Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?
Sun Microsystems

Sun's UltraSPARC III Processor Shipping 231

jzuska writes: "Sun Announced it's UltraSPARC III today. The link is here. 29 million transistors and 900MHZ, 9.6 Gigabyte/second processing bandwidth woohoo!!!" And this is an announcement of "volume shipments," not *coughcough* EventualWare. So lessee ... for boxes that you might reasonable put on your desk right now, there are how many architecture choices? (And how many of them run Windows?)

On a semi-related note, Sulka writes: "Motorola just introduced a new G4, the MPC7410. While this is a processor meant for embedded systems, it's essentially the same CPU as the G4 in Macs. What's significant is the price -- the 500 MHz version carries a $195 price tag. This is much cheaper than the Intel and AMD high-end offerings. I wonder how much the G4e is going to cost." Sounds like a cool basis for (awfully) high-end set-top devices at that price, but imagine what that will cost 12 months from now! Yoiks.

This discussion has been archived. No new comments can be posted.

Sun's UltraSPARC III Processor Shipping

Comments Filter:
  • It's becoming more and more apparent that not all submitted news is being "read" as we would think. The whole point of the site is that we find out about news really quickly, right? So something submitted 23 days ago should have been posted, right? Look through all of the threats and check how many people submitted stuff that was rejected even though it's the same as the main post. Isn't this important?

  • Was down at the Level3 colocation here in Denver, when I bumped into a guy setting up some Enterprise 4500's. I mentioned to him how our E420R's would mysteriously/instantly reboot. Of course, this is last week, and both he and I knew of the problems Sun is having with their processor architecture, and how it is related to "garbage" getting in the cache. Of course, the point is that the UltraSparc III's share a lot in common with the UltraSparc II's, esp. the ones found in their E10K servers, and the ones found in the E420R's, etc. He mentioned that these next-gen chips were delayed _precisely_ because Sun was having so much trouble with their existing chips, and that they share a lot in common. Of course, this is just one sysadmin talking to another. But from the troubles I've had with the E420R's and E220R's, I'm much more inclined to believe him.

    Sun claims they have top shelf equipment, and for the most part, they're right. But they should be ashamed of themselves for sitting a problem as big as this one has been (the whole Cache/corruption thing).

  • I've not heard of any 64 bit PPC chips. Are there any? I haven't been paying attention to that arena.

    SGI's selling Intel boxes. My company was considering a lower-end MIPs chip for a (barely) embedded device for a while. Apparently we dropped that idea because we didn't know if we'd be able to get the chips.

    There's no confidence in the Alpha -- every so often a story comes out that Compaq's going to kill it (eventually) or that Compaq and Samsung just invested another couple of hundred million in research. I'd be a hell of a lot more confident in it if Compaq would come out with a firm position on it. Alpha's got a lot of stuff going for it if you don't mind dropping $1500 for your processor (Last time I checked that was the going rate for a good one.)

    I don't really know why PPC hasn't caught on. Maybe because it doesn't run Windows. It's all about marketing and no one's marketing PPC's. Well, Apple is, though most of their attempts at marketing seem to be geared toward suing people who are posting rumors.

    *shrug* I'm not out to prove anything. This is all just the opinion of a guy who's been in the industry for over a decade.

  • Oops. I was dumb... I didn't read the headings on the table and and only looked at the INT table, thinking that the left column was integer and the right column was floating point.

    My bad!

    Disregard my last comment!!!!!
  • OK, someone is a bit of an angry unix zealot.

    My point is that you can't tell a machine's usefullness simply by looking at how many clock cycles it spits out, or its supposed advantages in RISC architecture (and by the looks of things, this AC doesn't even know what RISC architecture would do for him -- he's just spouting out stats).

  • Uh....

    In my new job, we use Sun computers. My observation is that they are amazingly slow (US==UltraSlow), have buggy compilers, and clunky GUI tools (which I don't use; I am in a 20 year time warp, back to vi and dbx)

    It takes 2-4 hours to compile our CORBA library, depending on which machine I use. In contrast,
    the cheap Win98 box I use as a X-client can compile that same library in 20 minutes. (In theory, if I used parallel compiles on our hugely expensive 8-way processor machine, the compile time would be comparable; too bad the compiler will crap out and generate corrupted files)

    Not impressed.
  • Perhaps the best idea I have ever read on /. Stefan
  • A new workstation, with really nice plastics: The Sun Blade 1000 [sun.com] and a server, the Sun Fire 280R. (No link yet.)

    From their website:

    "Sun ushers in the next generation of exceptional tools for technical professionals with the Sun Blade[tm] 1000 workstation. The Sun Blade 1000 system accommodates up to two superscalar, 64-bit, high-performance UltraSPARC[tm]-III CPUs. It features a high-performance, crossbar-switch system interconnect that provides high bandwidth (up to 4 GB/sec.) for today's and tomorrow's ultra-high-speed processors and graphic subsystems. It also delivers plenty of internal disk and memory and a 64-bit PCI bus for incredibly fast I/O. The Sun Blade 1000 workstation provides both USB and IEEE1394 interfaces for connectivity to the leading edge in third-party peripherals. With state-of-the-art high-end graphics, dual monitor capabilities, and support for Sun's advanced storage systems, this workstation is truly a powerful, flexible next-generation desktop."
    Does anybody have pricing yet?


    Datasheet: Sun Blade 1000 [sun.com]
    Whitepaper: Sun Blade 1000 [sun.com]

    Adam Sherman

  • Sparc: Possible contender

    Yes, Sun seems to be a little bit boring at first, but I think this is exaclty what the industry wants. And no changing roadmaps twice a year. ;-)

    Itanium: Probable winner

    I agree, too. Not because the're good, but because the're from Intel.

    Sledgehammer: Expected no-show

    We will see. But I don't expect gread marketshare. Well, maybe, if AMD get's more industry support for their x86 stuff in the meantime.

