Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
×
Apple Businesses

Dual G4 Mac Cube 131

ijx writes: "Those of you with a hardware-hacking bent should enjoy this - a Mac Cube modded to accomodate dual processors, courtesy of AccelerateYourMac. It seems that it uses the same dual-proc module as a Sawtooth G4 Tower. My question: will it melt?"
This discussion has been archived. No new comments can be posted.

Dual G4 Mac Cube

Comments Filter:
  • Cool! (Score:4, Funny)

    by darkov ( 261309 ) on Saturday December 01, 2001 @07:16AM (#2640776)
    Now you can set how brown you want your CD/DVDs betfore it pops them out the top!

    But seriously, a nice hack. Sounds blazingly fast. Maybe it would have sold better in this configuration?
    • I agree. If they were offering MP Cubes, I'd be very tempted to have opted for it instead of my single processor minitower.
      • Re:Cool! (Score:2, Informative)

        by Ziviyr ( 95582 )
        I agree. If they were offering MP Cubes, I'd be very tempted to have opted for it instead of my single processor minitower.

        I'd like to hear what tech support has to say about it when people call up asking why their cube is on fire.

        "Uhhh, yeah! Innit the hottest product we've ever sold? Wanna buy some really cool speakers for it? I can transfer you to sales! -click-"
    • Re:Cool! (Score:2, Interesting)

      by class_A ( 324713 )
      It's a shame Apple don't offer some ultra-dense server solutions built on Mac OS X Server as they obviously have the capability to produce some pretty robust systems.
  • Will it melt? (Score:2, Informative)

    by Anonymous Coward
    No, it's RISC-based. You ought to try one. No heat problems that a small fan can't handle. Unlike x86/CISC machines, there's not a huge chip with (around) 100 million transistors generating heat.
    CISC chip do put off heat, but it's VERY easy to control, and the wattage is lower than CISC. That's why RISC technology is used in many hand helds.
    • Is this a troll? (Score:4, Informative)

      by autopr0n ( 534291 ) on Saturday December 01, 2001 @07:36AM (#2640800) Homepage Journal
      RISC and CISC aren't technologies, they're design ideologies. And like political ideologies, they have been strayed from in the interests of practicality. Intel/AMD chips all use RISC-ish designs with instruction translation stages. And the PPC architecture has some very complex instructions (alti-vec?)

      The reason risc is more common nowadays is because it's easier to design and write optimizing compilers for. Software issues, not hardware ones.
      • Re:Is this a troll? (Score:4, Informative)

        by Elbereth ( 58257 ) on Saturday December 01, 2001 @08:40AM (#2640850) Journal
        He's got the general idea right, though. He might get mixed up with specifics, but something with less transistors will run cooler. The PPC chips run very cool. Athlons run extremely hot. Pentium III and IV chips don't run as hot as Athlons, but I'd hate to see what happens when the heatsink malfunctiones on a 2GHz P4.

        You can run a PPC CPU at temperatures unheard of with AMD and Intel CPUs.

        Not all RISC-based CPUs run cool, though. Because the PPC was designed for embedded use, it runs a lot cooler than a DEC Alpha. Try touching a fast Alpha CPU. Try touching a slow Alpha!
    • Re:Will it melt? (Score:2, Interesting)

      by Corrado ( 64013 )
      I recently turned off my Dual Celeron 333 because of a bad fan and I couldn't beleive the amount of noise I was dealing with every day! All those fans make quite a racket. Since then I have been on a quest to get a silent (or as close as possible) system.

      I thought about getting a used SPARC Station (or something similar), but I think this Dual Cube thing would be great! Man, I wish Apple would have sold this. I think I might have given up all my other computer equipment (except my near-silent FREESCO box :) to have it. *sigh*
      • Re:Will it melt? (Score:1, Insightful)

        by Anonymous Coward
        you should get a "overclock friendly" celeron and UNDERVOLT it instead. seriosly, im running a good old 300A at 300mhz/1.45 volts rigth now. big heatsink and no fan. works like a dream.
        i also disconnected the PSU fan, but thats a nother story. (tip: run an open case, and remove the lid on your PSU box.)
        i also use a voodoo3000. no fans at all then.
        and read up on storagereview.com witch harddrives are silent.
    • Notice the huge heat sinks he stuck in there? That's because it's HOT! Also, you do know that the G4 Cubes HAVE NO FAN right?
      • You did know-

        That's a stock heatsink modified to make contact with the two G4's instead of the single one that was stock.

