Sun Microsystems has discontinued development of two planned chips as it retrenches in a difficult era for the company. More at News.com.
Sun kills UltraSparc V, Gemini chips
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Sun would be much better off if it dumped its whole java and SunOffice operations. These areas are a massive financial drain on the company.
They may be able to find a buyer for StarOffice, but java is dying anyway.
well, after reading the article it doesn’t seem as bad as it sounded at first. If it helps them get back on their feet, then it would have been worth it.
As long as they continue work on Niagra they’ll be fine…
hello.
i thought sun had already stopped using their own processors?
any idea why i would think that?
As long as they leverage Linux and use it wisely to recoup potential loses, they’ll be fine.
That will give them time to focus on their strong points (hardware engineering and services) while they seize the opportunity to cut costs tremendously by sharing the cost developing software with the open source community. (Linux, {Java?}, {Solaris?})
Unfortunately, this is SUN we are dealing with.
Yeah, it doesn’t sounds that bad.
From the news, it looks like Gemini is a dual-core, just like current UltraSPARC IV. But the core of Gemini is older
(somewhat a testbed for UltraSPARC IV??).
So, from technology point of view, UltraSPARC IV already surpassed Gemini.
Do I understand that correctly?
For UltraSPARC V, it looks like the reason is
the new architecture requires customers too much effort for transitions.
May be Sun wants to avoid itself from the Itanium scenario.
… likes to hear opinions on this, cheers
maybe they should just let fujitsu handle the hardware while the focus on the software and then they could each create packages by mix and matching from the two companies?
hahhaha this is actually good. When I read the header i was like.. OH MY G*D
Sun was working on too many chips. You know they are developing them and testing them out. I bet they have several chips that you aren’t hearing about.
This is great for sun customers actually, we don’t have buy completely new servers. Sun was working on too many chips anyway. It would cost too much to make completely new hardware and support existing hardware.
Then again.. wouldn’t it big a big boost in sales if people purchased new systems? Offer upgrades to the same yet offer a new system that is great? I guess they may be right.
Interesting.
maybe you have been listening to Sun’s competetors and people who don’t know what they are talking about. There’s alot of those people who have nothing to do with the company and aren’t even sun customers that shoot their mouth off.
I think fujitsu and sun hardware should merge. and be the new “sun microsystems” and split off their software business and call it “Java software” or “Sun software”
Why not? issue stock for the 2 yet keep them under the same management like how Sprint and sprint wireless did.
Oh, and about open source. Why would you want to sell your rights off to the OS community? a million other companies would offer the same product! LOL. Seriously, people could get it free then.. Johnathon told me java IS profitable, he claims its a “spin from IBM” to get sun to get rid of it
As a hardware company, Sun has a great future ahead as the “Dell of Linux”. By selling Opteron systems, Sun could even sell into Windows accounts.
If Sun goes with AMD, they will have better price/performance than Intel for the forseeable future. So they’d be up on Dell.
On the software side, unless Sun open sources Solaris and Java both technologies are dead.
Despite all the FUD people keep spreading about “killing off” UltraSparc, UltraSparc has got by far the best story to tell in 64-bit market even though it would be nice if Sun could make it more competitive (Fujitsu can help them do it). It is going to take years (if ever) before Opteron will reach the same level of support and credibility as UltraSparc, Itanic will probably get slowly swept under the carpet, PA-RISC and Alpha are already dead and forgotten, Power is a great chip but it is having a hard time gaining more traction in enterprise market, which leaves UltraSparc as the only credible and proven choice in 64-bit market with a company that is 100% committed to its future. Bottom line is UltraSparc is here to stay for a long time and we could actually see it grow in the future.
sun can’t open source solaris due to license agreements with SCO. and solaris’ growth rate is increasing as sun has focused back onto it. I highly doubt solaris is dead because I use it at home and its AWSOME. Why not download it and try for yourself?
This is the only financially viable decision. The Sparc line, for better or for worse, is no longer commercially viable. Read that – COMMERCIALLY viable. That does not mean it “sucks” – it just means people have stopped investing in it. Sun server sales have fallen off of a cliff and they lag far behind HP and IBM by more than half.
The issue isn’t if Sparc/Solaris is DOA – the market has already determined it is – the true question is whether Sun should get out of the hardware business entirely. At the very least they should be pushing Opteron boxes as a way to promote Linux and their “Java Desktop”/StarOffice combo.
In any case Sun is going to be a smaller company – much smaller. The deal with Microsoft can be seen as Scott McNealy basically signalling to the world that he is starting to read the writing on the wall and he isn’t going to burn up any more shareholder value fighting a war he can’t win…and he can’t. When an architecture dies in the marketplace it is ugly but you have to leave it at the side of the road and move on. sell the intellectual property to someone who thinks they can revive it.