    Alpha: Dying, Also Ran

    Yes, Compaq seems to be quite happy with Intel chips, no make competition inhouse. ;-)

    MIPS: Out of production, Also Ran

    Still in production, but IMHO only with a vague future. SGI is bouncing from MIPS to Intel (NT and Linux workstations) and back to MIPS (Origin 3000). Looks for me like a looser in the last years.

    HPPA: Possible contender if anyone knew about 'em. Nope. Intel developed the Itanium together with HP, and even if HP has extended its HP-HP roadmap, the future at HP is IA64. For example, the new SuperDome is already prepared for IA64.
  • But that's exactly my point. What end user needs access to a massive disk array on the machine sitting in front of him. That kind of thing is sufficient for servers only.

    And if I really wanted to play around with large clusters of disks, I could mess around with software RAID in Linux, or even make a minor Beowulf cluster. My point is that speed is relative to the applications you are running, not determined by the raw numbers a PR spits out.

    See the book "The Hardware Software Interface" for more information.

  • First they were announced in 1997 to be released in mid 1998. [sunworld.com]

    Then they were "on track" for volume shipments by the end of 1999 [findarticles.com].

    Now they finally have some of them in products in late 2000 [cnet.com].

    Very impressive. Heck, weren't we supposed to have UltraSparc IV's [zdnet.com] by now?
  • Sun hardware does the following:
    Prevents your kid brother playing games on it

    That's just blatantly not true. Doom works fine on my Sparc Linux box, and we used to have multiplayer doom games at a previous employer running Sparc/Solaris. Of course, it was a little unfair, 'coz the guy with the UltraSparc had a huge advantage over the rest of us (who were using sun4m machines). But I guess your kid brother probably isn't going to be interested in Doom any more, and AFAIK, Q3A and Unreal Tournament don't yet run on Sparc hardware.

    As for comparing server performance, note that the original coment didn't explictly compare it to a PC. Much as I like Sun hardware, others (particularly IBM, and to an extent, Compaq) have been making some really nice PowerPC and Alpha boxen recently that gives Sun a run for its money in the price/performance game. Sun is heading towards being too expensive for what it does, but the strong brand name seems to make up for that in the market.

  • Sounds cool. One of the guys here at work had a translator that converted normal documents into jive, now that was hilarious :)
    I don't know about the author, but I always get a kick at the troll-biters. I really wish /. could be a forum for intelligent conversation, but until that happens, I can still get my fill of amusement out of the [flame-bait deleted] who haunt /.
  • Maybe this new processor will substantiate Sun's claims of being "the dot in .com" (now isn't this untrue, since ICANN is literally the com in .com?).

    I fear that any IT administrators who obtain a system with this new processor may get power-drunk, and start cutting bandwidth to 16K/sec, forbidding users from contacting tech support, et cetera. Just a worst-case scenario.

  • Yes, Compaq seems to be quite happy with Intel chips, no make competition inhouse. ;-)

    Come again? The Compaq sitting next to my Alpha has AMD inside. Compaq seems to have their eggs in plenty of baskets.

  • It doesn't matter if Suns aren't well suited for your use. They are amazing servers, and nobody really gives a shit if they run games or not. That's not what they're for.
    So you like Athlons. Fine. I like RS/6000s. Fine. And RS/6000s don't exactly have a lot of games available. But the old 43Ps we had at work outserved the fancy shiny new NT boxes in every way. That's what they're for. So bug off.
  • Asynch/Raw I/O is vital for databases -- your database has to know FOR SURE if what it thinks it's written to disk has ACTUALLY been written to the disk and not just to the cache. If your machine goes down after a transaction has been written to cache, but before the cache has been written to disk, you lose that transaction. In the database world, anything that breaks transactional integrity is a VERY bad thing.

    "The axiom 'An honest man has nothing to fear from the police'

  • Actually if anything, the reason that Sun gots its ". in .com" saying was because Network Solutions were running off of E10K's. Which as we know now, they switched from E10K's to RS6000 by IBM. IBM is technically the ". in .com" now...
  • You don't know that.

    What's the last game you saw running on Solaris?

    I thought so. People don't play games on Solaris, they do actual work.

    Nevertheless, have you tried compiling and running Quakeforge on a Sun Workstation?

    Neither have I.

    Therefore, you aren't qualified to make that statement.

    TY & HAND.

  • Isn't also one a RISC design with the other is a um huh non-cracker design?

  • Yeah, that article was pretty cool. I'd like to see more like that, actually - going into depth about the Sparc architecture, especially on the new UltraSPARC III (and future chips).
  • by larien ( 5608 ) on Wednesday September 27, 2000 @07:22AM (#750362) Homepage Journal
    Check www.spec.org [spec.org]; in short, Intel just got kicked into touch. A new US III 900MHz workstation is getting over twice the FP performance of a roughly equivalent Intel chip.
  • by jmv ( 93421 ) on Wednesday September 27, 2000 @07:23AM (#750363) Homepage
    It's not that much about an article being posted with your name... It's to see that today's "late breaking news" is something you've submitted a month ago and that was rejected... The posting of stories just looks totally incoherent sometimes.
  • And a Compaq 8500 which takes up less rack space than and E450 Holds 8 faster CPU's 16GB of Memory and 11-64bit PCI hot pluggable slots, and onboard hardware raid. And it's a hell of a lot less expensive.
  • by Blue Lang ( 13117 ) on Wednesday September 27, 2000 @07:26AM (#750368) Homepage
    http://cgi.eb ay.com/ aw-cgi/eBayISAPI.dll?ViewItem&item=451336496 [ebay.com]

    you can buy one of the first five of these machines, signed by "big daddy" scott m.

  • Good point. I'd agree with that statement.
  • by yakfacts ( 201409 ) on Wednesday September 27, 2000 @10:29AM (#750376)

    Don't get me wrong, the Athlon is a great chip.