        So he's not doing anything other than going with Apple's design decision. It could probably run with a cooler heatsink, but since it's not aided by a fan, and you want it to work in rooms that don't have air-conditioning (thinking of consumers who buy computers and then use them in stuffy offices) there's probably some tolerance designed in so that it doesn't overheat.
    • Comment removed based on user account deletion
    • by Billly Gates ( 198444 ) on Saturday December 01, 2001 @09:32AM (#2640900) Journal
      So why did aceshardware.com chose sun over x86 over a new server? [aceshardware.com]



      ..."The peak power consumption of the UltraSPARC IIe is 13W at 500 MHz, which minimizes cooling requirements for the chip and in turn a potential threat to server stability should the datacenter cooling/environmental equipment malfunction.
      In fact, one of the primary barriers to broad acceptance of the Itanium by OEMs is its large 130W power appetite -- literally 10 times that of our own UltraSPARC IIe."



      I am aware that cisc vs risc is a flamebait in alot of areas but risc was specifically designed to have less complexity in the chip in exchange for higher clock speeds. In many ( not all ) situations risc is usually %25-%40 faster on equilivant megahertz basis or can run at a slower megahertz and consume less power for a similar speed. X86 has alot of baggage in it and consumes alot more power then powerpc's, sparcs, and even alpha's and offers less performance or the same if its clocked ridiculously high. This would make it less hot then x86 cpu's running. I think the pentium4 and the athlon are almost an embarrassment to the electrical engineering community.

      • Hiya. I work for a company that works with Linux in embedded systems (Sun is, or at least was, a customer/partner/something at one time, so I don't want to name who we are out loud right here). We've dealt with Sparcs a bit, and can say quite emphatically that they are *not* (even close to) low-powered chips. Even their "embedded" reference board that we tested took something in the area of *80 watts*. Compared to the embedded x86 boards we work with (never mind the PPC, MIPS, SuperHitachi and StrongARM boards, almost all of which best the x86s by far in power consumption), that's absolutely hideous -- and perhaps part of why Sun's handhelds are using Intel's StrongARM chips rather than their own "embedded" sparcs.

        Saying that many CISC chips are lower power than many RISC chips is certainly true. Citing Sun as an example -- bad call.

        (Now, the system that my coworker was dealing with was evidently not the same one aceshardware reviewed -- but even 13W for the chip is way, way too much in embedded space. If you're looking for something to hold up as an architecture conducive to low-power chips, consider one of the examples I gave above).
        • by cduffy ( 652 )
          Saying that many CISC chips are lower power than many RISC chips is certainly true.

          Needless to say, I got that backwards. *sigh*...
          • If you crank down the speed of any CPU low enough, even Pentiums, they will get to the point where they don't need a heatsink just to survive. However, I don't think anyone cares about such underperformers/relics in this discussion.
            • Re:oops (Score:2, Interesting)

              by cduffy ( 652 )
              Underperformers need not be relics. Whereas many of the embedded chips I work with aren't even in the same playing field as the G4s being discussed here (though G4s *are* among the chips we support), quite a few of the systems we support *are* built for speed -- just with that not foremost in mind.

              Compare Intel's "mobile" chips used in laptops with those for the desktop market; the performance difference isn't all that great, while the difference in power consumption is large. These are still quite a bit more power-hungry than most embedded chips, but they illustrate that not all new development is done with performance as the primary goal. There's a lot more than underclocking which can be done to cool down a chip (don't look here for details, though -- I'm software, not EE).
              • by peter ( 3389 )
                Get your details here [realworldtech.com]. If you haven't read any of Paul DeMone's Silicon Insider columns, but you like computer architecture, you need to check it out. Well written, pretty much unbiased, and usually quite insightful.
    • Re:Will it melt? (Score:2, Informative)

      by SlamMan ( 221834 )
      This is obviously said from a person who's never sat extra close to his iMac because the building's heat is off on weekends. Just because there's not a fan doesn't mean it doesn't get toasty warm.
    • Strictly speaking, talking about RISC and CISC is irrelevant when comparing x86 and PPC, for one the "RISC" G4 actually has more instructions than the "CISC" Athlon (dunno whith the new SSE instructions though).