Java, Linux, Opteron. Those are the three words (in that order) Sun should be focusing on. Forget everything else and can anyone who doesn’t bolster the support for those techs.
>> Despite all the FUD people keep spreading about “killing off” UltraSparc, UltraSparc has got by far the best story to tell in 64-bit market
No one cares about “best in class” – Sun server sales are nosediving. Just another poster who confuses technical viability and commercial viability.
>> Power is a great chip but it is having a hard time gaining more traction in enterprise market,
IBM has more than 2x Sun market share in servers.
So does HP.
>> which leaves UltraSparc as the only credible and proven choice in 64-bit market with a company that is 100% committed to its future.
We already know Sun is not “100% committed” to Sparc by virtue of this story. And as Intel has shown, 64bits is a nice goal but what matters is PERFORMANCE which drives SALES. the 32bit P4 has no problems trashing the Sparc line. Look at Sun’s revenues and look at Intel. Clearly the market is telling you something about the pressing need for 64bit.
Sparc is open design. http://www.sparc.com. Its not like Sun is the only one using Sparcs.
>>> Power is a great chip but it is having a hard time gaining more traction in enterprise market,
>IBM has more than 2x Sun market share in servers.
>So does HP.
There’s for entire server market.
For UNIX market, Sun still #1
—-
I also agree that it’s likely to be a good move,
killing off something that may hard to sell.
Anyway, keep in mind,
the news just said that it just killing off
only some branches — NOT its entire processor development.
As I’ve heard in news/press release many times,
Sun still positions SPARC as its only platform for datacenter/mission critical tasks,
while puts IA to edge-servers and network appliances.
(it is not about size, an edge server can be bigger
than a datacenter server.
but the diffs are ‘criticalness’ and how to ‘scale’
— note: *both* SPARC and IA do scale, with different strategy, and both are doing well in their suitable job.)
Opteron and UltraSPARC has at least one thing similar,
it really cares about processor I/O.
And that serves Sun vision on Throughtput Computing.
a bottom line, this move is about focused, not abandon.
here is the ‘official’ url
http://www.sparc.org/
Sun focuses on a roughly 5 year plan. If you buy a Sun machine, chances are it will be supported for roughly 5 years after it has been discontinued. If you look around, you will probably see OLD sun machines still in use. I see Ultra 5s and Ultra 10s all of the time. Releasing 2 different processors in the same year, or even consecutive years won’t help Sun at all.
The UltraSPARC V, according to the article, would change things enough that customers would have to migrate. That’s a major pain for someone that has been using an E4500 for the past several years. Migrating between differing platforms is just too difficult in some areas, a lot of the areas Sun really focuses in.
My opinion, of course.
> No one cares about “best in class” – Sun server sales are nosediving. Just another poster who confuses technical viability and commercial viability.
The only reason Sun is lagged behind in server market share is increase in sales of very low end single processor servers, which is not even where Sun wants to be in the first place. This very low end of server spectrum has practically no COMMERCIAL (you really like that word, don’t you) viability for any company except maybe Dell — low spec high volume machines have practically zero margins and are pushed by HP , IBM and alike just get the bragging rights of increasing server market share. Middle to high end server shipments are actually going up at Sun with a lot of those displacing IBM mainframes and PA-RISC/Alpha dead-ended machines.
> the 32bit P4 has no problems trashing the Sparc line.
This may be true if you look only at some synthetic processor benchmarks not reflecting the real world applications. From system perspective UltraSparc machines are still beating the crap out of Wintel boxen sometimes being even cheaper, just look at Sun Fire v440/480/880 servers with Sybase and Oracle databases.
The Sparc line, for better or for worse, is no longer commercially viable. Read that – COMMERCIALLY viable. That does not mean it “sucks” – it just means people have stopped investing in it. Sun server sales have fallen off of a cliff and they lag far behind HP and IBM by more than half.
The issue isn’t if Sparc/Solaris is DOA – the market has already determined it is – the true question is whether Sun should get out of the hardware business entirely. At the very least they should be pushing Opteron boxes as a way to promote Linux and their “Java Desktop”/StarOffice combo.
There you are again. Sun just released the UltrSPARCIV line.
Sun shelves UltraSPARC VI in favor of The Rock
http://www.theregister.co.uk/2004/02/12/sun_shelves_ultrasparc_vi/
Sun is moving UltraSPARC to the throughput computing era. In case you haven’t noticed IBM, Intel and AMD are all headed that way. Sun may have missed the ILP performance era but the new era is the TLP (thread level Parellelism) era. TLP is the natural evolution of processors. Large OOO Execution processors are not longer viable and are reaching thier limits. Notice how the P4 architecture pretty much got stuck at 3.x GHz for a year. Intel had to add more cache. Prescott can go at most 5GHz but its extremely long pipeline starts to be problematic and the clock rate increase is going to hit diminishing returns.