    But a Sun UltraSparc 60 running 360Mhz machine completes a setiathome packet in four hours. An Athlon 450 "faster" machine completes the same packet in 20 hours.

    Which is faster? (Hint 4

  • I just had a horrible idea.

    Why should Apple fight turf wars with Intel? Wouldn't it be nice if Sun invited Apple to port a version of Aqua to this new killer chip to run on top of Solaris.

    Can you imagine this chip on corporate desk tops? (Not Apple's market so no threat to the consumer base.)

    Just a thought.
  • are there any significant motherboard/general architecture (sp?) changes that would make this board incompatible with linux/sparc?


  • by DonkPunch ( 30957 ) on Wednesday September 27, 2000 @10:45AM (#750386) Homepage Journal
    Praise whichever Bruce Perens is your favorite, instead of always dwelling on negativity.

    AC, you have hit on an important philosophical point.

    It is perhaps symptomatic of our Western monotheistic culture that we believe there can only be one Bruce Perens. As a result, we create many warring factions -- each convinced that their's is the "true" Bruce Perens.

    The truth, of course, is that ALL of these groups are right. Bruce Perens is all-encompassing. He is what we want Him to be. We may think we believe in different Bruces but in reality, your Bruce and my Bruce are the same Bruce. Who am I to say that your Bruce Perens is not the true Bruce? Have I actually talked to Bruce? Of course not. I think I have, but I have no proof. The Bruce Perens on Technocrat and the Bruce Perens with a dot on Slashdot are equally worthy.

    The ultimate truth is that Bruce Perens is all of us and we are all Bruce Perens. And remember: In the future, everyone will be Bruce Perens for 15 minutes.

    Will the real Bruce Perens please stand up? Perhaps He already is, AC. Perhaps he already is.
  • So lessee ... for boxes that you might reasonably put on your desk right now, there are how many architecture choices? (And how many of them run Windows?)

    I'd be willing to wager that my Athlon 1000 mhz, 768 MB RAM, 40 GB HD desktop machine running Windows 2000 would blow this thing out of the water...

  • Come now, the UltraSparc volumes would make intel and amd roll over laughing.

    True, but then Sun hardware's all about quality not quantity. Although I don't think they'd mind selling SparcStations at PC volumes ...

  • Does 8MB L2 cache mean anything to you?
  • by scrytch ( 9198 )
    > Remind me when was the last time you've seen ANY version of Windows running on Solaris? (and no, I'm not talking about Soft Windows - although I'm not sure if it was for Solaris, and I'm not talking about WABI either)..

    So um, did you know that both of those are operating systems? If you're not talking about SoftWindows, what *do* you mean? Windows running on the, uh, Solaris CPU architecture or something?

    Windows doesn't run on FreeBSD or MacOS either...
  • That's funny, because none of the stats you cited indicated that you meant for game play. You talked about CPU, RAM, and HDD. If you were really talking about that, then you should've included graphics card stats. Of course, none of that would've given us a clue since you were just talking about "blowing it out of the water."

    Face facts. A Solaris machine has far more memory bandwidth, far more i/o bandwidth, far faster math calculations (especially FP), and has far more support for multi-processing, including more advanced bus-negotiation schemes than little old SMP.

    What you mean to say is that your little graphics card could blow the unmentioned Solaris box's graphics card out of the water (if the unspecified machine you are spouting off about even has one). Fine. Whatever. You're just trolling and spouting nonsense when you've never actually had any experience with what a real piece of hardware can do. It's not like your little machine would have a chance against a 4-way or more Solaris box doing OpenGL in software.

    Further, does UT even run on Solaris? Stupid OS...

    No, stupid UT. It's not Sun's fault that it hasn't been ported to Solaris. It's the fault of Epic for realizing that people buy a Solaris box for work, not for screwing around when they should be in class learning about how real computers work.
  • > For example, can Intel hardware currently run 200-prosessor SMP-systems?

    Sequent seems to think so. Been doing that for many years now. Not that you run that many processors in a SMP configuration, but I'll assume you use that term for any MP setup.
  • > Further, does UT even run on Solaris?

    When I was at Sun, the engineers were pestering Valve to let them port HalfLife to Solaris. Big time Team Fortress addicts there, there's a whole network of quake servers at sun. Quake2 was also ported, but it was OpenGL only, and you needed a pretty beefy framebuffer to handle it (Sun framebuffers are kind of like Matrox, great 2D quality, lousy 3D speed)
  • What magic bus will prevent memory and disk performance hits in the "real world"? Sun's got a good bus, and very good floating point perfomance. Floating point perfomance won't do it on it's own.

    Now how much did you say that mobo cost? Ironically, I could only find Intel mobos on the Motorola ATX page [mot.com]. Nothing showed up on pricewatch. 11,700 matches for G4 Motherboard on google, barf, I give up.

    Help me out, I'm as interested as the next guy in cheap computers.

  • > Maybe this new processor will substantiate Sun's claims of being "the dot in .com" (now isn't this untrue, since ICANN is literally the com in .com?).

    They used to be the dot in .com. and .org. and .net. ... that last dot, since the root nameservers ran off sun boxen. No longer, it's IBM boxen now. It was always just a goofy and amazingly stupid slogan, but it's not what's clever, it's what's catchy. God I hate marketing.
  • by ToLu the Happy Furby ( 63586 ) on Wednesday September 27, 2000 @10:58AM (#750407)
    SPECfp2000 results are available for the UltraSparc-III 900 MHz. It scored a 482. Pretty damn quick, especially when you consider that its score is more than 50% higher than the Pentium-III 933 Mhz, which got a 305.

    Wrong comparison. First off, the proper score to be using is SPECfp_base, not SPECfp_peak. In case you didn't know, while recently some pretty ridiculous SPEC optimizations have been sneaking into the compilers used for SPEC_base, they are at least optimizations included in the standard compilers; SPEC_peak numbers include optimizations which would break any other code, and thus are not considered widely applicable.

