Now is about the time Sun Microsystems was meant to start telling the world of the wonders inherent in the Rock processor. No such luck. Instead, Sun and Fujitsu are celebrating the arrival of Version VII of the SPARC64 processor. The four-core SPARC64 chip will ship inside Sun and Fujitsu's midrange and high-end Unix servers, …
"No one else can do this ... "
Well IBM seem to have made a pretty good job if it.
The Power Series has kept pace with or ecxeeded Moores law every year.
Boxes featuring the promised chips arrive on schedule.
And they are 100% binary compatable with the venerable
power "chip set" boxes from the early 90s.
Sun shops are begining to look like VAX shops.
Nice hardware in its day, not the best value for money now,
but it's easier to upgrade the current systems than port them to
another platform. However new systems are hosted on a more
And what about Rock?
How come Sun has gone so quiet about Rock? I wish they'd give an update on how the transactional memory and scout threads are coming along. If they can't deliver competitive silicon on time, they should shut down the Microelecronics division and get in bed with Intel.
Sun not talking to you?
Surprising! At least the old marketeering droids used to welcome a challenge, if they caught someone passing on unwanted info they used to go on a campaign to bring them back into the fold. Looks like the latest marketeering droids just want to bury their heads in the sand and sulk.
Nice to see Fujitsu still have what it takes, I always did like the old SPARC64s, but not sure if it is enough to save Sun. Meanwhile, Fujitsu are still working on Tukzilla systems, so they obviously aren't betting the bizz on SPARCVII....
Would hate to dissapoint!
As does Paris !!!
To any Sun staff reading the comments
Please don't get your nose out of joint over Ashlie's article, it was nothing more than a discussion piece and was treated as such by anyone who made comments. What was also obvious in the comments was the high regard in which the readership of The Register holds Sun.
The Register is one of the most respected and unbiased technical journals on the net and I for one would be disappointed if up to date news from Sun was unavailable.
1/ The outside option - Sun can Rock at the last minute, buy Transitive with their last savings, release Solaris on Power/Itanium as a kludge of Solaris SPARC code on-top of Transitive on top of a copy of Red Hat Linux for Itanium/Power, and try to be the next Microsoft without the drag of hardware engineering.
2/ The better option - Sun start a back-up plan, ports Solaris to Itanium (job already half done) and opens secret discussions with FSC to badge their Tukzilla servers in readiness for when they can Rock.
3/ The flying pig option - Schwartz actually sees sense today, cans Rock, and sells a merger to FSC, with Solaris struggling on via SPARC64 and Transitive on FSC Itanium servers.
4/ The more likely Sun MO - Sun carries on, struggles to release Rock late and with a featureset and performance already ecipsed by Power, Itanium, Xeon and Opteron, and Sun dies a slow and painful death, with many pink slips going to the real workers whilst the management desperately hold on as long as they can.
A Great Upgrade to Sun's M-Series
People love to complain on blogs, and I must admit I have my own bias, but early response from real customers who have already purchased and installed the latest quad core M-series servers has been great. Check out the benchmarks on Sun's web page including SAP and Linpack (HPC) and see for yourself what the excitement is about.
Re: A Great Upgrade to Sun's M-Series
Well, well. I'm glad to see a Sun executive taking time to comment on our story even if they're refusing to speak with us in an official capacity.
Solaris on Itanium?
Rock has more chance of seeing the light of day than Poulson. By the time Sun can port Solaris to Itanium, HP will be scrambling to stuff Westmere-EX Xeon's into the QuickPath sockets in its Tukwila Superdomes so they can have a high-end server to sell. So if Sun just keeps Solaris humming on Intel x86, it will run on every Itanium server out there (once the servers are converted to run Xeon).
No more droids and Blue Koolaid.
Hmm - as far as memory serves me most of the Sun Marketing Droids got shot last week - there is nobody left to hold the conference and until they employ new droids and indoctrinate them there would no other conferences.
@John Anderson - really? IBM can do all of that??? Their boxes ship on time? They are binary compatible? They still give out free blue-Koolaid? You drank some of it?
... unless you were sarcastic?
RE: Solaris on Itanium?