TLP and the Return of KISS
http://aceshardware.com/read.jsp?id=60000312
Every chip manufacturer has announced multicore cpus by the end of 2005. T
I see people making claims that if sun wants to “survive” that it should make Java and Solaris OSS. Another guy saying they should be Like Dell/Windows and sell Opteron boxes. Maybe you should need permission to publicly share your opinion. Not to be rude but Dell is a MAJOR vendor of Linux servers on x86 hardware. It is pretty hard to compete with Dell/Linux on price and Sun is not in a position to do so. The battle for the small/mid range stuff is already lost against the Xeon/Linux(and Windows!). Sun still has a strong market in Big iron high availibility systems. Thinking it is Good Buisness for Sun to throw away everything it has to join in on the OSS/AMD march is nothing short of your shortcomming and inibility to make reasonable judgement.
In any case Sun is going to be a smaller company – much smaller.
I doubt it, because…
The deal with Microsoft can be seen as Scott McNealy basically signalling to the world that he is starting to read the writing on the wall and he isn’t going to burn up any more shareholder value fighting a war he can’t win…and he can’t.
Teaming with Microsoft just means he has started a new “war” — just not against Microsoft. I’m sure you can realize what it means that Sun will be offering better Windows interopability than *every other* Unix vendor on the market.
No matter how Sun spins it everybody knows that Sun will never make a Return On Investment for java. Open sourcing java won’t help them out in any way obviously, but they will never even be close to recouping the costs of java development no matter how many J2EE licenses they sell.
IMO, sun is a hardware company that really should be a sofware company. Sun just can’t afford the billions that go into chip R&D.
Making sure Java is as popular as possible is smart for Sun because Java takes beefier hardware to run and this is good for Sun’s bottom line.
It seems Sun is the only company that doesn’t get it. Keeping Java from evolving is the same thing as killing it. The world moves on even though Java stands still.
Sun doesn’t have any long term sustainable competitive advantage. Their time is done and they are just playing out the game.
> Sun is on billyg’s new mission to support everything that is proprietary and kill everything that is open. It is up to IBM now to give the customer choice as all Sun will give them is proprietary solutions — either Windows or Solaris. If the future of the enterprise is open, it won’t run on Sun.
Dude, whatever you’re smoking there has some serious paranoia inducing effects on your brain. Sun didn’t sell out to MS and if anything Sun’s software will be less proprietary in the future. Why do people come up with drivel like this when Sun is one of the best friends OSS ever had. After all Sun is the second biggest contributor to open source after UC Berkeley. On the other hand IBM is just wrapping itself in Linux and OSS flag trying to make an impression of an “open” company at the same time still delivering the most closed and proprietary software/hardware out there.
In a recent SUN thread here is OS news, some SUN zealot
was going on about how “is not unreliable”, with the arguments that SUN has a firm plan for the future,
and it already has “several Sparc chips” down its pipeline.
Well, here we are a week later.
and it already has “several Sparc chips” down its pipeline.
Well, here we are a week later.
If you didn’t read my previous post sun still has several SPARC chips in the pipeline. Funny.
To Niagara and Rock family of processors have been in the news for a while now.
Architecting the Future: Dr. Marc Tremblay
http://aceshardware.com/read.jsp?id=55000245
Dr. Marc Tremblay Sun Fellow. You should read the article.
IMO, sun is a hardware company that really should be a sofware company. Sun just can’t afford the billions that go into chip R&D.
Sun had about 5 billion in cash before the microsoft deal now they whould have about 7 billion in cash. They don’t fab thier own chips. How many billions would they need to do R&D if 7 billion isn’t enough?
blah (IP: —.dsl.snfc21.pacbell.net)
Java, Linux, Opteron. Those are the three words (in that order) Sun should be focusing on. Forget everything else and can anyone who doesn’t bolster the support for those techs.
Really what pray tell are Sun’s current customers to do with there billions invested in SPARC Solaris?
Let’s take HP. You said HP has more server marketshare than Sun. Did you forget that HP is now compaq and HP together. Also HP has been trying for years to get thier customers to move off of Alpha, VMS and tru64 to HP-UX linux on thier intel boxes itanium and x86. They haven’t. infact they still have thier Alpha server line and are releasing one new version EV8 for the handful of customers that still use Alpha boxes.