    Second, you clearly ought to be comparing the US3-900 to the P3-1000. Yes, the GHz P3 was unavailable for 6 months after it was supposedly "launched", but it is available now, or at least as available as the US3-900. Plus, the P3-933 numbers you quoted were hobbled by the horrible i820 chipset (yes, the i840 is only used for high-end workstations, but that's the market we're talking about here, right), and were obtained using an older version of Intel's Fortran compilers. Now, many people have complained that Intel's new compilers are so good as to call into question the usefulness of the SPEC_base benchmarks, since successive versions of the compiler have shown remarkable improvement in SPEC scores on otherwise identical computers. Still, they meet SPEC's rules for base scores, and everyone optimizes their compilers for SPEC, and most importantly the SPEC tests seek to benchmark not CPUs but entire platforms, and the compiler is an extremely important part of any platform.

    Thus, the numbers you should have quoted are:


    US-III@900: 427
    P-III@1000: 327

    Looks a lot less impressive, doesn't it. Especially when you consider the fact that any chip with an ISA less than 20 years old ought to beat the pants off x87 in SPECfp, due to x87's crippling 8-register stack-based FPU implementation. Once the x86 chips finally phase out x87 in favor of SSE2 (coming with Intel's P4 and later with AMD's K8 "Hammer" family), they will finally have a decent platform for double-precision fp, and their SPEC_fp scores should rise accordingly.

    In any case, considering the US-3 is destined primarily for the server market, the more important SPEC benchmark is not SPECfp but rather SPECint. Let's check those scores, shall we...


    US-III@900: 438
    P-III@1000: 438


    Of course, the real strength of Sun's UltraSparc line is its tremendous scalability. Yes, you're overpaying for a 1-way Sun system, or even 2- or 4-way, but what you're paying for is the headroom to later buy a 64-way machine without having to completely switch your architecture. Fine. Considering the poor scalability of x86 and Compaq's awful support of the Alpha platform, it's completely understandable that IT departments continue to overpay for Sun boxes. Intel's flubbing of IA-64 thus far has given Sun a 3 year reprieve, and it'll be another year and a half before a real IA-64 CPU (McKinley) shows up on the scene.

    Still, don't try arguing that Sun can compete with anything (except HP, which desperately needs to release a new processor) on straight price/performance. The US-3 closes the gap quite a bit from the extraordinarily outdated US-2, but not all the way. Sun's done a decent job squeezing performance out of an in-order design, but when Intel releases the SPEC scores for the P4 5 weeks from now they're going to make it (and indeed everything else) look extremely bad.
  • I don't think the question is so much, "How many of them run windows?" but, "How many of them run MS Office?" It sort of puts the situation in a whole new, scary, borgish perspective. Sure, we could use a Sparc to do our work, but nobody would be able to read our documents.


  • I'm sorry, it's a queue, not a stack, but thanks for playing.

    I really doubt the articles are processed in the order they arrived. It looks more like a FIMO (First In Maybe Out) to me.
  • Simple answer: it doesn't

    The SPARC III processor is FAR faster (don't look the the Mhz!), have more cache, faster bus - so it's WAY above what Intel or AMD has to offer..

    Ofcourse - on quantities of shipping - thats a totally different stories. I guess it's something like for every single processor that Sun sells - Intel sells at least 100 if not more.

    But then again - Sun is targetting to the high end segment (ofcourse, they'll happy to sell you an Ultra machine as a workstation - which will costs you twice what you pay for an X86 machine)
  • I think its great that Sun spends a lot of time and puts out some quality hardware. It's rare to hear Sun coming out with a new chip, but it seems like Intel or AMD are putting out chips every other week. I guess its an entirely different market though.

    It would be nice to see Sun produce mobo's for the new chip that support standard PC hardware. So you can build your own system type stuff instead of buying packaged Sun systems that cost an arm and a let.
  • On the linux-kernel mailing list, Anton Blanchard of linuxcare posted a boot log of 2.4.0-test9 running on a Sun E10000.

    Total of 24 processors activated (19149.62 BogoMIPS).

  • by HeUnique ( 187 )
    Remind me when was the last time you've seen ANY version of Windows running on Solaris? (and no, I'm not talking about Soft Windows - although I'm not sure if it was for Solaris, and I'm not talking about WABI either)..

    Why the hell a person would like to run Windows software on Sun SPARC based machine? for Office 2000? Outlook?
  • What page generates this postmodern tripe?
  • Here's [sun.com] the details about this chip...

    Now lets see here... The 9.6GB/s is *only* cache coherency. Addresses, not data. So I wonder how they're inflating this number? As a line address is sent out on this bus, I bet they are counting the entire line size.

    The cache BW is only 4.8 GB/s, and off chip is just 2.4GB/s. I'm not impressed. IBM has has been shipping machines with more than that for years. Heck, even Intel is going to catch Sun soon. S/390 G5 and G6 off chip BW is over 3GB/s. Infact, they're developing a machine available soon with 40GB/s data BW to the cache, and these caches are huge (like +128KB) low latency L1's, not the whimpy ones Sun is shipping (gee, notice they don't talk about the size?). Size matters. S/390 is the real choice for serious enterprise computing. And remember, IBM supports Linux across the entire line, from your wristwatch [ibm.com] up to the biggest and baddest boxes in the world... [ibm.com]
  • Anyone care to guess when an ATX mobo (single, dual or quad) will show up on pricewatch?
  • If NVidia produces fantastic boards thats great for them.
    But they could, and probably should release boards for the UltraSPARC III arcitechture.
    The processors are much better for calculations and openGL than Intel -processors.
  • According to this [sun.com] page, the 750 MHz models have a maximum power dissipation of 70W. While it does have the ability to reduce the clock speed by 1/2 to 1/32 for EnergyStar compliance, you're still looking at a possible max power consumption of 560W.