      Anyhow there's not much to say about the powerconsumption, G4 clearly rules (and therefore runs cooler and quieter, no fans...).
      • sadly you're WRONG. The 7400 / 7410 G4's were totally fanless but ALL of the 7450 / 7440 G4's (yep, TiBooks too) have some kind of on-chip active cooling. The Quicksilver machines sound like someone is hoovering next door...
    • Re:Will it melt? (Score:5, Insightful)

      by Anonymous Coward on Saturday December 01, 2001 @12:04PM (#2641126)
      --I don't understand why this comment got modded up. RISC vs. CISC has nothing to do with how much power a device uses. RISC arcs have a wide range of power consumption characteristics, from your low-power ARM to your space heaters like the Alpha.

      As I recall any of the Alpha line of microprocessors (a pure RISC design) uses significantly more power than any of your desktop x86 processors (I believe in the 100W or more range).

      I imagine the reason the conclusion is incorrectly drawn is due to the fact that x86 is one of the only (if not the only) high performance CISC architectures out there. Since he has only one sample point, the poster must have assumed that "all CISC designs consume more power than RISC designs".
      • I don't think the original poster is as right as he thinks, but RISC designs can use less power because they don't have to devote any die space (ie. transistors) to translate CISC instructions into RISC-like instructions (which are common to all x86 designs currently). This means RISC ISA implementations can use fewer transistors inherently and thus consume less power.

        Of course, this doesn't mean you're wrong either. RISC designs can suck up juice too.
    • The chip is not likely to be the cause of melting - the DVD-ROM is. My housemates and I went on a CD copying spree one evening and would have copied about a hundred CDs using two burners, a ton of hard drive space (buffering) and about five CD-ROMs, including a G4 Cube. After some six hours of running the DVD-ROM constantly, it literally burnt the last CD we put into it. The CD came out slightly blackened and we decided to call it a night.

      Notably though, the PowerBook G4 didn't have this problem - it's fan didn't even turn on.

  • Interesting (Score:3, Interesting)

    by flonker ( 526111 ) on Saturday December 01, 2001 @07:29AM (#2640791)
    Modifying a machine to be dual processor without swapping out the motherboard? Anyone have any more info on this? The article was kinda skimpy on the details, it assumed we knew already.
    • Re:Interesting (Score:2, Redundant)

      by Anonymous Coward
      The processor is in a socket.

      Most Apple machines have been so, since the biege G3, in a ZIF socket.

      Even Powerbook G3 is in a socket, and can be upgraded. Among the very few laptop that can be upgraded to a better processor, hope Intel do copy that idea
      • Well, you can't really stick two CPUs in one CPU socket. From the pictures it looks like the chips are on cards that can be swapped out.
    • Re:Interesting (Score:5, Interesting)

      by Osty ( 16825 ) on Saturday December 01, 2001 @07:39AM (#2640805)

      Modifying a machine to be dual processor without swapping out the motherboard? Anyone have any more info on this? The article was kinda skimpy on the details, it assumed we knew already.

      This is a by-product of the PPC architecture used by Apple. You're thinking in a PC mindframe, where the CPU sockets are built directly onto the motherboard, and to do a dual setup you need a mobo that supports two CPUs. A Macintosh is different. The CPU is on a separate card. The dual CPU card is the same as a single CPU card, but with two CPUs on it. It connects to the motherboard in exactly the same way. Now, I've never taken a mac apart or done a CPU upgrade, so I don't have much authority here, but that is the gist of how it's done. PC architecture vs. Mac architecture.

      • I seem to remember a 'build your own computer project' (Byte mag?) that worked similarly. Imagine a CPU bus. Add as many CPU's as YOU need. Just stack as many as needed in a box. Well maybe not THAT many.

        Ok where can I get a quad CPU board... Hell with that! Give me a 2^8 mod - would have to rename thou. Hmm.. Beo,G4,256...BfG2k anyone?