Alpha doesn’t even show up on server market indicators. SPARC the most succesful 64-bit chip to date.
Large enterprises don’t work like the PC enthusiast market. They don’t upgrade every year to the latest and greatest. It would be too expensive for customers to move to linux x86 from Solaris/SPARC. They would have to rebuy all thier licenses for thier software, imagine moving a Oracle database to linux on opteron they were expensive when they bought it for SPARC and are still expensive for linux on opteron. The cost of moving platforms is non trival. Even enterprise linux from redhat does in the $18000 dollars/cpu price range for that level of support.
We already know Sun is not “100% committed” to Sparc by virtue of this story. And as Intel has shown, 64bits is a nice goal but what matters is PERFORMANCE which drives SALES. the 32bit P4 has no problems trashing the Sparc line. Look at Sun’s revenues and look at Intel. Clearly the market is telling you something about the pressing need for 64bit.
I think in another thread at OSnews I answered this question. Perofrmance is not what sells, it sells in the gaming market. RAS sells in the server market. When you add all the RAS features to you average x86 hardware the cost adds up. SPARC chips have ECC in every register inside the chip and every thing is protected. They made a mistake once but they learned from it an the USIII lines are ECC protected at every level.
Contrast that to the PC industry. I was at a PCI-SIG annual conference last year. There was a whole presentation by Intel on desgining PCI-express for the server market by building in RAS features into the SPEC. There were so many times the speaker talked about how it is today that any error just causes the machine to reboot and go on and how it was unacceptable for the server market and they are fixing this in PCI-Express.
So may be you should stop the FUD an understand Sun’s market. Sun has been a leader in Scalable, highly reliable systems. The x86 industry and Intel is just catching up. Intel is finding out the mentality of the enterprise customers with Itanium. Itanium’s adoption has been so-so to be nice to it. Enterprise customers don’t just flock to a new sarchitecture just because it performs well on a few benchmarks.
May be you should gain some realworld expereince before you start predicting company and market futures.
“If you didn’t read my previous post sun still has several SPARC chips in the pipeline.”
Yeah, until next week, when SUN cuts them off too.
Anyway, no point in arguing, SUN will be irrelevant then bought in a year or two.
Sun had about 5 billion in cash before the microsoft deal now they whould have about 7 billion in cash. They don’t fab thier own chips. How many billions would they need to do R&D if 7 billion isn’t enough?
They keep those billions to use them to pad out their quarterly looses.
In case you haven’t read their previous announcement, they are in fact CUTTING OUT their R&D budget.
Those 5 billions where gained during the dot-com bubble era that ended around 2001. More than likely most of it is already gone.
They keep those billions to use them to pad out their quarterly looses.
In case you haven’t read their previous announcement, they are in fact CUTTING OUT their R&D budget.
Those 5 billions where gained during the dot-com bubble era that ended around 2001. More than likely most of it is already gone.
Bullshit. If they were padding thier quarterly loses they wouldn’t post loses every quarter, would they?
http://finance.yahoo.com/q/bs?s=SUNW&annual
PERIOD ENDING Net Tangible Assets
30-Jun-03 $6,074,000 <—- it’s still there
30-Jun-02 $7,515,000
30-Jun-01 $8,274,000
Yes they are cutting out thier R&D budget, Analysts were claiming they were putting too much into it anyway. They are cutting 350 million/year from a 2 billion dollar budget. The processors they cut I think will be a big chunk of them.
“McGowan said in a conference call Friday morning that Sun expects to reduce both administration and research and development expenses by $500 million from about $5.2 billion in fiscal 2004, which ends June 30, to $4.7 billion in fiscal 2005.”
They are not killing R&D they are trimming R&D that won’t translate to revenue. A smart move by the way. The microsoft deal is a positive thing for Sun, even though I hate microsoft.
Sun can now market its desktop software stack and compete against Novell, Redhat and IBM with an edge, being able to interoperate seemlessly with 95% of the worlds desktops. Sun’s deal with MS makes it possible for them to license Office’s latest file formats and offer more compatibility with MS Office than say Openoffice. It’s compertitors either have to buy StarOffice from sun to bundle with thier stack or lose out. buying StarOffice increases Sun’s competitors cost for thier desktop software and either makes thier competitors software more expensive or makes their margins lower. MS gains a competitor and fends of Anti-Trust litigations. Sun and MS still compete for Office dominance but Sun has the potential to own the linux and non MS enterpriser desktop software market. This is one example I can think of where this deal is mutually benificial to Sun and MS. There are many many more.
Please think and research before you let you ignorance show.
Sun had about 5 billion in cash before the microsoft deal now they whould have about 7 billion in cash. They don’t fab thier own chips. How many billions would they need to do R&D if 7 billion isn’t enough?