    Better hold off on that fantasy until we perfect a portable micro-fusion reactor.
  • The problem with Sun hardware is that it's simply too expensive for what it does.

    While you no doubt are right in many respects, I would like to challenge your statement on one end: reliability. Having been working in an almost exclusive Sparc/Solaris-shop, which then (due to the costs of hardware) graduately moved towards an Intel/Solaris-platform, I have also seen the maintenance costs and the downtime increase. When purchasing (almost) top-of-the-line server-hardware based on the Intel/PC-architecture, then the reliability may come close to that of Sun/Sparc. But then again, so does the price.

    For regular office and lab workstations, reliability might not be too much of an issue for most (after all, a lot of people aer used to 3-4 reboots/day from whatever other OS they are running at home). However in a tightly administrated setup, where ordinary users have no privileges to reboot machines, Sun/Sparc-hardware does have its advantage over Intel/PC-equivalents. Sure, it costs more - but so does network administrators.

    Now, for me - at home - I have an old Sparc Classic. Equipped with an extra netcard and some large SCSI-disks and a dat-drive, it makes for a decent file-server and acts as a filtering router between my local network and that of my upstream. The machine, while old - even ancient by todays terms - has no problems dealing with the at times rather intensive IO-load it is subject to. This due to the (then) innovative S-bus and crossbar-switch, which was default in Sun-gear (at at time when the ISA bus was the hottest in PC's). Even today, Sun's larger machines (creator, enterprise) are equiped with the same io-architecture, making them well suited for io-intensive tasks. IMHO, the IO-architecture of most Intel-based systems I have seen is no match for that of the Sun's.

    Of course, Sun has released PC's also (Ultra 5, Ultra 10) which are little but a PC with a non-intel-compatible CPU but with all the defects of the PC-architecture (the PCI-bus, to name one such defect). Choosing between such a Sun Ultra 5 (f.eks.) and a Pentium-III-based PC wouldn't be too hard....

  • 560 watts... A wafer should be able to cook my pizza, heat the toaster, run the central heating, and keep the grill running. All in *cough* software. :)
  • Yes, that old boss would just love to have the biggest most expensive toy in the world on his desk so he could upload his palm pilot data, write papers, and other things any old 486 could handle. Well, OK, so long as he needs tech support, I'll keep on smiling.

  • If you measure the transactions per second on a high end dual CPU linux / BSD system .. or the cost per 'page view' on a java app server .. and get a cost per transaction, or a cost per page view.. you would be shocked at what a cluster of well designed and load balanced linux boxen will do over a few enterprise sun servers.

    Though we sun may have 'reliable' hardware that has uptimes of years, I have a mail server running RH 6.0 that has about a 1 year uptime on it. (I know I should be patching kernel, etc.. but it's running fine for what it does..).

    I have web servers that stay up and running for 90-180 days without so much as a hickup .. and are rebooted for nothing more than hardware and kernel upgrades.

    If you buy good linux supported hardware, you shouldn't have any abnormal problems with faulty junk. (But if you buy cheep garbage, of course you will have problems...)

    As for point of failure, I would be my server farm any day on a pack of 5 linux boxen that have proper failover than 1 monster sun behemoth.

    To make it simple, sun sells enterprise solutions with an enterprise pricetag. Most corporations could get along fine with something that commodity hardware and linux, but corps seem to be funny that way and love to spend their enterprise budgets.

    As to me and my company, we are quite happy with our sun-free environment. With large disk support comming and improved raw-io to disk, the database argument will not be really there.

    IA64 will change things as well.


  • The problem with Sun hardware is that it's simply too expensive for what it does.

    Let's see. Sun hardware does the following:

    Runs reliably for years.
    Doesn't suffer the incompatability problems of cheap PC parts
    Keeps the room warm
    Costs a lot
    Prevents your kid brother playing games on it
    And (most importantly) looks good

    You're right when you say that PC's can perform most tasks a Sun workstation can do, but I've yet to see a PC come close to a Sun server for performance, scalability or reliability.

  • although off-topic for this bit of news, this
    thread is important - this is why I stopped
    submitting articles - after submitting several
    items and seeing them appear a week or two later
    from someone else, I decided it wasn't worth my

    before this is written off as whining, consider
    that the overall effect is to discourage
    submissions, reducing the usefulness of \.

    of course, I must be a masochist, since this post
    itself is most likely a complete waste of time,
    since it will do nothing to improve the situation
  • Replying to myself, I've got ES40 on the brain, but I guess that the DS20 would have been a more valid comparison.
  • You perhaps forgot that the UltraSPARC III is scalable up to 200-processors, and has twice the floating point performance of an equally clocked AMD or Intel -chip.

    Let us put an UltraSPARC III with 10 processors up against your Athlon system, and we'll see.

    The AMD and Intel -chips are great for Joe Average, and average needs, but these processors
    cater for entirely different needs. Even a single-processor UltraSPARC III -box would trash the athlon when it comes to professional openGL-performance.

    Don't get me wrong, you have a great system, but it's not the beginning and end of everything.
  • In my new job, we use Sun computers. My observation is that they are amazingly slow GUI tools (which I don't use; I am in a 20 year time warp, back to vi and dbx)

    The Sun software leaves a lot to be desired (run OpenBSD or SparcLinux rather than Slowaris), but the hardware is fantastic. You can even run SparcLinux on an E10000 - and apparently it compiles the Linux kernel in ~20 seconds. I doubt if your Win98 box can match that ...

  • Same here - not impressed. Sun has people hooked and they don't know better.

    When you're talking about *real* installations doing *real* computing work then Sun has people hooked because their hardware scales well. You could of course take the Yahoo! route of having massive amounts of simple Intel boxes, but for seriously intensive tasks a couple of big Sun machines are a lot less hassle. Rather than having to piss around with RPC or CORBA to synch up many computers working on the same problem, you can simply code your application(s) without regard for parallel computing issues.


