SPARC

Actually this is one of the few open source chip architecture available. SPARC International was created in 1989 as an independent, non-profit organization to oversee and guide the SPARC evolution.

Several companies like Fujitsu produce Sparc compatible chips for many years. Actually Fujitsu one core chips are faster then Sun's own and are equal to Opteron in most computational benchmarks like SpecInt.

They pushed forward the SPARC performance envelope and keep pace with competing architectures.

Fujitsu 5th generation 1.89 GHz SPARC64 processor shipped in September 2004 was the leader in Sparc market until T1 came.

Old News ;-)

SANTA CLARA, Calif. - October 2, 2006 - Sun Microsystems Inc. today announced the creation of an independent OpenSPARC Community Advisory Board.

The board is chartered with setting the direction for OpenSPARC, a community that fosters the creation of tools and derivative chip designs based on Sun's UltraSPARC T1 processor. The community now includes a new GNU/Linux distribution, Gentoo Linux, which is supporting UltraSPARC T1 in the latest release, and the first published derivative of the chip design from Simply RISC.

"To grow we can't just rely on upgrading our installed base, we have to go after new customers, and that's what open source helps us do," said John Fowler, executive vice president of Sun's systems group.

Editor's comments:- it sounds like such a good idea you may ask why hasn't Sun done it before?

The answer is Sun did. But most of the oems which picked up the SPARC technology ball in the first decade of SPARC got eased out of the game.

The first commercially available SPARC servers were shipped in 1987.

In 1989 Sun, with other partners set up SPARC International as the industry body to disseminate technical information about SPARC, and facilitate the licensing of technology and trademarks.

Fujits...Who - A Primer on Fujitsu's SPARC Heritage - article

Two years ago, in July 2002, I wrote an article called Sun Users Waiting for Fujitsu? in which I said that Fujitsu might be the only hope for Sun customers who were suffering from Sun's inability or unwillingness to push forward the SPARC performance envelope and keep pace with competing architectures. Little did I imagine that it would actually take so long for this to happen.

In October 1993 in my article Are Sun's Days Numbered? I ran a report that Sun and Fujitsu were talking about collaborating on future SPARC chip developments and merging product lines. We had to wait till June 2004 for confirmation of this in a joint Sun and Fujitsu press release.

Also in June 2004 came the announcement from Fujitsu that it would be shipping servers using its 5th generation 1.89 GHz SPARC64 processor in September. This is a SPARC processor which is 60% faster than any single SPARC processor which you can buy today from Sun, and it's the start of the roadmap for all forseeable future high performance SPARC servers from Sun.

Here's the important part.

If you have investments in SPARC servers then, looking ahead, it's going to be Fujitsu that delivers future enhancements in SPARC chip technology and not Sun. There's no one else left in this market who will come in as a white knight and save the SPARC market from being a dinosaur. You have to feel comfortable with that or start making plans to switch to another platform.

Most of you won't be familiar with Fujitsu's track record in the SPARC market, and that's why I put together this article to pull together some snippets of SPARC History from our archives.

Some Past SPARC Milestones by Fujitsu

Fujitsu's SPARC products have left a weak impression in the minds of most Sun users. That's partly because the company's past efforts have been fragmented, disjointed and sometimes unsuccessful in the market.

Within the SPARC systems segment the company has created and then killed or stealth marketed brands more often that Buffy the vampire slayer has saved the world from armaggedon. Fujitsu's SPARC systems companies and brands have included:- HAL Computer, turboSPARC, ICL, Amdahl, Fujitsu Siemens (in Europe), Fujitsu Technology Solutions (in the US) etc.

1993 - Fujitsu's European subsidiary ICL previewed its GoldRush Megaserver, a 64 SPARC CPU capable server, originally rated at 6,000 transactions/second. It ran Unix, but not Solaris. ICL later merged into Fujitsu Siemens Computers.

1995 - Fujitsu owned HAL Computer launched the industry's first workstations based on a 64 bit CPU, HAL's independently designed first generation SPARC64. HAL's workstations did run Solaris, but failed to make a dent in the Sun compatible market. Within a few years, Fujitsu closed down HAL and merged its server and chip technology into the short lived Fujitsu Technology Solutions, which then became Fujitsu Computer Systems.

1997 - Fujitsu Microelectronics's 32 bit TurboSPARC was a user installable upgrade chip for Sun's SPARCstation 5

2001 - Fujitsu's independently designed SPARC servers didn't suffer from the cache design problems in Sun's own systems (see the article Unsafe At Any Speed?). That was first sign that Fujitsu's combination of semiconductor and computer experience could deliver a vastly more reliable SPARC server than Sun's inexperienced designers who made a fundamentally bad design decision.

