A look at the July Power Macs now that we know the Xserve specs

145791012

Comments

  • Reply 121 of 238
    [quote]Originally posted by Amorph:

    <strong>

    In that case, nothing will change anything, so you're also disagreeing with yourself.

    [ 05-16-2002: Message edited by: Amorph ]</strong><hr></blockquote>



    I am? I sure didn't think I was.



    From what I see, Apple's PowerMac sales have dropped 23%, and still, there is no forward momentum in their hardware development. Their sales must not have not dropped ENOUGH for them to take action. I guess it's a question of numbers.



    Remember Fight Club? If there aren't enough car accidents to be cost-impactive, the automobile manufacturer doesn't have a recall. It's a movie, I realize, but it describes American business practices to a tee.



    If Apple continues to make enough money on their iMac/iBook lines, they can effectively offset the losses from their PowerMac line with minimal impact.



    [quote]

    They have to work with what they have to work with. If the CPUs they have available are better suited to PowerBooks and iMacs than to high-end towers, Apple would be foolish not to concentrate its efforts on PowerBooks and iMacs.



    The reason I see their focus as component-driven, rather than as Steve forgetting power users, is that Apple is targetting higher end customers than they ever have.

    <hr></blockquote>



    I agree completely. I don't think that anyone is "forgetting power users". I'm simply suggesting that no one should blow good money on machines that are 1+ years behind the hardware curve.



    [quote]

    And (I can't resist asking) have you actually tried a dual-GHz PowerMac? Is the performance actually inadequate for your needs in practice? I'm not saying it isn't, I'm wondering if you've given it a chance. I have, and as far as I can tell the thing just blazes. As some of the engineers on this board have discovered, Apple's 133Mhz bus holds its own against some DDR implementations on the PC side. Quality of implementation matters.

    <hr></blockquote>



    Ask away! No, I have not used the 2xGig Powermac.



    I have used the 800MHz PowerMac, and concluded that $1600 is far too much to spend (for me, personally) on hardware that is outdated right out of the box.



    I agree that 512MB of 133MHz SDRAM can be as fast as 256MB of 266MHz DDR, all other things being equal. But I do not perceive Apple has having "quality" in mind when they made the decision to keep their 133MHz bus implementation. Rather, I imagine that they did it because they do not have a cost-effective/plausible implementation that would allow a 266MHz bus to be implemented for current PowerMac designs.



    And, yes, I believe that an 800MHz G4 PowerMac is beautifully-fast compared to my 350MHz G3 PowerMac. But that is completely irrelevant. My 1Gig Duron is beautifully-fast compared to my 350MHz G3, also. And it cost $1000 less.



    The REAL question is whether the 800MHz G4 is worth the money - now AND a year from now.



    [quote]

    And it's quite possible that people who look at nothing more than spec sheets do not realize that Apple isn't as far behind as they imagine. Spec sheets are marketing; they're designed to make the product look as good as possible. On the other hand, good engineers pride themselves on doing more with less. This is why there is no love lost between engineers and marketers, generally speaking.

    <hr></blockquote>



    I'm curious, do you work in the engineering field?



    [quote]

    Are they behind in some areas? Sure. Woefully behind in running a couple of professional applications (I'm thinking 3D here). Could Apple solve those problems by throwing hardware at them, even though some of them may be specific to the software? Of course. Hardware was eventually able to make even System V UNIX run fast. But I suggest sitting down in front of what they're offering now, if you haven't, and giving it a fair shot first.

    <hr></blockquote>



    How, exactly, could I give the 2x1Gig PowerMac a fair shot without shelling out $2999?



    [quote]

    Faster? Name one ATA HDD that comes anywhere close to transferring 66MB/s sustained. If Apple put ATA/100 on a PowerMac tomorrow without telling you, you'd never know they had. I expect Apple to leap up to ATA/133 soon, not for the added bandwidth, but because ATA/133 can address volumes bigger than 137GB, and hard drives are getting up there.

    <hr></blockquote>



    Whoa! Back up!



    What I said was that ATA/66 is outdated. I would challenge you to find a Dell, Gateway, or HP in the PowerMac price range that still used an ATA/66.



    And be careful about the word SUSTAINED. There is NO ATA HD out there (to my knowledge) that has a SUSTAINED data transfer rate. It's the same reason why Firewire wipes the floor with USB 2.0. A peak, unstable data rate of 800MB/s is no match for a sustained data rate of 400MB/s.