That 7 billion doesn’t matter when they’re bleeding something like 900 million a year. Sun has to look at the future and decide if in the long run they’ll be able to compete with the likes of AMD, Intel, and IBM in the high-end chip market. I don’t see it happening.
I saw a post mentioning Alpha processors and HP releasing an EV8 version. I think that chip was basically killed and was included in what they sold to Intel. They released the EV7, which was was aan EV6 with an integrated memory controller (and on die L2 cache), much like how an Opteron is mostly an x86 with a memory controller. They were going to release an EV79, which would have had (I think) faster connections between processors and a faster memory controller along with a die shrink . . . but they cancelled it and are now just coming out with an “EV7z” or something with fewer improvements. Instead of just being an improved 21264, the EV8/21464 would have been very new. A great article, though old, can be found at http://www.realworldtech.com , from back before Intel called SMT hyperthreading and Alpha still had a future. Other interesting reading is a paper (I lost it when my hard drive crashed) about vector extensions to EV8 called Tarantula. It was given at a microprocessor forum even after the EV8 was cancelled. I’m sure it’s on the internet somewhere.
That 7 billion doesn’t matter when they’re bleeding something like 900 million a year. Sun has to look at the future and decide if in the long run they’ll be able to compete with the likes of AMD, Intel, and IBM in the high-end chip market. I don’t see it happening.
The last I checked AMD wasn’t in the high-end chip market, opteron scales only upto 8 cpus, Sun is an AMD partner now selling opteron boxes. Sun invested in a startup that makes opteron boxes. So no competition there.
Intel isn’t in too hot of a spot in the high-end arena, Itanium isn’t selling in any volumes. Opteron is eating at thier lowend and 64 bit market. Thier own x86-64 chips can canabalize itanium. So Intels highend future doesn’t look too good.
That leaves IBM. IBM is Suns major threat. Sun is trying to deal with that threat by moving SPARC to throughput computing and geting into that market early. IBM is also moving it’s high end Power chicps to multicore SMT soon. Sun is now doubling down on software and hardware and realign its expenses by cutting jobs and unnecessary R&D spending to return to profitability.
Sun infact has been late to respond to the market deteriorating, IBM, cisco and HP have layed of many more thousand employees than sun. sun till date has only let 11000 employees go while still hiring. Scott McNealy wanted to save jobs in anticipation of market recovery. His softness is porbably what brought Sun to continuous loss. McNealy’s rivals on the other hand ruthlessly cut 25000 + jobs and brought their companies back to the black. Business wise McNealy might have made a wrong decision but moraaly was it correct?
I wouldn’t write Sun of yet.
“SPARC, MIPS they don’t even hold a candle to the performance of the regular Intel P4 [Pentium IV] you can buy for 1/10 of the price. The SPARC in particular is so slow, it’s embarrassing. If I were Sun I would put a bag over my head.”, Linus Torvalds in a Q&A session on a Linux Geek Cruise in late 2002.
linus torvalds on Intels IA-64 same interview
“On IA-64…
I dislike IA-64. I think it’s a losing strategy. Intel is already eating itself where the P4 outperforms the IA-64 by factor 25-30% which is huge on SpceEngine. SpecFP doesn’t matter, and has never mattered. Intel has 98% of the server market, that’s because most of the volume is in really small servers. And now Intel noticed they couldn’t expand much more, so they’ve decided to do all the same mistakes in the RISC community, thus begat IA-64 My personal hope is that IA-64 just withers and dies because there is no point. It performs badly, it’s expensive and it’s an all new instruction set. “
So…. What’s your point? Linus is very opinionated (surprise, surprise) and likes the pentium 4.
alpha -> itanium
ultrasparc -> amd64 -> ultra’s from fujitsu
I doubt ultrasparcs will be finished, it’s just that fujitsu will
take over. r&d for chip fab and development costs an arm
and a leg, that’s what happened with the alpha processor,
just a damn good processor, but cost too much to support
I’m glad to see sun work on these things. if it’s one thing that
the world doesn’t need is too much choice for microprocessors.
We already have enough, powerpc, x86, x86-64, itanium, alpha,
ultrasparc, via, transmeta
Do you know what a pain in the ass it is to port from one
architecture to another?
I think sun’s moving in the right direction, software and services,
that’s always been their strength, the edge they’ll have over dell
and hp.
go sun!
I think Torvald’s point is that SPARC is overpriced crap, and he’s actually more than qualified to make that judgement.
I don’t understand the point of your quote. What was that supposed to prove? That opinion of IA64 is a widely held one in the industry. There’s a reason IA64 is usually referred to as ‘Itanic’. Why do you think Intel has finally introduced the IA32e?