  • I don't doubt that PC hardware is going to be cheaper price / performance wise, but it's not as expensive as you say. The Ultra 5 that I was looking at cost $2000, not $8000+. Throw in the PC card and you've got two systems for the price that I paid for my current Athlon setup.
  • If either the Alpha or MIPs were viable processors, one of them would have ended up owning the server marketplace. That hasn't happened. And the industry is getting psyched up for the Itanium. I expect it WILL end up taking over a good chunk of the high end market where the others have failed. The only progress I expect to see from the Sledgehammer is that it might drive prices down a bit depending on how it performs and when it shows up.

    Lets recap, shall we?

    Sparc: Possible contender
    Itanium: Probable winner
    Sledgehammer: Expected no-show
    Alpha: Dying, Also Ran
    MIPS: Out of production, Also Ran
    HPPA: Possible contender if anyone knew about 'em.

  • Baah. Use them all the time here at work, esp. for development. You wouldn't believe how much better it is to be compiling on one of these than on a PeeCee.
  • I'd be willing to wager that my Athlon 1000 mhz, 768 MB RAM, 40 GB HD desktop machine running Windows 2000 would blow this thing out of the water

    Of course it would. Suns have always been poor at running graphical, interactive applications (an example of something like this is an office productivity app), even with a Creator card installed. Similarly, if you want high framerate 3D rendering (for example, to play Unreal) then you'll also be disappointed.

    But ask a Sun to manage several thousand processes at once with heavy i/o (for example, an industrial-grade mail or database), or point it at a large floating point job (like rendering a movie-quality scene, or computing material stresses in an FEA/CAD application) and it will leave your PC in the dust.

    Did you see the bandwidth on this new SPARC? And the number of registers it has? That's an edge your PC can't match.

  • > Check www.spec.org; in short, Intel just got kicked into touch. A new US III 900MHz
    > workstation is getting over twice the FP performance of a roughly equivalent Intel chip.

    Though the 900MHz UltraSPARC III an amazingly high performer in specfp2000 (482 peak, for those not up to date). But it is hardly twice that of current x86 chips! The top line AMD and Intel x86 chips are both above 330 peak. That makes the fastest UltraSPARC III 46% faster than the fastest x86 chips, not 100% faster.

    However, the 900MHz USIII may not be available at the intro. It may only achieve server volumes around, say, a couple months from now. So, you really should be comparing it against x86 chips which will be available around the end of November. What will we have then?

    Intel (barring delays) will have announced their 1.50GHz and 1.40GHz Pentium 4. Rumoured (pretty much confirmed, btw) specfp for the 1.50GHz Pentium 4 is 524, which is well above what the USIII is published at. Mind you, the P4-1.5 is not expected to achieve volume production for quite a while after intro, but this very high score means that even a lower frequency P4 should outperform the USIII-900 score.

    AMD will likely have available for purchase by then a 1.30GHz (1.20GHz at the worst case, likely) Athlon, possibly based on their incrementally improved Mustang core, and probably outfitted with 2.133GB/s PC2100 Double Data Rate SDRAM. Additionally, they are slowly attaining optimized compiler support thanks to the efforts of DEC. I cannot say what the score will be, but I expect that it will be north of 400, even with current compilers.

    Of course, all of this is moot. Sun doesn't need to be faster than x86. The two platforms do not compete in the same market segments. On the contrary, Sun has to compete against the Merced (Itanium). And given Intel's comments about the processor (they have openly stated that spec2000 is not a relevant benchmark for Itanium, which means that it likely performs like crap), I don't think Sun will have much of a problem competing against Intel in terms of specfp performance until the release of McKinley (IA-64 generation 2) in late 2001 or more likely 2002.


  • Looking at the actual speed of the CPU in benchmarks, you should be comparing to a 1 GHz Pentium III. According real world test, the 500 MHz G4 should be compared to a much higher MHz PC counterparts.

    Intel Pentium III 1 GHz, $750 http://www.pricewatch.com/1/3/2448-1.htm [pricewatch.com]

    AMD Athlon 1 GHz, $445 http://www.pricewatch.com/1/3/2219-1.htm [pricewatch.com]
  • It was called "OpenStep". It was a complete port of the NeXTStep API layer (the thing we now call "MacOS X", or at least 80% of it) onto the Solaris kernel. It was a tremendous flop, largely because after signing on with great enthusiasm, Sun killed the project (probably realizing that if people actually used openstep, they'd have an easier migration path to non-sun hardware), and damn near killed off NeXT in the process.

    I feel safe predicting that Steve Jobs would sooner fly to Denmark to marry Bill Gates in a ceremony broadcast around the world than even consider inking such a deal with Sun again.

  • Hunh?
    Standard PC hardware?

    Sun workstations support:
    standard mouse
    standard keyboard
    standard svga monitor
    PCI network card
    PCI video card? (unsure)

    SCSI peripherals of all kinds.

    How much more 'standard' can you get?

    The only thing they DON'T support is IDE.

    And of course, they probably don't have support for your latest-and-gratest soudnblaster live or anything, but you don't need it. It's not a gaming platform.

    Build your own? About the only thing you have to buy from Sun is the MOBO, processor, and the case (you DO want a case, no?)

    Also.. where is the market for this? Companies that buy Sun don't WANT to build them themselves. Part of the reason sun is relaible is because of quality control; something you can't do if joe average is building his own high-end workstation.
  • by Will The Real Bruce ( 235478 ) on Wednesday September 27, 2000 @07:40AM (#750478) Homepage
    Slashdot sucks, film at 11.

    News for nerds, with sane submission queues [kuro5hin.org]

    A great place to troll [zdnet.com] (heck, even their articles are trolls!)

    An appropriate place for song parodies [poppyfields.net] because everyone loves song parodies!