    ATA/100 has the potential to write at 100MB/s in random bursts. Are you telling me that there's NO software out there that would require 100MB/s bursts? Are you telling me that there's NO speed to be gained from 33MB/s MORE data throughput? Are you telling me that there WON'T be software out there that may require this within the next year?



    [quote]

    If you really want speed, forget the onboard controller and get <a href="http://www.macgurus.com/graphics/customraid.html"; target="_blank">one of these</a>.

    <hr></blockquote>



    Sorry, I don't see the need to pay $2999 AND shell out more cash for more speed. It's a personal preference, but the top of the line machine should offer top of the line performance, hands-down.



    Please don't misunderstand me, Amorph. I'm not saying that Apple Computer has crappy hardware. What I'm saying is that, for the money, they should utilize more powerful, longer-lasting hardware that is comperable with their price-point.



    I do consider SOME Apple merchendise worthy of a higher price-point. Apple's flat panels are gorgeous compared to what the PC world can offer. I would GLADLY pay more for them.



    Respectfully,

    -theMagius
  • Reply 122 of 238
    thttht Posts: 3,953member
    <strong>Originally posted by G-News:

    Like I said, SCSI was a good choice for the non PowerMac macs...then again, before 95 none or only very few configurations came with a drive bigger than 500MB anyway...</strong>



    Macs used a lot of media drives. SCSI was better at the time. In addition, the PPC 601 PowerMacs (7100 and 8100) used &gt; 500 MByte drives.



    <strong>But after 95, EIDE would have been the better, and most importantly MUCH CHEAPER, implementation for many customers.</strong>



    Agree here. But you hyperbolized by saying "always". For a time, Apple was using 5 MB/s SCSI while PC's mostly had IDE. And Fast SCSI v EIDE is very debateable.



    <strong>which, btw has a maximum transfer rate of 16.7MB/sec, not 13)</strong>



    PIO 3 or Multi-word DMA Mode 1 = 13.3 MB/s

    PIO 4 or Multi-word DMA Mode 2 = 16.6 MB/s

    Ultra DMA (ATA/33) = 33.0 MB/s



    <strong>And I wasn't trying to say that the Beige G3 was revolutionary for the industry, certainly not, but it was for Apple, and for the Mac market, and that counted.</strong>



    No, I wouldn't agree with that. If Apple wanted to be technologically superior, they had a lot directions they could have gone to, but decided to follow the pack instead of lead.



    And isn't it a logical contradiction. If Apple wants to be revolutionary, they should be better than the industry. Being revolutionary within there own market doesn't help them otherwise. The object of everyone's disappointment is that Apple isn't competing with PCs is it not?



    <strong>And btw, the G3 266 creamed the 604ev 350 in integer performance,and vice versa in FPU.

    For some apps, the G3 was the top level machine, not the mid range thing...even across platforms.</strong>



    Yes, very true, but we've been saying for a long while that if the 604e was the architecture Moto and IBM used for future development instead of the 603e, Apple could be better off. The addition of backside cache, improving the integer units the same way, et al, should have given a 604 based processor increasingly better performance as well. The 604 was 4-way superscalar and had a lot more buffers and such for concurrent execution. It had much more potential.
  • Reply 123 of 238
    thttht Posts: 3,953member
    <strong>Originally posted by jeromba:

    Do you really think that waiting for a Dual G4 1.4 Ghz with a FSB @ 266 Mhz it's silly ?</strong>



    It's borderline. A 0.13 micron 7450 based G4 can clock at 1.4 GHz, but I would expect the yield to be borderline for the first run. It may not be high enough to ship. A DDR bus, or something of the same bandwidth, would be the same sort of thing. I think it's waiting for a 0.13 micron product.



    The 7450 is a 7 stage execution pipeline architecture. It will always clock lower than the Athlon (10 stage), P3 (10 stage) and P4 (20 stage) on the same process. The only way a PPC processor will clock higher than the Athlon is to have a &gt; 10 stage pipeline processor on a clean fab or to fabricate it on a better fab. That's just the Athlon, Intel's P4 clock rate won't be overtaken, and we should not expect it to nor want it to.



    All G5 expectations are silly. There is no evidence of a G5 processor whatsoever. None. No descriptions of its archiecture. Not even a set of goals they are trying to reach. Any G5 statement is pure fantasy as far as I'm concerned.



