the ppl.
Lets focus on keeping_it_open and keeping_it_free. SGI wasn't in the game for you, or me. And fwiw, the mips chip could do considerably better after a SGI divestment.
Your talking about the lowest lowend workstations.
What about the research community, that need highend visualation/simulation systems? like Onyx2 and Origin 2000? are there any alternative that can match Onyx2 Infinite Reality2 systems? they are outstanding, the best availble in the world. Every car maker, oil company, drug/medecine maker use Onyx2 for realtime visualation/simulations of thier datasets.
so old != bad
Even if you insist no Mac is ever a workstation (debatable), there are things like RS/6000s etc etc, and of course the newly viable dualboot Mac running linuxppc. I just set my dualboot box up last weekend, and by God it's totally viable- you just have to find tar.gzip versions of things, which is good for you anyway, knowing how to compile things :) :) :) :) ;) isn't it good to encourage types of linux use that tend to create rabid frothing delighted fans? :)
To _my_ viewpoint things are massively looking up for PPC Linux, because I am _running_ it, and I'm quite savvy about Mac issues and would _know_ if MacOS was going to be jerking around linux, and it ain't- MacOS thinks linux is A/UX partitions and leaves them alone, COEXISTS happily with them. Can you say the same for Windows? Does Windows consider linux partitions valid data, or mysterious garbage? Also, we have 'BootX' which replaces the need to mess about with Open Firmware- that's an ease-of-adoption issue and the effects _will_ be known. And I personally am already helping Mac newbies do things like kill kdm and install Window Maker.
PPC Linux lives! Things are looking up for x86 alternatives. _try_ one! Pick up an old PCI Mac and make it a RISC Linux workstation. It's fun to have sweet hardware and a built-in respectable basic SCSI bus! It's never been easier to try Linux. Even if you download linuxPPC (no CD or floppies or anything) you only have to drag 2 files into your system folder, drag the BootX control panel/extension onto the system folder (they will be put in the right places) and reboot- undoing it all is as easy as dragging the files back out again, it's _never_ been so easy to try Linux, or so safe and convenient.
Expect more PPC linux. I'm telling you, the nature of linux is to route around suckiness in the world and find the good answers. Well... looks like linux can get the most out of Apple hardware- and the Mac side _helps_ by being cooperative and adaptable, and the whole process is a _dream_ of simplicity and convenience.
:)
There. heh. er, can you tell I'm enjoying linux and having a good experience with it? I'm just as much of a newbie as some of the PC guys who had hellish troubles and failed and gave bad press reports. Ponder that for a while
Take a closer look. The dead products are the Ultra30 (replaced by the Ultra60), the Ultra-1 (Which is OLD, the UltraSPARC 1 is three generations back!) and the SparcStation 5 (REALLY Old...FOUR generations back.)
Sun is selling its new Darwin systems, and they are selling better than any workstation line in Sun's history. The Sun Ultra5 is the low end system, you can pick one up for less than $3000 if you shop carefully, the Ultra10 is the next level up, and cloned by Tatung and a few others, and the Ultra60 is the latest and greatest and outruns Xeon like a bullet train outruns a steam locomotive.
Even -BETTER- news for Linux fans are the UltraSPARC-II AX motherboards...roll your own UltraPenguin system for less than 2 grand! Put it in that Black Archy case from colorcases.com...
SoupIsGood Food
SGI's Ric Bleuzzo has made it abundantly clear that he is moving everything from MiPS to IA-64 at the high end. IRIX is going to be ported to IA-64, and workstations will all be IA-32 running NT. This was supposed to happen with the advent of merced, and it was supposed to have already happened, but it looks like he'll have to wait until late next year when (if) McKinley comes out.
I wonder where SGI will get fab space to build those R14000s if MiPS tells them to go screw? MiPS is a very succesful company, and supplies their low end chips everywhere...it's not as if they are hurting for buisines...
SoupIsGood Food
Geeze. First you can't get a Sun Workstation (supported but no longer sold) at all anymore. Now MIPS is on it's way out. Who's next? If Compaq follows Sun's and SGI's lead, they will can the Alpha workstations too...