You are obviously unaware of Itanium's history - Sun was originally onbaord, then decided to go it alone with SPARC. At that point, they already had a Slowaris porting effort underway on Itanium in the labs, so this would provide a sound base for a quick escape to Itanium.
And I hate to burst your bubble (OK, actually I don't), but HP's Itanium range is the only enterprise server range that is growing in market share and revenue, the rest are stagnating or diminishing. Simply check out the Reg articles on quarter results. Add to this Sun's woeful record of bringing chips out late or killing them (remember UltraSPARC V?) and I'd say there's much more of a chance of Poulson making it.
PS: Ash - don't be too hard on Mr Hamilton, it must have taken a lot of guts to not post as AC, more than your average exec. Mind you, he must have been busy shooting the old droids - only two sentences of Sunshiner prose!! Maybe he was saving the best for his blogs. Or polishing his CV...
Ashlee-Do you have an axe to grind with Sun?
Its interesting to see how you give praise and glory to IBM and its Power6 "monster" I believe you called it, but don't give any justice to the Sun SPARC product line and especially the M-Series. Any biases here?
Sure, the M-Series is not ROCK based, and yes, the world will have to wait till next year to see ROCK, but the M-Series appears to be the current leaders in server performance across a wide range of benchmarks, even beating IBM's Power6 Monster! And when will IBM or Itanium release quad core??
Why haven't you talked about these new benchmarks? Why haven't you questioned IBM on why it took them a year to rollout the Power6 product line? Why has IBM slipped Power6 by 2 years when you look at their 2004 roadmaps showing 2006?
Solaris has maintained #1 Unix marketshare for over a decade. Solaris is now running on the widest range of platforms on the planet. Can't say that for AIX or HP-UX.
Re: "No one else can do this ... "
I think what they mean is that no one else can do binary compatibility across this scale of servers (from volume to enterprise). IBM certainly can't, HP definitely can't (and who would want to with HPUX or Windows anyway?). The point is binary compatibility. No one else does it as well as Sun.
RE: Solaris on Itanium?
Wait for a response from Matt Bryant. He's a true Sun hater. He rips on Sun every chance he gets even if the article isn't about Sun.
Matt said "but HP's Itanium range is the only enterprise server range that is growing in market share and revenue, the rest are stagnating or diminishing"
If you sell 1 egg one year and 2 eggs the next, that's 100% growth. Your shipments have grown much faster than your competitors that are selling 500,000 the first year and dropped to 450,000 the next.
http://www.itjungle.com/bns/bns022108-story01.html - review by gartner of major server vendors of 2007.
"RISC-Itanium Unix servers fell on a global basis for 2007 at 13.8 percent in shipments but grew 1.7 percent in revenue for the year," explained Jeffrey Hewitt, a research vice president at Gartner who does the server counting, said in a statement accompanying the numbers." "By Gartner's count, seven of the top 10 server makers in 2007 had shipment increases, but IBM, NEC, and Sun Microsystems did not. Hewlett-Packard is still the king of server shipments worldwide, thanks to its acquisition of Compaq back in 2001, and managed to push out 2.64 million boxes, up 16.8 percent and giving it a 29.8 percent share of the global 8.84 million shipments in 2007. Gartner reckons that ProLiant branded shipments rose by 17.2 percent, Integrity shipments rose by 56.8 percent, and NonStop shipments climbed by 13.8 percent"
As far as Solaris being slow http://www.datacenterknowledge.com/archives/2008/Jul/03/1_million_messages_a_second_for_trading_system.html.
That beat out Linux on an HP system of very comparable configurations. HP systems couldn't even come close. So Solaris not only performed but out-scaled the competition in a standard topology in an apples-to-apples comparison.
To add to that, Sun just isn't about SPARC anymore: "Sun Fire X4450: Challenging the HP and IBM bastion" http://www.anandtech.com/IT/showdoc.aspx?i=3316
Anyone who thinks SPARC is expensive, hasn't been keeping up on price. You can get a SPARC server for about $2800: http://www.sun.com/servers/coolthreads/t1000/index.xml.
When you compare it to other RISC or Itanium based servers, the price/performance on SPARC is pretty good and very comparable.