    [ 05-16-2002: Message edited by: THT ]</p>
  • Reply 124 of 238
    rickagrickag Posts: 1,626member
    tiramisubomb

    <a href="http://www.corporate-ir.net/ireye/ir_site.zhtml?ticker=mot&script=411&layout=-6&item_id=164832"; target="_blank">http://www.corporate-ir.net/ireye/ir_site.zhtml?ticker=mot&script=411&layout=-6&item_id=164832</a>;



    This is from a Motorola press release the day before.

    [quote]



    Motorola Inc. (ticker: MOT, exchange: New York Stock Exchange) News Release - 9-Apr-2001

    ------------------------------------------------------------------------

    Motorola's 'HiP7' Ushers in Launch of Full System-on-Chip Development Environment for Tomorrow's 'Smarter' Products



    "Motorola is currently running embedded microprocessor cores on this advanced 0.13 micron process. Production is expected to begin in second quarter of this year in MOS13, Motorola's most advanced 8-inch facility in Austin, Texas. With the completion of this new process, Motorola continues to be an industry leader in copper process capabilities, building on the successes of two previous generations. "<hr></blockquote>



    I've done searches on Motorola's web site and still can not find a complete processor using HiP7. I also regularly check out their press releases, hoping for information about HiP7 and the G4 and G5.



    Granted, I know nothing of this technology and very easily could be missing Motorola parts being manufactured using the 0.13µ process other than embedded processor cores.



    Any links to product summary sheets would be appreciated. I would think the longer Motorola has been using HiP7 the closer this process becomes being implemented on a G4.



    Actually, I was very disappointed in Jan. when the G4 wasn't upgraded to HiP7 since they had had what 8 months manufacturing experience using HiP7 @ least on a minimum of embedded processor cores if not complete processors.
  • Reply 125 of 238
    rickagrickag Posts: 1,626member
    By the way, I tend to agree w/ THT that any expectations for a G5 within the next 12 - 18 months is wishful thinking.



    Maybe the delay in using HiP7 on a G4 is partly due to Motorola using the smaller die to extend the pipeline, or add/beef up the floating point, or incrreased on die cache, or wider piped altivec or or or or or



    crap I'm tired of waiting.
  • Reply 126 of 238
    stevessteves Posts: 108member
    [quote]Originally posted by crayz:

    <strong>Of course it should also be noted that the new G4s are already breaking the 200fps barrier on the toughest games when used with the high end graphics cards



    Are you joking? In Q3A a DP 1GHz w/ a Radeon 8500 or GF3 gets only 150FPS at 640x480. Q3A is not one of the "toughest games" either, and that limit appears to be because of the processor/bus, *not* the card. Try running RtCW on *any* Mac with *any* card and getting 200FPS with *any* settings....dumbass.</strong><hr></blockquote>



    Crazy,



    If you're going to call someone a dumbass, you should at least have your facts straight. You've just corrected someone for saying that Macs have now broken the 200fps barrier in the toughest games. You then go on to site Quake3 as one of those games.



    That said, I'd like to bring your attention to the following link:



    <a href="http://www.xlr8yourmac.com/Graphics/geforce4_ti/geforce4_titanium_Mac.html"; target="_blank">http://www.xlr8yourmac.com/Graphics/geforce4_ti/geforce4_titanium_Mac.html</a>;



    Please note the DP 1GHZ G4 on with Geforce 4 Ti achieving 201.3 fps at 640 x 480 in HIGH quality mode (32bit, trilinear filtering, 32bit textures, etc.).



    Also note that earlier graphics cards were not far behind on the same system.



    Further, these numbers are consistent with the 1.8 - 2 GHZ P4s with Geforce 3 Ti's I've seen in PC Gamer.



    In short, the orginal poster was correct. You should get your facts straight before you go on to call others a "dumbass", dumbass! <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />



    Steve
  • Reply 127 of 238
    rickagrickag Posts: 1,626member
    and another thing.



    While pondering THT's post concerning the 133 MHz MPX, it occurred to me that, at least in many Altivec aware programs, the current G4's keep up with and occasionally surpass the speed of X86 machine over twice the rated GHz speed.



    Me thinking... after the G4 crunches the numbers, it must pass the results throught the same 133 MHz MPX bus to the monitor or CD burner or DVD burner, right??



    Unless my reasoning is totally bassackwards, and the G4 733 truely does say, encode a DVD 70% faster than a 2000 GHz Sony Viao , whatever, this same 133 MHz MPX bus must be handling the data and can't be too awfully bad, can it??