        With all that artic silver thou, I would think that it would be difficult to remove.
      • Re:Interesting (Score:4, Interesting)

        by SlamMan ( 221834 ) on Saturday December 01, 2001 @10:06AM (#2640948)
        Exactally right. We got a dual 500 processor card donated to us because the machine was damaged in flood (there was liquified ceiling tile corroding the rest of the motherboard, but the card was high enough to stay out of it). All we had to do was pop it in the slot that had out single processor 400, and off it went, rendering out movie files at extra blazing speed.

      • here [xlr8yourmac.com] is an article from the same site that shows photos of what the connector looks like.
      • Won't that slow it down, by making it pass over some sort of bridge, though?
      • "Now, I've never taken a mac apart or done a CPU upgrade"

        I have, several times and your post is completely accurate. I don't know much about PC upgrades, and wasn't even aware of the difference with this. Apart from a horrific experience trying to install a CD-R in my nephew's Compaq, I've never had experience with non-Mac PC hardware. No Mac upgrade (CPU, PCI card, video, USB, memory, etc.) has ever taken me more than 5 minutes. I appreciate Apple's attention to detail in this regard.

    • by jmenezes ( 100986 ) on Saturday December 01, 2001 @02:45PM (#2641471) Homepage
      This has been an easy thing to do with Macs for years.
      Unlike PCs, which are instaled in ZIF sockets, Macs have had a few diferent ways of connecting the processors, but with one unique, and very advantageous difference.
      Since the PCI powermacs came out, the PPC processors have been on daughtercards (except a rare few consumer class machines which had it soldered onto the mobo, but thats a different story).
      These daughtercards can have several proicessors on it, and can easily be interchanged.
      Unlike PCs, which have the entire processor bus on teh mobo itself, with separate traces for each possible CPU, powemacs usually have the processor bus itself as a daughtercard connector, so there is no difference between a mobo for a 1 cpu machine and a 4-way machine (Such as the Daystar Digital mac clones, which used the same mobo as the PowerMac 9500 of the time, just a 4-way 604e CPU)
      These daughtercards have seen a few different formats over the past few generations. The original PCI powermacs had a slotted daughtercard which was simply inserted like any expansion card, and could be used to upgrade any of those machines with up to a G4 cpu, made by some upgrade manufacturers. Then with the Beige G3 machines, came the Zif socket, but which also supported multiple CPUs if necessary, although not many upgrade companies made them.
      Then, with the Sawtooth G4 machine (the one which finally brought AGP to the mac) the CPU went onto a daughtercard which was connected via a 423(?) pin rectangular socket, which can accomodate several CPUs.
      This socket is what was used in the MP G4 machines, making it possible to use the same Motherboard in all their desktop G4s, simplifying the product line.
      The iMacs also use a similar, but incompatible socket, which could theoretically support multiple CPUs.
      -=-=-=-
      Also, for information on how to Overclock these daughtercards, and several different Macs in general, visit http://violet.berkeley.edu/~schrier/mhz.html
      for instructions and some interesting reading.

      And for those wanting to overclock a Mac....
      beware: you need to make motherboard/daughtercard modifications to do so.
      • The iMacs also use a similar, but incompatible socket, which could theoretically support multiple CPUs.

        Yeah, but the boot ROM is soldered onto the CPU card in iMacs. (this may be different on the newer Kaheis(sp?) I'm not real sure). When you buy one of the G4 upgrades they ask you to send in the old card so they can pop-off the CPU and replace it with a faster one(sneaky ain't they?)
        • Very true, the boot ROM is soldered on the CPU card for older iMacs.
          but at least one upgrade manufacturer had a very interesting workaround for this, they provided their own boot rom, and an additional flash rom.
          Before installing the upgrade card, you would run a utility which would make a dump of your Apple-provided ROM, and save it to a special place on the HD.
          You could then proceed to install the upgrade card, and upon boot-up, the manufacturer-provided rom would search a specific area of your hard drive and get the ROM image, and copy it over to the Flash ROM, and the upgrade is complete.