I think Torvald’s point is that SPARC is overpriced crap, and he’s actually more than qualified to make that judgement.
I don’t think he is qualified to make that judgement. He is not a cpu architect. He can have his opinions. Look at transmeta the company he used to work for, it should have made a killing in the market place becuase “linus” workred there, it didn’t. It failed miserably.
Linus doesn’t even work on the linux SPARC port. What qualification does he have to judge CPU architecture, name one CPU he architected that is used in this world. I doubt Linus would himself claim to be an authority on CPU architecture. Your post had nothing to contribute to this discussion, that was the point of my post..
If you want to treat linus like some Computing God go ahead. Just because he doesn’t like SPARC, means any thing for the future of SPARC. Companies aren’t going to suddenly dump SPARC because Linus Torvalds thinks its crap.
I think a lot of these comments are poorly motiviated, lack understanding of many of the underlying issues, and aren’t very thoughtful or insightful.
Sun has repeatedly stated that they believe in R&D, are a systems company, intend to keep their SPARC processors viable, and will push for open standards.
That will give them time to focus on their strong points (hardware engineering and services) while they seize the opportunity to cut costs tremendously by sharing the cost developing software with the open source community. (Linux, {Java?}, {Solaris?})
I think this poster has their companies confused. IBM is focusing on its Hardware engineering, services, and middleware. Sun is focusing on putting together end-to-end solutions that don’t lock customers into a proprietary architecture, or system model. The fact that they have editions of Solaris available for x86, x86-64, and SPARC means that if customers are concerned about being stuck with a platform they don’t want, they can migrate to another one. The price of a Solaris license is much cheaper than one from RedHat, so it’s hard to argue that Linux is more affordable.
I don’t see how this poster expects Sun to cut costs by ditching Solaris in favor of Linux. The development of Solaris is years ahead of Linux, provides critical features an functions to Sun’s customers, and is the platform upon which their customer-base runs all of its applications. Switching to Linux would be suicide as all of their customers would desert in favor of whatever company came along with the ability to run older Solaris apps.
Further, I don’t understand how the Open Source Community can provide a Linux to Sun that meets their product goals. Sun will still have to spend engineering time and money modifying Linux to meet their customer’s needs. In fact, it makes much more sense for Sun to retain their Solaris engineering expertise. They already have a product which provides significant value to their customers. Another key point, is that features that are just being added to Linux have been in Solaris for years. I have friends who work at Sun and have pointed out numerous examples of mistakes the Linux kernel developers have made, that Sun also made when first attempting comparable features in Solaris. These have settled and been fixed for years, yet the problems exist in Linux. It would be a step backwards for Sun’s customers.
Please don’t construe the above as Linux bashing. It is not intended as such, rather a statement of where Linux’s enterprise-class feature-set is in relation to other commercial offerings.
maybe they should just let fujitsu handle the hardware while the focus on the software and then they could each create packages by mix and matching from the two companies?
Well, certainly if you’ve read the Register over the last couple of months it appears like this could be a possibility. However, I think this statement is pretty drastic. Assuming Sun is able to forge a strategic alliance with Fujitsu, I would imagine that Sun would continue to develop its Throughput-Computing processors (TPC) for future applications. In the short term, Fujitsu could potentially offer great value to Sun in replacing space just vacated in their product line, and also helping develop large ccNUMA machines. It looks like Sun might have similar plans to do something like this on the x86-64 side, since they acquired Bechtolsheim’s company, but I would imagine that their currently entrenched SunFire 4800/6800 and E10,13,15K customers would like to see future SPARC development in that vein of architecutre as well.
The point here, though, is that with Jonathan Schwartz, a software guy, appointed to the position of President and COO, Sun is changing its focus to make sure that its strategy draws appropriately from both hardware and software. The assertions by various posters that “Sun is basically a hardware company” is simply false, and is indicative of a larger misunderstanding of what Sun does and how they make their money.
sun can’t open source solaris due to license agreements with SCO.
I’m not a lawyer, but once again, people I know who work at Sun have indicated that their deal with SCO was a purchase of intellectual property, more than it was a license aggreement. Notice that Sun is shipping Linux via. JDS and SCO hasn’t made any attempt to sue them for violations of license agreements. Based upon what I know, Sun paid a lot of money in the early 90s to purchase unadulterated irrevokable rights to SysV, and they can basically do whatever they want. They don’t pay SCO regularly for the privelege, it was a one time deal, and best of all, they can do whatever they want with it.