    Please let me know if my reasoning is truely bassackwards.



    And if the G4's are really handcuffed on this measely little 133 MHz MPX bus, great things could be in store for a G4 @ 1.4 - 1.6GHz on a frontside bus capable of 266MHz.
  • Reply 128 of 238
    Does anyone know how a 3 DP Xserver workstation would compare to a DP 1GHZ G4? With everyday tasks, with MAYA complete, with Final Cut Pro? Just looking into making myself some sort of workstation.

    Thanks,
  • Reply 129 of 238
    lowb-inglowb-ing Posts: 98member
    Apropos the apple leak (or whatever it was) about 100% performance increase. I don't think there's even the slightest chance for 100% increase in clock rate. If the above "rumor" is true, I predict quads for MWNY. I guess that means either there's a new G4 coming soon, with a much faster bus, or apple has developed a system controller that lets the processors use individual buses (instead of sharing the same bus) while still letting them talk to each other (through said controller) about things like memory sharing asf. That way they could get away with using the current G4 design (though speed bumped, hopefully).

    I don't know alot about processor design, of course, so anyone who does is welcome to enlighten me on this subject.
  • Reply 130 of 238
    bellebelle Posts: 1,574member
    [quote]Originally posted by THT:

    <strong>[qb]All G5 expectations are silly. There is no evidence of a G5 processor whatsoever. None. No descriptions of its archiecture. Not even a set of goals they are trying to reach. Any G5 statement is pure fantasy as far as I'm concerned.</strong><hr></blockquote>

    It's always a pleasure to find a voice of reason when reading threads in Future Hardware.
  • Reply 131 of 238
    hornethornet Posts: 76member
    Well its obvious the sales haven't slipped enough for apple to do their homework. Their loss, we will get it eventually. The sales are expected to fall even further this quarter, and even more if nothing comes out in July that is decent (and 1.2ghz is NOT in any way, shape or form, decent, unless its a quad G5 at 1.2)





    FYI, on the topic of "have you even tried the new powermacs??". Yes, I have tried the 800 and 933. My thoughts?

    slow

    I wouldn't spend anymore than $1200 on the 933, with the performance I felt it had. It was SLOW at OSX. The only nice thing was the ASD17, beautiful. But beauty can't hide slowness, and what I felt, was OSX not a whole lot faster at all than my G4 933. Note I didn't do anything else other than IE, and finder stuff, and for what its worth, that performance would feel adequate for the bottom of the line ibook at best.



    really, it was quite shameful..... oh well.
  • Reply 132 of 238
    powerdocpowerdoc Posts: 8,123member
    [quote]Originally posted by Belle:

    <strong>

    It's always a pleasure to find a voice of reason when reading threads in Future Hardware. </strong><hr></blockquote>



    Yes and no : the G5 appear in the Motorola's roadmap since a long time. There is even a G6 on this roadmap. And the roadmap describe a 64 bit chip with I/O based upon 0,13 SOI process. It's not a precise info about the architecture, but the G5 is on the way ; the question is when : personally i won't bet a cent for G5 at MWNY 2002.



    [ 05-17-2002: Message edited by: powerdoc ]</p>
  • Reply 133 of 238
    g::mastag::masta Posts: 121member
    guys. .. just my 2c:



    the Xserve is something that Apple cannot stumble with. they have to be 110% sure that everything inside that 1U work of art will work 110^% of the time. EVERYTHING inside there has Apple's peace of mind stamp.



    IMO we might yet 266Mhz bus speeds, ATA-133, Firewire2, as for the chip, i fully expect 1.4Ghz, 512Mb L2 cache, Altivec2 (we can but hope). let's not forget that Apple consider the 4x AGP Radeon for their Xserve, so they gotta be pretty confident.



    Peace,

    G <img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
  • Reply 134 of 238
    crayzcrayz Posts: 73member
    If you're going to call someone a dumbass, you should at least have your facts straight. You've just corrected someone for saying that Macs have now broken the 200fps barrier in the toughest games. You then go on to site Quake3 as one of those games.



    Are you blind? I specifically said "Q3A is not one of the 'toughest games' either"



    And no, I am not all that impressed by the most expensive and powerful Mac ever created barely breaking the 200FPS barrier in Q3 only when using a tweaked config file.