Things are NOT looking up for x86 alternitives. Well, I guess the good news is that at least AMD is making things look good _in_ the x86 world. :-(
www.sun.com then hit "products and solutions" then hit Workstations, and it droped me right to that # tag about no longer making X, Y, and Z....
So, I spook too soon, I happened to come across this SGI thread on /. about the same time I was looking at that, and who'da thought you were suppose to scroll _UP_ to get to where they were suppose to send me ;-)
Spoke too soon... It felt like a dark day, but I guess it's not _that_ bad.
We can hope. What right do I have to say it? I work on it every single DAY.
yadda
There are only a few companies where I truely respect their hardware. SGI is one, Digital is the other. This is such a sad day for SGI, cause in my opinion Mips made them who they are today.
Oh well.. the only thing that can make me regain favor is for them to port linux to their new workstations.
x86 is a standard that _needs_ to be killed. Look at the snivling piece of garbage, its been around since 1977 or something and has its roots in the 8088. The only thing keeping us from moving on to faster processors with better instruction sets and a smoother design is Microsoft! I know the linux development wizzes will port the kernel to these new processors in a week!
The standards that we should be supporting right now are not made by intel! SGI letting go is a step in the _wrong_ direction!
I guess this means that Irix dies as well.....
Do not read this
..about how bad it is to be dependant on a single proprietry vendor? Did the SGI execs come out of the Andy Grove business school?
Who made the decision, BTW. Does anyone know whether it was a marketing/manager type bod, or an engineering based decision (I wager the former.)
Christian
Considering that:
a) Merced is not out
b) nobody really knows when it will be out
c) nobody knows how well it really will perform, not even intel
d) the price difference in R10K and PII
comes mostly from volume
I would say that the aricle this is in reply to is pretty much rubbish.
Let's parrot the previous statement with a little bit of twist to show how ridiculous this is.
"UNIX is an O/S that _NEEDS_ to be killed. Look at the sniveling piece of garbage, it's been around since 1970 and has its roots in MULTICS. The only thing keeping us from moving to faster O/Ses with better fileystems and a more streamlined kernel is Linux! I know the [BeOS, Apple, Microsoft, etc] development whizzes will port UNIX functionality to their own O/Ses in a week!"
See how ridiculous it sounds? Things EVOLVE, people. And evolution does not neccesarily entail replacing things. If it ain't broke, don't fix it! UNIX/Unix clones have been in development since 1970. NT is supposedly (ripped-off VMS code aside) a "ground-up" system and I bet that XENIX is more scalable. (Tongue in cheek statement)
x86 in and of itself isnt bad. However ISA architecture *is* a problem, because the same evolution that took place in the processor did not take place in the bus, processor cache, et al.
I gave you a dollar. Where's my $0.98?
Three Step Plan:
1. Take over the world.
2. Get a lot of cookies.
3. Eat the cookies.
The SGI universe is ending. Time to move on.
I'm not anonymous, and this isn't a flame, but I agree, that "First Post" stuff is childish.
Gee.. maybe SGI's port of 64bit Irix to the Merced might be a 64 bit solution. Aren't Sun and HP also creating a 64 bit os for merced?
I believe Linux might be a non-denominational version that would be a base for attack from the unix vendors, but don't think these well established companies will just throw their own os/2 at linux.
as complicated as the x86 chip is getting, think about it this way... the human brain as we know it has been evolving for many millions of years. It seems to be more of a CISC type thing than a RISC thing, in that it has many built-in instructions. So maybe it is not bad that the x86 has been growing for twenty years.
this is truely a sad day. MIPS was vastly superior to anything intel ever had or will have for years to come.
First, SGI *IS* committed to open source development. I can cite two specific examples:
- Samba 2.0 getting development help
- Linux on SGI/Indy's (http://www.linux.sgi.com), which a lot of us worked very hard to distribute.
SGI has, personally, helped me with documentation on the Indy, loaned me a machine, and hosted SGI/Linux. Don't be so quick to talk!
Also, there *IS* an HPPA port of Linux starting. http://www.puffingroup.com/parisc, for starters.
- Alex
I'm first, get on line
"Where is my mind?"