2002 - Fujitsu's SPARC servers started making headlines for their performance, as in this December 10, 2002 news story... "Fujitsu Technology Solutions Inc. today announced best-in-class results in industry-leading benchmarks on its PRIMEPOWER 850 SPARC compliant, Solaris compatible servers."

2003 - In October the pieces of the jigsaw started to come together. We reported... a news story today in JapanToday.com speculates that Sun Microsystems may reduce development costs by pooling resources with Fujitsu on the design of future high end SPARC servers.

In Retrospect

I'm reminded of the 25 year old battle between Intel and AMD in the x86 compatible chip market. In 1980 AMD was persuaded by Intel to drop its support for the competing 16 bit Z8000 processor which AMD was making as a second source to Zilog, and instead to support Intel's 8086. For over 20 years AMD struggled to make money from its x86 processors, and then Intel made a terrific blunder by launching a 64 bit CPU which was not x86 compatible. Now AMD owns the 64 bit x86 market and is on a roll with that product.

In 2001 Sun lost its repuation for making reliable servers because of its badly designed cache memory which was sensitive to alpha particle radiation and which didn't include adequate error checking and correction. In the period 2001 to 2004 - Sun lost the performance lead of SPARC over Intel by late to market and badly executed semiconductor design. Fujitsu, a world leading chip company, now has products which can put SPARC back onto a competive track. Sun has tacitly admited that it will take a back seat in chip design and stick to tweaking the Solaris OS and marketing. Both jobs which Sun does well.

Past setbacks with some SPARC products haven't stopped Fujitsu's support for this technology. They have kept plodding on when dozens of other server companies have exited the SPARC market. Without that long term investment and consistency - SPARC wouldn't have much of a future today. ...Fujitsu profile, ...Sun profile

History

Surviving the Solaris x86 Wars 2004 - 2006 - article in the SPARC Product Directory

November 16, 2004 article by Zsolt Kerekes editor / publisher of the SPARC Product Directory - updated Jan 25, 2006

They say that elephants never forget. The wise old elephants of the herd help the survival of their species by remembering important stuff like where to find water in the rare years of drought. The young ones who are keen to find their own way and play with new toys are rude about the older generation thinking to themselves - they're just like mammoths. But now is not the time to challenge their wisdom. Not yet.

It's the same with the IT world. The old managers remember when they were screwed by supplier X a long time ago. So the young bucks have to wait till the old guys have moved on. The young IT guys say "What can I do with such dinosaurs running the place? My senior management won't let me try anything new." But their time will come - when they can experiment with new toys and make their own mistakes.

This is the first in a series of articles which will look at the long term strategic business and market implications of Sun's Solaris x86 platform. Yes - it has a long term. There was a time not that long ago when you couldn't say that. But there will be winners and losers in the community of users, IHVs, ISVs, and Sun VARs. And some uncomfortable truths will be revealed in later articles when we come to discuss the subject of losers. For now, this first article is just a reminder that not every platform in the Intel Architecture market has been a long term runaway success. By reminding younger readers about past failures I'm setting the scene for a cautious reappraisal of the future for the Solaris x86 server market.

As an old elephant myself, I've had cause to re-evaluate Solaris x86 four times in the past 18 years.

This is a subject I have viewed from many angles and in many market conditions. So I just want you to know that I'm not just a staff writer who was reviewing the latest PC game last week and has been asked to write a few hundred words about Solaris 10. There are plenty of those around. And I'm not caught up by Sun's hype machine either. I've been there a few times in the past - swept along by enthusiasm for an open systems standard called SPARC - in the early 1990s - which spawned this directory. SPARC became a whirlwind for a while which peaked in supplier numbers in the mid 1990s and is now heading towards a different kind of market. Actually I believe the prospects for SPARC over the next few years are pretty good - and I was the first editor/analyst publicly saying that Sun's revenue would get back to growth this year. But this article is not about SPARC. It's about a distant relation. So let's get back to that.

If we look at the Intel Architecture market today (and that includes AMD's Opteron) we see only successes. So that may lead some of us to believe that there's a Midas touch operating here. Surely the market for x86 deskstops and servers is so big - you might say - that there's room in it for a lot of success. After all, if you only get 1% share of the $250 billion PC market - that's a revenue you might be happy with for starting a new platform business plan. Let me warn you - that way disaster lies. That's the kind of reasoning which naiive entrepeneurs put in their business plans and some naiive venture capitalists used to swallow. But computer users are like gazelles - they can move fast sure - but they live longer if they go with the herd. They've noticed that those adventurous types who used to wander off to explore interesting looking paths on their own mostly got eaten by lions. So let's get back to the meat of this article.