    Tom's has an Athlon XP 2000+(a much cheaper system than a DP 1GHz) running Windows XP getting nearly 220FPS at 1024x768. That's the exact same card and the exact same game run with the exact same settings, and its getting over 35FPS more than the much more expensive PowerMac. That means that the PowerMac, either the CPU, the OS, or some other part of the system(like the AGP bus) is limiting the ability of the Ti4600 to do its job - BECAUSE THE MAC IS JUST TOO SLOW.



    And I repeat: Q3 is an old game, and hardly "one of the toughest" out there.
  • Reply 135 of 238
    moogsmoogs Posts: 4,296member
    [quote]Originally posted by powerdoc:

    <strong>

    Yes and no : the G5 appear in the Motorola's roadmap since a long time. There is even a G6 on this roadmap. And the roadmap describe a 64 bit chip with I/O based upon 0,13 SOI process. It's not a precise info about the architecture, but the G5 is on the way ; the question is when : personally i won't bet a cent for G5 at MWNY 2002.

    </strong><hr></blockquote>





    Not to pick on you PowerDoc, but I'm going to reiterate something I've said many times before: the Motorola "PPC Roadmap" IS NOT a legitimate source of evidence. Thus THT and Belle are correct in their assertions.



    The roadmap does not provide any useful information in terms of gauging what is actually happening with their PPC lineup. The roadmap was released years (not weeks) ago by some marketing schmo at MOT. Not by the head of SPS, not by their engineering team - by a marketing wonk. That's all you need to know in order to discern the "roadmap's" true purpose (to stoke user interest, not to inform).



    You don't actually think anyone at Apple uses the roadmap for anything but paper airplanes, do you? Companies who buy PPC products in large lots have all the inside information about what's coming in a few months and what isn't, because they are the ones who actually get to test those chips before we ever see them. The roadmap is for us, not for Apple or anyone making embedded systems - another sign telling you the document's true purpose.



    [ 05-17-2002: Message edited by: Moogs ]



    [ 05-17-2002: Message edited by: Moogs ]</p>
  • Reply 136 of 238
    powerdocpowerdoc Posts: 8,123member
    [quote]Originally posted by Moogs:

    <strong>





    Not to pick on you PowerDoc, but I'm going to reiterate something I've said many times before: the Motorola "PPC Roadmap" IS NOT a legitimate source of evidence. Thus THT and Belle are correct in their assertions.



    The roadmap does not provide any useful information in terms of gauging what is actually happening with their PPC lineup. The roadmap was released years (not weeks) ago by some marketing schmo at MOT. Not by the head of SPS, not by their engineering team - by a marketing wonk. That's all you need to know in order to discern the "roadmap's" true purpose (to stoke user interest, not to inform).



    You don't actually think anyone at Apple uses the roadmap for anything but paper airplanes, do you? Companies who buy PPC products in large lots have all the inside information about what's coming in a few months and what isn't, because they are the ones who actually get to test those chips before we ever see them. The roadmap is for us, not for Apple or anyone making embedded systems - another sign telling you the document's true purpose.



    [ 05-17-2002: Message edited by: Moogs ]



    [ 05-17-2002: Message edited by: Moogs ]</strong><hr></blockquote>

    Well i start my post with yes and no meanings that what they say was not perfectly true but was not wrong either . The roadmap is not the bible and is for consumer only , but in the past the MOTOROLA roadmap use to be nearly correct. It's not a big bet to say that the G5 will have IO tech since the 8450 chip have them (the reserach have be already done : they just have to put in their new desktop chip (or high end embedded chip if you prefer), for 64 bit i don't know : if the G5 come in one year may be yes, sooner may be no (there was a thread about this some times ago).



    So i think Mot will follow roughly the roadmap with one exception : if Apple design his new chip himself (let's call him the powerapple PPC chip ), in this case the next generation of PPC chip by mot is already there : it is the 8450 and all further developpements will be based upon this new architecture.



    I agree with Belle and THT for one thing : we will not see G5 tomorrow, if the day came we should be in 2003 and the G4 has not finish his developpement (DDR memory controller, and larger cache can be added without too much R&D. )

    BTW after this latest revision of the G4 , i don't see how Mot can improve this architecture with this old core : too short pipeline, old FPU unit .They have to go in a brand new architecture, like Intel has done with his P4.
  • Reply 137 of 238
    sonnyssonnys Posts: 4member
    The way people talk about their computer specs, you'd think they were talking about d*ck size or something.