          As for the current and next iMacs to come out, however, i am not aware if the ROM is still on the CPU daughtercard or finallty made its way to the motherboard.
  • Given the rumoured low-heatness of the coming G5 processors I wonder if Apple's going to revive the Cubes or if Sonet could make an upgrade card?
    • No way is Apple going to bring back the Cube: it didn't sell well before and it won't sell well again. And there are probably too few Cubes out there to justify Sonnet making an upgrade, unfortunately.

      Apple's will be on to other stuff [terminal34.com] soon anyways.
      • Well, since the cube uses the standard ZIF processor module (that's why he was able to fit the dual proc card in), Sonnet will not need to make a special cube version, just one that isn't exceedingly large. Just like you can actually fit smaller AGP cards into the cube (like a dual headed one).
    • Apparently Apple sent a memo out to its retail partners telling them not to throw out their cube displays... I guess there's always hope :) Too bad the things weren't appreciated until it was too late.
  • Comment removed based on user account deletion
  • Won't melt. (Score:5, Informative)

    by gig ( 78408 ) on Saturday December 01, 2001 @09:42AM (#2640910)
    The G4/450's that are in this Cube only use 7-11 watts each. Compare to 50-70 watt Athlons and Pentiums, and you can get an idea of why this works without a fan. Standard Cubes run cool, so there was some room there. He plainly states that he is monitoring the CPU temperature with a utility app and it's cool enough. It may not work in Florida if you don't have air conditioning, but that's why he checked with the CPU temperature utility.

    The G4 towers have a fan, but they are there at least partially because the box has room for three more hard drives, one more removable drive, and four PCI cards in addition to the stock stuff, so you have to leave a big margin for error. The fan switches off when the machine sleeps, though, and the boxes don't run hot. Also, the power supply is inside a G4 tower, but it is outside on the Cube.
    • Re:Won't melt. (Score:5, Informative)

      by barzok ( 26681 ) on Saturday December 01, 2001 @10:36AM (#2640989)
      Those Cube cases are made of GE Lexan, the melting point of which is far higher than you could possibly generate even with a dual Athlon.
      • True, but if you heat up your computer too much, it starts to have issues. In general, your components will fail a lon time before the case - the only time I've seen the opposite take place is when a computer survives a fire.
      • How did this get modded as being _that_ informative. Sure it's true (well, I have no reason to doubt it), but it's not really useful. The fact that the case hasn't melted yet has little to do with whether the silicon still works. High heat will let the doping additives in the silicon migrate, which is not good when there are feature with a size of 0.15 microns, or whatever the process size is. In other words, heat up a chip too much, and some of the transistors might stop transisting.

        This is why people worry about overclocking, for one thing. They would just try it if the worst that could happen was that the case could melt! Heating up a semiconductor increases its conductivity, and makes it more likely that you will burn something out. (Well, part of the burnout risk from overclocking is from the higher clock speed directly (which produces higher currents), not just the resultant heating.)
    • Hmmm.. I trust actual temp probes more than onboard sensors. But I guess its close enough.
    • Re:Won't melt. (Score:3, Insightful)

      It is also noteworthy that when Apple realized they had to use a fan in the G4 tower, they did a very good job of engineering an especially efficient and quiet cooling system. It is not possible for me to hear the fan on my G4 tower, sitting approx. 6 inches from my left foot, over the noise from the god only knows how many fans in my roommate's HP, sitting about 20 feet away.
      • Which G4 do you have? I have a 466 mHz Digital Audio, and the fan is almost as loud as my old PowerComputing clone! I'm thinking of replacing the fan with one of those quiet ones.

    • IIRC the Powerbook G3 Firewire, aka Pismo also uses the same processor socket as the G4 Tower and Cube. Mmm, a dual processor laptop, what could be better? Or what could be warmer? Anyone want to guess how much power the laptop would draw like that? I wonder how long the battery would last on such a beast?

      Then again you have the other extreme. Lets put the processor card from the Powerbook G3 in a G4 Cube and see how long it would take before the victim noticed.
  • by cjhuitt ( 466651 ) on Saturday December 01, 2001 @10:10AM (#2640955)
    Would this be a potential solution for the cracks that were supposed to be in the Cube cases? Get it just warm enough to meld them together again...
    • Please...

      Those "cracks" were faint mold lines in the curves of the plastic. It is extremely difficult to bend plastic like that and avoid some defects.