CP/M - from Digital Research, was the market leading operating system in the Intel Architecture desktop PC market - back in the late 1970s. In those days - a lot of CP/M machines were actually running on Zilog Z80 processors. These were like an enhanced Intel 8080 - with more instructions and a single 5 volt supply rail. It was a 2nd generation 8 bit microprocessor world. The overall market leading PC at the time was the Apple 2. But unfortunately Apple standardised on the 6502 processor from Rockwell - which was a technical dead-end. So it wasn't easy for Apple to do a follow up machine.

Back to the Intel world... so successful was CP/M that when IBM's PC designers went shopping around for an OS to port to their 8088 based machine (which was an 8086 16 bit instruction set compatible processor with a lower cost 8 bit external data bus BTW) - IBM went first to talk to Digital Research. What happened next is part of the folklore and myths and legends of the PC world. Some facts are disputed. What's not in dispute is that a little company called Microsoft got the deal for supplying the new operating system. Although Digital Research later came out with a DOS clone - called DR-DOS - they were out of the game.

PC junior - following the initial success of the original IBM PC in the early 1980s - IBM saw a potential market in the home. Instead of lowering the price of the PC - IBM's marketers came up with the "brilliant" idea of making a low cost compatible consumer version. Before the launch of the PC junior - Fortune magazine ran a major feature listing all the companies whose fortunes would be made by the new product. But it was a dud. The keyboard was unusable, and the base PC had been deliberately designed to make upgrading to a business class PC difficult. Although this is an example of a much hyped but failed hardware platform - it containes the useful warning that analysts can be entirely wrong about the success of new computer product lines - even when they come from a well known source and it seems that nothing can go wrong. In a consistent move which in retrospect looked a lot like corporate suicide - IBM later compounded their hardware strategy failures by coming up with the PS/2 - which in its purest form had a bus called MCA - which was incompatible with the PC-AT.

Xenix - there was a lot of excitement in the market in the mid 1980s when 16 bit processors got enough disk space and memory to start running Unix. Systems based on Motorola's 68020 32 bit processor got a head start in the market because they were binary compatible with the 16 bit 68000 - which already had a bigger memory addressing range than the Intel chips at the time. And the 68020 also had hardware support for virtual memory management. Sun Microsystems' own Sun-3 workstations originally used Motorola chips. But most IHVs had big investments in Intel. So where could they turn to source a multi-user operating system? The surprising answer was Microsoft. I joined an ISV and VAR called Databix in 1986 which was using Zilog boxes running Xenix as its main Unix development platform. That was a problem because the Zilog Z8000 was a hardware dead end. (Within a year I had switched our Unix platform to Sun. But not the Sun platform you might think - see later.) Microsoft stopped developing and supporting Xenix which later got transferred to SCO.

OS/2 - "A better Windows than Windows." So ran IBM's hype machine as it signed up ISVs to support its own windowing environment for the 80286 processor generation of PCs - which had more memory than the original 1MB limit of the original 8086. Whole books have been written about this subject too. I had a bunch of good software people evaluating early version of OS/2 in the late 1980s. It was unstable, slow and misconceived. It is possible that if IBM had fixed on the 386 processor as the entry level OS/2 machine - it might have got there faster. But the compromises involved in making it compatible with Microsoft's stuff while at the same time retrofitting it to the low cost low performance (by then) 286 were too much. Although IBM marketers tried to convince analysts for years that it had an installed base of millions of users - most of those OS/2 machines were dual boot PCs - in which the OS/2 just never got booted. ...Later in 2004 - is Sun inflating its own Solaris x86 market numbers with its own dual boot servers? That's something we'll return to later.