    Numbers aren't everything. The reason OS X apps seem sluggish, especially web browsing and the like, is simply because the programs are not fully written to properly take advantage of OS X.



    There is going to be a long road ahead in ridding ourselves of Carbon apps (which, for all intents and purposes, are legacy apps) and transitioning not only to Cocoa but also in getting programmers up to speed on how to take best advantage of the services the OS has to offer.



    For all of you clamoring for the latest and greatest specs, go buy a PC. Those of you who don't feel defined by your possessions and can measure your computer purchase by more realistic yardsticks (productivity, ease-of-use, balanced system performance, user experience), keep your Macs.



    Macs have never had leading-edge processing hardware, and I doubt they ever will. Apple has no competitors, as it is the sole producer of hardware that runs Mac OS, and it will never feel compelled to play the Intel/Windows numbers game. Whatever you think you're giving up in processing power (which is a misnomer, as Macs are inherently faster for me because they allow me to do more work in less time) you are gaining in things like built-in Firewire, Gigabit ethernet, and a kick-ass OS.



    Playing the performance numbers game is small-minded. Look to the bigger picture of workflow, troubleshooting, and productivity, and you'll see that Macs win hands-down at any speed.
  • Reply 138 of 238
    amorphamorph Posts: 7,112member
    TheMagius wrote:



    [quote]<strong>I am? I sure didn't think I was.</strong><hr></blockquote>



    Well, you said that people needed to stop buying PowerMacs, I replied that people needed to stop buying them and tell Apple why, and you said that would be ineffective. Ergo, not buying PowerMacs is ineffective.



    [quote]<strong>From what I see, Apple's PowerMac sales have dropped 23%, and still, there is no forward momentum in their hardware development. Their sales must not have not dropped ENOUGH for them to take action. I guess it's a question of numbers.



    Remember Fight Club? If there aren't enough car accidents to be cost-impactive, the automobile manufacturer doesn't have a recall. It's a movie, I realize, but it describes American business practices to a tee.</strong><hr></blockquote>



    I don't get my sense of economics from Hollywood films, and since I saw nothing about Fight Club to recommend it, I ignored it.



    Remember that Apple is aggressively moving into markets that cannot subsist on consumer products or laptops, and they are trying to grow market share. In both these cases, subsisting on increased sales of their other lines is a stopgap solution at best. It might be what they're doing now, but I'll bet it's not because they want to.



    [quote]<strong>No, I have not used the 2xGig Powermac.</strong><hr></blockquote>



    It's about 30% faster than the dual 800, to give you some idea. (That based on xlr8yourmac's tests.)



    [quote]<strong>I agree that 512MB of 133MHz SDRAM can be as fast as 256MB of 266MHz DDR, all other things being equal.</strong><hr></blockquote>



    Busses and memory controllers being unequal (as they often are), 256MB of SDR SDRAM can outrun 256MB of DDR, or at least keep up in the majority of cases. MaxBus has robust support for streaming.



    [quote]<strong>But I do not perceive Apple has having "quality" in mind when they made the decision to keep their 133MHz bus implementation. Rather, I imagine that they did it because they do not have a cost-effective/plausible implementation that would allow a 266MHz bus to be implemented for current PowerMac designs.</strong><hr></blockquote>



    What part of "quality" implies that they should use an implementation that isn't plausible? :confused:



    They've eked an astonishing amount of performance out of their current architecture, far more than most - if not all - PC motherboards ever got out of their single-pumped busses. If they switched to DDR, and the net result was a single-digit performance improvement sometimes, Apple would be laughed off the boards. They implemented DDR for their servers, despite the 133Mhz MaxBus, because it made sense to have all that extra memory bandwidth for I/O; now that they have a memory controller that can use DDR efficiently, they'll implement DDR in the towers when they have a processor that supports it. To claim that they're holding back because the iMac is selling well is just strange.



    [quote]<strong>And, yes, I believe that an 800MHz G4 PowerMac is beautifully-fast compared to my 350MHz G3 PowerMac. But that is completely irrelevant. My 1Gig Duron is beautifully-fast compared to my 350MHz G3, also. And it cost $1000 less.</strong><hr></blockquote>



    Might as well ask another question: Is processor speed the only item of concern for you? Or hardware speed, for that matter? Do you mean real-world performance or numbers on spec sheets?



    BTW, you can check out a dual GHz by going to a VAR or an Apple Store, if there's one around. That's how I did it. The folks in the Apple Store load them up with pro apps, and they're more than willing to let you take them for a spin.