The issue isn’t if Sparc/Solaris is DOA – the market has already determined it is…. Java, Linux, Opteron. Those are the three words (in that order) Sun should be focusing on.
Which market? What are you talking about? Sun continues to make sales of SPARC equipment and Solaris is going strong. You obviously haven’t been paying much attention to what the Sun intends to do with Solaris 10, or if you have, you have failed to understand what value it provides to existing and potential customers. Linux is not a panacea or a cure-all for any company. I find it encouraging that Sun is working Linux into its strategy and is using it to challenge Microsoft on the desktop, but it’s certainly not appropriate for every corporate problem the software world is facing. Suggesting that open-source is the answer to all of enterprise computing’s problems is a naive, if not a specious claim.
I got capped at 8,000 chars.
Here’s the rest.
We already know Sun is not “100% committed” to Sparc by virtue of this story. And as Intel has shown, 64bits is a nice goal but what matters is PERFORMANCE which drives SALES.
I don’t see how you could possibly make this claim. Sun has just released US-IV as an in-place upgrade for existing customers and has cut extraneous processor development projects so it can focus on areas which is expects to be of great importance to their customers. As this article indicated, a number of SPARC processor projects exist, and a number of other companies continue to manufacture SPARC processors. To suggest that Sun is not %100 to SPARC is unfounded and alarmist.
The second half of your claim is equally rediculous. If I had a chip that was twice as fast as anything else on the market but 100 times more expensive, I don’t think that I’d be selling many chips. Performance is not the only thing that drives processor sales. The reason x86 has experienced such great sales and adoption is that it offers very acceptable performance at incredibly cheap prices. Notice that Sun has amended their product line to include competitive Price/Performance chips from Intel and AMD. Sun’s TPC initiative is a different approach to the same idea, by offering incredibly parallel performance for many different concurrent workloads on one chip, they’re effectively planning on attacking this price/performance problem from a different direction down the road.
In fact, suggesting that Intel’s clock-speed based marketing is what drives their sales is pretty short sighted. Many of their server processors are clocked lower than their P4s, have bigger caches, and are more tuned to work in a server environment. That said, AMD sells chips that can beat the crap out of Intel’s, or at least be competitive. I’ve benchmarked 2.0gHz Opterons against 3.06 gHz Xenons and P4s, and found that Opterons can compete in just about every situation except those which are a measurement of raw-processor bound operations. Like Sun, AMD realizes that there’s a multifaceted approach to building processors, and the clock-speed isn’t everything. That’s not to suggest Intel doesn’t, they’ve hired some of the best and brightest people in the EE, ECE, and IE field. However, their marketing division still insists that they build processors with the highest clock rates, regardless of the actual improvement found.
Maybe you should need permission to publicly share your opinion. It is pretty hard to compete with Dell/Linux on price and Sun is not in a position to do so. The battle for the small/mid range stuff is already lost against the Xeon/Linux(and Windows!).
Heh. Perhaps you should.
However, be that as it may, Sun competes pretty well with Dell in the small-size sever business. Sun sells 1 & 2U systems at a comperable price to Dell. They may not make the same margins that Dell does, I don’t really know, but to suggest that Sun can’t compete there is silly. They are, and they do. I would argue that the battle is just beginning. Sun has been supplimenting their low-end Intel stuff with AMD’s Opteron which allows them to take a good position for attacking exsisting low and mid-range vendors.
I think Torvald’s point is that SPARC is overpriced crap, and he’s actually more than qualified to make that judgement.
I don’t think Torvalds is qualified to make such a statement at all. He is entitled to his opinion, but you should avoid getting sucked into his cult of personality and ought not to take everything he says as gospel. As far as I can tell, Linus has never designed a microprocessor, or if he has, it hasn’t made it to market in the civilized world. Linus is a geek icon, but doesn’t hold much status in the business world. Most companies aren’t going to go out and make business decisions based upon Torvalds’s opinion. It makes much more sense, especially in technical matters, to go research things for yourself and develop balanced opinions based upon facts and comparisons. Holding a strong opinion based upon ignorance is an easy way to develop bigotry, prejudice, misunderstanding, and misinformation. This also applies in the non-technical world.
>> Really what pray tell are Sun’s current customers to do with there billions invested in SPARC Solaris?
They can get stuffed along with all of the buyers who put money into IRIX, DEC/Tru64 etc etc etc. There are no assurances in life – buyers are free to dump their entire IT budget into a dead architecture if they like.
The rest of your post is offtopic blathering about conferences and ECC and everything else that doesn’t matter.
>> Companies aren’t going to suddenly dump SPARC because Linus Torvalds thinks its crap.
No, they are dumping it because it is a dead-end architecture with poor performance promoted by a company that is evaporating.