      Of course, as soon as people heard "crack" everyone thought of the unit falling apart. The mold lines were actually very difficult to see. In fact, when ZDnet.com ran an article on the cracks they had to put the unit on a lightboard in order to effectively display them. Only with light shining directly into the plastic were they even noticable.

      As always, people overreacted to it and caused a big fuss. The Cube was and continues to be a great machine. Maybe if Apple had priced it a little better from the get-go, it would have seen more success. It made sense to have a Mac that was a step up from the iMac in power and was able to take a monitor of your choice.
      • I have seen cubes in stores and unfortunately, they all exhibit these cracks, and they were definitely not mold lines. Each cube had cracks in a different part of the case and they were curved, unlike mold lines, which are straight. There aren't a lot of them but they are there.
  • Heat and mods. (Score:3, Informative)

    by Stenpas ( 513317 ) on Saturday December 01, 2001 @11:20AM (#2641034)
    It's not really too much of a surprise he can do this without the computer blowing up. PPCs are designed to be cool (as in temperature, but also in neatness I guess) processors. Just put a huge heatsink on it, and you're set. My G3 upgrade that I overclocked by 40mhz (won't go any higher than 280mhz) maxes out at 43 degrees celcius. It's been awhile since I checked the specs out, but I believe it can handle a maximum is 90 degrees. That doesn't sound to relevant, but remember, the G4 the Cube uses is basically a G3 plus altivec. I'd be concerned about the other componets melting before being concerned about the processors even if it is in an extremely tight space.

    Ok, enough of the boring stuff. Here are some links to keep you guys semi-entertained. The Cube goes into the Museum of Modern Art [nytimes.com]. For those without a NYTimes subscription, MacObserver summarizes the article [macobserver.com].

    Now for the fun stuff. Case mods! Check out the Kleenex Cube [applefritter.com] and the Blue Smoke Cube [applefritter.com]. Also, here's an All American Cube [xlr8yourmac.com].

  • by firewort ( 180062 ) on Saturday December 01, 2001 @11:33AM (#2641051)

    Apple stopped using ZIF a few years ago.

    In the tower machines, pre-cube, it went like this:

    G3 Yosemite - zif G3 processor, Blue and White minitower.

    G4 Yikes! - Zif G4 processor, Graphite and clear, PCI graphics.

    G4 Sawtooth - no zif, Apple used a large D shaped white connector and used screws to stabilise the board that carries the processor. AGP graphics.

    G4 Mystic - same processor interface as Sawtooth, first of the DP series, with 450DP and 500DP. Also known as Gigabit ethernet.

    G4 V'Ger - 466, 533DP, 667, 733. Same as gigabit ethernet, with 133mhz system bus.

    G4 - Quicksilver - 733, 800DP, 867. Same as gigabit ethernet model with 133mhz system bus.

    G4's from Mystic or newer won't work on Yikes!.
    G4's from V-Ger and newer won't work on Mystic (133mhz bus versus the 100mhz bus...other differences..)

    XLr8.com sells a dual G4 upgrade card, but it only works on Yikes! and Yosemite (ZIF).

    I wish I had better details on where the modifier of the Cube got the extra processor card from a Mystic, and how he added the extra pad to match the dual processors. Was it just held in place with thermal goo and pressure from the retention of the heatsink?
  • by Anonymous Coward
    It is related to Cisc vs Risc but that is not the main reasoning. Risc and Cisc processor may run hot but Intel and AMD are not making Cisc processors but rather RISC-like procesors with MicroOps that allow them to convert x86 instructions into internal instructions, this extra baggage along with the units and added hardware so the performance is good brings up the heat. So it is like super fast to counter it's instruction set so Risc Processor certainly can be built cooler. Cisc does not exist anymore, complex ISAs exist but the processors themselves are not, straight up CISC simply cannot scale.
  • Mike has run xlr8yourmac for years with the same plain format. More websites should take their cue from him and keep it simple. Hyperlinks and simple graphics make for quick pageloads. His is an excellent example of a site with huge amounts of content and a simple format. You don't need massive loads of graphics to make a excellent site.
  • If it does get hot, will it self-repair the cracks and bubbles in the Cube's casing?

If you think the system is working, ask someone who's waiting for a prompt.

Working...