    I realize slashdot has this all in one place, but the moderation system and the userbase is so pathetic these days that it's a wonder we old-timers bother to stick around. I've had this account for less than a week, and I'm already disgusted with slashdot! It used to take months for that to happen...
  • +else if (about(art,"mp3")) {
  • BUZZ
    I'm sorry, it's a queue, not a stack, but thanks for playing.
  • by Scooter[AMMO] ( 98851 ) on Wednesday September 27, 2000 @07:41AM (#750484)
    Back in February, Ace's Hardware [aceshardware.com] had a really great in-depth article [aceshardware.com] on the UltraSparc series.

    It starts by covering the history of the SPARC architecture, and what their naming conventions mean (eg. what is the difference between a US I, a US II, and a US III). It then looks at the design decisions that were made for the US3, which included previous UltraSparc binary compatibility, reducing load latency, pipelining, branch prediction, and scalability. The dicussion of all these topics are rather technical.

    The article is long, and the techno-babble may scare off some, but if you have any knowledge of basic CPU operation, particularly of RISC cores, or if you are just curious about some of the quirks related to designing a CPU, you'll eat that article up.

  • I'm talking UT frame rates here, people. Not server page views.

    I have my own priorities, and in this case my machine blows their's out of the water. (Further, does UT even run on Solaris? Stupid OS...)

  • Hrm. Indeed. I've not given much thought towards staroffice - it's always seemed to be very much unnecessary glitter. But if it works, it works. :) I actually considered installing it again today, because I've had an increased need to view those cursed .doc files... pashaw on standardizaion.


  • That was my point, to show that you just can't do a generic comparison based on the MHz of the CPU. The kiddie who posted the first message (the one to which I replied) claimed that any x86 chip would blow the UltraSparc away 'cause "its faster d00dz".

    My comment is only relevant when read in context. It should not have been moderated to +4. But the mods are often clueless....

    You might want to tone down your inflammatory language since you don't know anything about me. Just a hint.

  • by Otis_INF ( 130595 )
    And how many of them run Windows?

    What does windows have to do with these new Sun CPU's? nothing at all. So why refering to them then? Ain't this editor's note a bit too much offtopic and a bit too much of a 'biased' seed to control the discussions?

    Be fair. The new sparc is an awesome CPU. it will be combined with extrodinairy cool hardware in the new sun servers, and will run a kickass operating system, solaris 8. No need for offtopic rants and raves about microsoft and their products.

  • The athlon 450 doesn't exist

    The Athlon must be the 500Mhz then. There is one K-6 and one Athlon and they are 50Mhz apart. They are on the same desk. I don't pay a lot of attention the the fine details of PC hardware; I'm tired of dealing with junk.

    Find a clue

    A "me too" comment so soon?

  • As the second generation of the SPARC 64-bit architecture, the UltraSPARC III chip provides complete binary compatibility between applications written for previous generations of the architecture, delivering unmatched investment protection for Sun customers upgrading to next-generation systems. Furthermore, Sun's continuing development of the UltraSPARC II processor -- underscored by the announcement of the UltraSPARC IIe processor on September 11-- will extend the economically useful life of systems based on this architecture well into the future.

  • sorry about that... one problem of filtering out comments... the one you replied to I didn't see as I filter anything under 2 (and the d00dz comment had been moderated down). It looked like you were just making a flamebait statement...before I go off in the future, I'll try to look for context :)
  • Did anyone else read this? (I heard it earlier in a NDA.) You can mix different speed CPUs in the same system. Talk about major cool. So if you're in a tight squeeze, and you need some extra juice, and you've got an older or newer processor laying around, use it! Yummy.
  • Now, if only they'd etch a cluster of them onto a single wafer, they might even catch on.

    (Wearable 8-processor UltraSPARC III's would make for a decent FreeCiv client. It'd also keep you toasty-warm in winter.)

  • there ARE Sun clones available

    Yup, and I wasn't saying otherwise, although I suppose it could easily be inferred from my comment. I have even come across a few clones like the Opus ones, but not Fujitsu's.

    For info on all things Sun, chekc out www.sunhelp.org, which includes great info on older Sun models if you're ever thinking of buying one.

  • The only thing they DON'T support is IDE.

    I was just looking over their workstation lineup (got a few K to spend) and I noticed that the Ultra5 appears to have an EIDE hard disk and CD-ROM. Apparently it is supported.
  • To make it simple, sun sells enterprise solutions with an enterprise pricetag. Most corporations could get along fine with something that commodity hardware and linux, but corps seem to be funny that way and love to spend their enterprise budgets.

    Given that the contracts a good deal of data centers sign with clients specify hefty penalties for downtime other than scheduled maintenance, I don't think that Sun gear is overpriced for what it delivers. Its not just a matter of buying good hardware, its a matter of buying hardware with failure rollover.

    Also, most people don't quite get the idea when they see the term: enterprise. Enterprise class machinery typically deals with multiple boxes with multiple CPUs and tens or hundreds of gigs of RAM with failure rollover and other nifty features.

    And the cost is mostly in the hardware. I'm sure someone could design and build Linux box that had as much hardware as most of the Sun boxes in our datacenter, but I'd doubt it would be much less expensive. 20+ CPUS and twenty or thirty gigs of RAM per machine starts to add up when you have a cluster of fifteen or twenty boxes for just one client.

    Now if you're only talking about clustering a few Linux boxes each with two or four processors, then, yeah, Linux will likely get you much better price/performance with the same sort of stability. Unfortunately, this isn't quite what Sun has in mind when it targets the Enterprise market.

  • "Doesn't suffer the incompatability problems of cheap PC parts"

    For the most part this is true...but check out the SCSI connectors on the Ultra 1 and Ultra 1 Creator. Yay for gratuitous incompatibility

    Ahem. I remember trying to get a printer lead for my Sparc and HP Deskjet. The normally reliable Black Box peripheral people assured me they could get the right lead for ... 60 pounds sterling!!! The lead turned out to be the wrong one, and I wasn't even entitled to a refund, just a poxy credit slip. Eventually I got one from a very small Sun reseller for a tenner.