… most frequent word in this thread?
seems like everybody love to be a fortune teller (ha)
However, be that as it may, Sun competes pretty well with Dell in the small-size sever business. Sun sells 1 & 2U systems at a comperable price to Dell. They may not make the same margins that Dell does, I don’t really know, but to suggest that Sun can’t compete there is silly. They are, and they do. I would argue that the battle is just beginning. Sun has been supplimenting their low-end Intel stuff with AMD’s Opteron which allows them to take a good position for attacking exsisting low and mid-range vendors.
Considering that their x86 servers are not assembled by them, and the motherboards and other associated components from third party vendors, the over head is very low. Where SUN make the money is via support contracts, software subscriptions, the “frills” that surround their products.
The fact remains, *ENTERPRISE* customers are *WILLING* and *ABLE* to pay the premium for SUNs hardware and software, heck, just recently SUN made a multimillion sale to Air New Zealand which will run IBM Websphere, once an exclusively IBM company. Telstra has moved closer to SUN with the deployment of JDS and new Solaris servers.
There are *NEW* customers coming on board for SUN all the time, the fact is, the only thing SUN isn’t doing well is promoting every new customer they get vs. IBM which boast about any customer which comes within a 100kilometre radius of the company, even if all they want is directions to the local pub; they’ll promote that as a sale of a service.
Microsoft on the otherhand will approach companies and offer them a discounted installation of their software then claim that their “software solution” won over the customer, when anyone who is as cynical as me would realise that it is a crock of crap.
HP is just as bad, for all the screaming we hear from the cheap seats in the Linux crowd, if it weren’t for the printer business and their miniscule services side, HP would be loosing money like there was no tomorrow.
As for SGI, it bleeds red every quarter, the only thing that makes the share price run high is the mistique that goes with the name Silicon Graphics and the fact they hype Linux.
About the only company making a decent amount of money off their core business is Dell, which is a 0 R&D company. They realise that R&D costs money, risky and possibly might not return profits immediately, and hence, decide to take the low cost route by pushing all the R&D off to their suppliers and simply assemble computers that kinda work.
They can get stuffed along with all of the buyers who put money into IRIX, DEC/Tru64 etc etc etc. There are no assurances in life – buyers are free to dump their entire IT budget into a dead architecture if they like.
The rest of your post is offtopic blathering about conferences and ECC and everything else that doesn’t matter.
What is dead about SPARC? it seems to be still selling. It will keep selling no matter how much inexperienced drivel you spew. ECC matters along with other RAS features in an enterprise enviroment, Sun’s mainstay. The fact that you can not come up with an intelligent rebuttal to what I said shows your ignorance.
Customers are not stupid. HP still sells tru64 and VMS machines becuase customers invested in it. But that is alittle too much for you to comprehend, so much so that you have to resort to name calling, nothing about my post was offtopic. I was trying to explain to you Sun’s market and how your drivel is so filled with ignorance about the real world.
I don’t think he is qualified to make that judgement. He is not a cpu architect.
That’s kinda like saying that anyone who has some clue wrt computers, but aren’t actually chip engineers, aren’t fit to judge whether the 6502 is suitable for modern computers or not.
Your post had nothing to contribute to this discussion, that was the point of my post..
Haha, nice try. Fact of the matter is you thought Linus was talking about AMD64 in that quote, not Itanium.
That’s kinda like saying that anyone who has some clue wrt computers, but aren’t actually chip engineers, aren’t fit to judge whether the 6502 is suitable for modern computers or not.
That is not what you posted. You said linus thinks the SPARC is crap so it must be. I questioned the “if linus says so” attitude. Linus says a hundred things about a thousand things, he has the right to have any opinion he wants. But you posted his quote as if it was a gospel and he was the ultimate say in what matters in computing. He is not. He is a good technical lead and his greatest contributon to computing was Open Sourcing linux, He hasn’t contributed any new technology to the computig world. He made a UNIX
clone open source. I should say he is a good leader. But most of where linux is today is becuase of the contributions of other really good developers. Linus was just instrumental in making linux open source and it gaining momentum. He is a good developer and a damn fine leader, without whom linux wouldn’t be where it is today. But he is no authority on CPU architecture.
Your post had nothing to contribute to this discussion, that was the point of my post..
Haha, nice try. Fact of the matter is you thought Linus was talking about AMD64 in that quote, not Itanium.
What lead you to that conclusion?
“inus torvalds on Intels IA-64 same interview” That’s the first line in my post.
I will say it again your post contributed nothing more to this discussion. In fact I would go so far as to call it flamebait.
“Java is dying”
I doubt that. Java will be the only remaining living technology if Sun dies.