SunOS (which became Solaris) - was another Intel Architecture dud in the 1980s. But it didn't look that way at first. If, like me, you were looking for a well integrated Unix platform on Intel Architecture in 1986 you could do a lot worse than choose Sun's 386i. It ran SunOS and some DOS applications in a compatible window, and it had slots for PC-AT disks. There the similarity to a PC ended. Instead of low resolution PC graphics it had high resolution Sun graphics. Instead of low performance disk drives it came with high performance SCSI disks - and a lot more RAM than the PCs of that vintage. Although the 386i was rubbish as a graphics workstation (the graphics was too slow) and despite the fact that the DOS emulation was too slow and useless to run most DOS applications - the 386i was a low cost entry level database server which could outperform Sun-3's and low end VAXes for a fraction of the cost. When I first signed up as a Sun VAR in the 1980s - it was the 386i which I used as the trojan horse to replace our aged Zilog boxes and eventually displace the HP and DEC minis. We were using Motorola's VMEbus SBCs as our main real-time platform - but needed a Unix with a future to coexist with our multiprocessor applications. Although the first SPARC systems had come out - the original Sun-4's didn't inspire any more confidence in me than the myriad of other RISC based Unix servers which were hitting the market at the time. Sun was supporting 3 hardware platforms with its SunOS at the time:- Motorola, SPARC and Intel. My guess was that Sun's long term strategy would be to run with Intel. I was right - but there was to be an 18 year discontiuity before that became a firm reality.

We kept asking our Sun VAR guy - when was the next model coming? There was talk of a 486i. That's all it turned out to be. Talk. Sun's SPARCstation re-engineered the SPARC processor into a low cost high volume manufacturable product with a new bus called SBus. Sun end of lifed the Sun-3 and 386i products - and my engineers made a painless transition to the 6U VMEbus plus SBus compatible SPARCengine 1. That was a good move for Sun, and a good career move for me as in 1991 I started my own company to publish the SBus Product Directory. (Which transitioned into this SPARC Product Directory.) This was the first time that Sun end of lifed Solaris x86. There would be other times and missed opportunities in the future.

I kept an eye out for Sun in the Intel space for several years after that. Sun did look like it might resurface in Intel space when it acquired a company called Interactive Unix. Unfortunately when Sun did later launch a shrink wrapped OS based on the IX acquisition - the product was completely unrelated to SunOS. Sun didn't understand the needs of PC makers and their hardware drivers were out of touch with the reality of hardware bing shipped. The lack of compatibility with SunOS meant that the platfform was a standalone with little 3rd party support. As a result Sun was unsuccessful at getting market momentum. So this strand of Sun on Intel story came to a marketing dead end too. That was the second time that end of lifed Solaris x86.

During the late 1990s - the runaway sucess of the SPARC platform meant that few people seriously questioned whether Sun should have another go at commercialising a version of Solaris on Intel. I wrote an article in 1999 Should Sun Microsystems make its own brand of "Intel Inside" PC's? - because I could sense problems coming to Sun's business. My self defense mechanism as a publisher had been to launch STORAGEsearch.com in 1998 and reduce my dependence on the Sun market - just in case Sun crashed and burned. Sun's revenue growth did hit a brick wall in April 2001 - and it took 3 years to recover as a company nearly half the size.

Sun launched another version of Solaris x86 during the late 1990s - as a non commercial product to lure universities onto Solaris and thereby to their SPARC platform. It was never promoted aggressively and it was positioned as a hobbyist system - despite its great technical potential. During the period 2001 to 2003 - Sun decided a third time make extinct this latest Solaris x86 evolution too as Sun wrestled with the demons of an its unsuccessful Intel Linux business based on the acquisition of Cobalt Networks. Sun started to change its mind about Solaris x86 when user pressure force it to accept the idea that this product so unloved by Sun's marketers, but with a base of over 1.1 million licenses in Q1 2003, could possibly save the company by fragmenting the Linux market and creating a reason for new customers to buy Sun servers.

But future of Solaris x86 still looked uncertain as late as Q2 2004 - because users had learned by now to distrust Sun and the apparent ease with which it could terminate its proprietary products. It was only the promise by Sun in Q4 2004 that it would make the Solaris platform open source - that marked the end of the phoney war. The real marketing war for establishing Sun's Solaris x86 business, its true aims and eventual targets will play out in the next few years.

Copyright © 1996-2008 by Dr. Nikolai Bezroukov. www.softpanorama.org was created as a service to the UN Sustainable Development Networking Programme (SDNP) in the author free time. Submit comments This document is an industrial compilation designed and created exclusively for educational use and is placed under the copyright of the Open Content License(OPL). Original materials copyright belong to respective owners. Quotes are made for educational purposes only in compliance with the fair use doctrine.

Standard disclaimer: The statements, views and opinions presented on this web page are those of the author and are not endorsed by, nor do they necessarily reflect, the opinions of the author present and former employers, SDNP or any other organization the author may be associated with. We do not warrant the correctness of the information provided or its fitness for any purpose.

Last modified: February 28, 2008

SPARC

Old News ;-)

Recommended Links

History