  • Sun is trying to retain the upper end market (8+ CPU servers) against other UNIX/Linux servers. and trying to move downstream to the personal/small business servers.
    MicroSoft is trying to move into the large server market with NT products that work on 8 or more CPUs.

    If customers are using Solaris, they aren't using NT.
  • It seems like Compaq has been shipping the ES-40 with 667 MHz EV67 CPUs for ages. Does anyone know how well the ES40 mid-range server would compete with this new 280R? I realize the 280R isn't actually shipping yet, but it boasts almost twice the memory bandwidth of the ES40 (9.6 versus 5.2 GB/s peak). OTOH, the ES40 can take 4 CPUs, 32 DIMMs, 8 disks, and 10 PCI cards. Hmm.

    For that matter, when is Compaq's next revision of their mid-range servers supposed to come out. I'd at least like to see them move to the 750 MHz CPUs.

  • My first encounter with Sun products was when a SparcStation 1+ was offloaded onto my desk. My employer at the time wanted an FTP server, and the budget wouldn't run to a new machine. Consequently, the sys admin put together the SS1+ from a box of Sparcs abandoned by the company typesetters.

    That machine ran for years, with the only downtime occuring when I switched from SunOS to Linux (RedHat 4.2 - the first distro I remember that actually outshone the commercial Unices).

    Since then, I have worked on a mix of Sun and Intel hardware, but have always favoured the former for its reliability. While PC's are cheap, thanks mainly to the proliferation of clones and myriad peripheral manufacturers, many reliability and performance problems stem from the subtle incompatabilities of PC parts.

    Most Sun computers contain nothing but Sun manufactured parts - although many parts are simply rebranded third party bits. For instance, my CD drive is a Toshiba with a Sun fascia, but it has been tested thoroughly for compatability and reliablity. This means higher prices, but when I plugged the CD drive into my Sparc 5 back in '96, I had a greater expectation that it'd work than friends plugging CD drives into their PC's.

    This 'monopoly' on hardware often garners criticism from Apple's detractors, especially after they pulled the plug on Mac clones. However, Apple hardware shares the increased reliability of Sun equipment in part thanks to this 'monopoly'.

  • Sun hardware is NOT expensive for what it does,
    and Intel based systems cannot compete in the fields that the UltraSPARC is aimed at.

    For example, can Intel hardware currently run 200-prosessor SMP-systems?

    Sun hardware is expensive, but for the added expense, you get safety and reliability, that a lot of companies ARE willing to pay.

    Of course, for the average Joe User, the UltraSPARC III will never be an issue.
  • That's exactly my point. As a server Solaris is pretty good. As a game machine it absolutely stinks.
  • Summary of most posts:

    Its not X86, its cool.

    Its Sun, its cool.

    Does it run Linux?

    I can't wait to get a Boewulf cluster of these

    Sun sucks. Alphas rule.

    Sun sucks, IBM power pcs rules

    Why would this chip be on a desktop since its a server chip?

    Slashdot sucks, I posted this a long time ago. Lets moderate stories.

    Lets see if I can post a link to Sun [sun.com]to get moderated up as informative.

  • The problem with Sun hardware is that it's simply too expensive for what it does. Alpha & Power systems are cheaper (in the UK anyway) and Intel based systems can perform 99% of the tasks that Sun systems can do.

    So, sorry if I'm not over the moon.

  • Unlike Intel's processor with 3 i's, maybe Sun's WILL actually make the internet faster (from the server side of course, to reduce the "demands of the Net Effect" i.e. /.)
  • Apples, Oranges... man you are showing yourself being clueless with a comparision like that. ever look at RC5 results... on a 400Mhz USII I think it gets about 700Kkey/s a PIII650 gets like 1.8 Mkeys... the 700 to 1800 is far more than the Mhz diff (the PIII would have to be at over 1GHz to be just spped scaling). The Sparc may just do the instructions for Seti better than the athlon. why Alphas smoke on DES but not on RC5.
  • I always thought the biggest mistake Sun and (formerly) DEC made was not flooding the market in the early '90s and going after the desktop, back when SPARC and Alpha were much, much, faster than Intel/486 and Motorola/680x0 and Un*x was much better, more stable than windows or macos (of course, un*x is still better, but IMO the gap is smaller). I guess they were worried about giving up their big margins on servers.

    I think there is another window of opportunity here for Sun and Compaq/Alpha to strike. They are 64 bit already. Open Source is closing the applications gap and removing hardware-dependency. It would be cool if they could win by selling midrange CPU/mobo combinations at prices similar to what you see on pricewatch for midrange Athlon and PIII.

    OK, It'll never happen, but I can dream
  • SPECfp2000 results are available for the UltraSparc-III 900 MHz [spec.org]. It scored a 482. Pretty damn quick, especially when you consider that its score is more than 50% higher than the Pentium-III 933 Mhz [spec.org], which got a 305.

    Of course, if you consider cost, it nearly evens out.. but people don't buy Suns cause they are cheap.
  • by jmv ( 93421 ) on Wednesday September 27, 2000 @07:20AM (#750567) Homepage
    Slashdot article selection:

    while (articles_left())
    article art = get_article_somewhere_in_the_stack();
    if (about(art, "microsoft"))
    else if (about(art, "big corporation") && (rand()%2 == 0))
    else if (rand % 10 == 0)
  • That's for applications; I'd be surprised if sparc/linux works perfectly without some tweaking. Bear in mind, part of the Solaris 8 release was to add support for US III chips.
  • by Anonymous Coward
    Yes, Linux is incompatible with the new platforms but the port should already be under way.

    There's the new processor, main bus, UPA64S/PCI bridge, USB for keyboard/mouse, 1394, fibre-channel controller, tod chip... just to list a few of the architectural changes that impact privileged software. At the user
    level it's fully compatible with existing apps, of course.

Houston, Tranquillity Base here. The Eagle has landed. -- Neil Armstrong