Just check out the following to see how Itanium and SPARC compare:
There's a reason why the undergraduate and graduate school Computer Architecture bible moved the Itanium processor discussion from the main book to the online Appendix in favor of the UltraSPARC T2 processors:
Computer Architecture: A Quantitative Approach By John L. Hennessy, David A. Patterson, David Goldberg, Krste Asanovic
Universities in China are adopting SPARC as their primary architecture to design and develop: http://news.softpedia.com/news/China-Universities-Join-Sun-039-s-SPARC-Partnership-Program-79587.shtml
There's also a reason why SPARC has been so successful since SPARC International was formed. It was never closed or proprietary: http://sparc.org/.
"SPARC International fosters innovation of SPARC by offering testing and branding programs, and, by promoting and protecting SPARC and SPARC-related brand names. The organization maintains this openly and cooperatively defined technology by using its membership fees to ensure that SPARC maintains continuity with the industry standards of binary compatibility. It is this organizations sole responsibility to clarify these definitions which are made available for free download from our web site. You do not have to be a member to download. However, becoming a member provides your company the choice to test and brand products based on the SPARC architecture."
Sun's opensourcing of its UltraSPARC T1 and T2 processors (http://www.opensparc.net/) has led to Universities in China adopting it as the main processor of focus and how there are already two deritivative designs S1 (http://www.linuxdevices.com/news/NS4310115400.html), and
People are starting to deploy SPARC beyond just traditional servers: "Wind River to port Carrier Grade Linux to Sun's UltraSPARC T2 CMT processor" http://findarticles.com/p/articles/mi_m0ECZ/is_2008_April_17/ai_n25343268
and also beyond just Sun's own chassis: "Themis Partners with Sun to Make Sparc T2 Blade Server": http://www.itjungle.com/tug/tug060508-story04.html
At the end of it all, Matt will probably call me a Sun droid or a Sunshiner or something along those lines. Perhaps. Maybe. It doesn't matter. He's pretty close "behind" Hurd from what I've been seeing reading register articles and comments. Don't care what he will reply with. I don't intend to respond. I like Sun products and I think they're pretty good. Compared to the 330,000 man IBM or the 250000+ man HP, the 30,000 man Sun does a pretty damn good job. And that is without a real marketing organization. Anil Gadre sucks and should be fired and a real marketing organization should take over. I just listed a ton of articles and unless you've been talking to someone from Sun, you probably didn't know all that happened. How much does Sun marketing suck? How much does Anil Gadre suck? How much time do you have?
And amid how bad all the vendors have been struggling in the present U.S. economy, Sun hasn't done great. But for the most part did all right: http://biz.yahoo.com/rb/080715/sunmicro_forecast.html?.v=4
The reality of Itanium has been quoted time and time again:
http://www.channelregister.co.uk/2006/02/24/clabby_cured_again/ - Negative view point of IDC on their consistently rosy depiction of Itanium
http://www.theregister.co.uk/2004/09/18/ibm_plots_idc_mistakes/ - IBM plots out why IDC has been wrong about their Itanium prediction.
http://www.channelregister.co.uk/2008/01/10/ibm_itanium_five_years/ - Arcticle in which IBM states that Itanium has 5 years to live.
http://news.softpedia.com/news/Dell-Says-Bye-Bye-Itanuim-8651.shtml - Article about Dell discontinuing Itanium servers.
Analysts generally view Sun positively
If you look at Gartner's recent (Jun 27, 08) review of Sun, they rated Sun very positively and that includes their SPARC products:
IDC said the same thing about Solaris gaining momentum and being the third most strategic operating system. The order was Win2K3. Z-OS, Solaris, Linux, AIX, HPUX, SuSE, OpenSolaris and finally others. Linux and AIX actually lost traction, while Solaris gained traction:
- Opportunity selfie: Martian winds have given the spunky ol' rover a spring cleaning
- Spanish village called 'Kill the Jews' mulls rebranding exercise
- NASA finds first Earth-sized planet in a habitable zone around star
- New Facebook phone app allows you to stalk your mates
- Battle of the Linux clouds! Linode DOUBLES RAM to take on Digital Ocean