    [quote]<strong>The REAL question is whether the 800MHz G4 is worth the money - now AND a year from now.</strong><hr></blockquote>



    My 450MHz G4 was worth the money when I bought it last year, and I still haven't come close to tapping its potential. I suppose it depends on what you're actually trying to do.



    [quote]<strong>I'm curious, do you work in the engineering field?</strong><hr></blockquote>



    In software, which has the same relationship to marketing that the hardware guys do. Plus a great many of my friends and coworkers are hardware guys. The "more with less" doctrine is more especially critical in hardware, especially where clock speed is concerned, because high clock speeds introduce a double handful of complications to motherboard design. But it's not uncommon in software, either. In fact, on the programming newsgroups, someone with a loaded, up-to-the-minute rig is most likely an abject newbie, and his uncapitalized "how do i clear the screen in c k thx bye" is probably going to be handled by someone on an 8500.



    [quote]<strong>What I said was that ATA/66 is outdated. I would challenge you to find a Dell, Gateway, or HP in the PowerMac price range that still used an ATA/66.</strong><hr></blockquote>



    Sorry, but what you said is that you wanted something faster. If Dell, Gateway and HP ship those... so what? The theoretical speed advantage they offer is irrelevant, and they offer no increase in storage capacity, unlike ATA/133. So we're back to real performance vs. theoretical numbers.



    When I wanted more space, I bought an ATA/133 FW enclosure. I have my original 30GB drive in there now, but if I wanted to add a whopping 160GB drive tomorrow I could, and bigger drives as they become available. I can also buy another enclosure and daisy-chain it for even more storage. The FW drive is every bit as fast as the internal drive, so there's no loss there either.



    [quote]<strong>And be careful about the word SUSTAINED. There is NO ATA HD out there (to my knowledge) that has a SUSTAINED data transfer rate. It's the same reason why Firewire wipes the floor with USB 2.0. A peak, unstable data rate of 800MB/s is no match for a sustained data rate of 400MB/s.



    ATA/100 has the potential to write at 100MB/s in random bursts. Are you telling me that there's NO software out there that would require 100MB/s bursts?</strong><hr></blockquote>



    Whether there is or not is moot if there isn't the hardware to provide those bursts.



    Also, I doubt that there is. There is software that benefits from it, of course, which is why people buy big SCSI RAIDs like the ones that I linked to, but nothing I've heard of that requires it. Real-time video, perhaps? Off to MacGurus you go, because you're not going to get it from your ATA bus.



    [quote]<strong>Are you telling me that there's NO speed to be gained from 33MB/s MORE data throughput?</strong><hr></blockquote>



    Maybe a fraction of a second here or there reading relatively small files from RAM cache (which OS X does to some extent anyway by caching files in main RAM, bypassing the drive bus altogether), but nothing like the 33% increase in theoretical bandwidth. You won't get that until sustained reads from ATA drives get faster (what's a sustained read from an ATA drive? Well, copy a 1GB file on an optimized disk. There you go. ).



    [quote]<strong>Are you telling me that there WON'T be software out there that may require this within the next year?</strong><hr></blockquote>



    With almost 100% certainty. Any software that failed to run without being able to access an ATA HDD at that rate (i.e., which required that bandwidth) would find itself with a vanishingly small market. People who really need that kind of speed fork out for SCSI RAIDs. That's where your read/write rates of 100MB/s are, and you'll pay dearly for them.



    It's possible that Apple will introduce something like the ATA RAID technology on the XServe, although I'll bet they have two channels instead of four if they do. Even then, people who want real speed will get a SCSI RAID or an XServe RAID (when those appear). If disk performance is critical, you don't under any circumstance rely on a single ATA drive to provide it. It's just never done.



    [ 05-17-2002: Message edited by: Amorph ]</p>
  • Reply 139 of 238
    keyboardf12keyboardf12 Posts: 1,379member
    Well said sonny.



    Also, as avi said the compilers have not been optimzed for OSX. With jagwire we will see all apps get an almost free speed increase with a recompile.



    I don't own the fastest g4 on the planet, But I would put up the work toe to toe I do with it with any PC geek that justs buy the latest 2 gig pc so he can play quake faster.



    Sure there are some PC users who do work that needs the mhz but pound for pound and mhz. for mhz I'll bet mac users do far more creative (meaningful?) work than any gighz crazed pc weenie.



    [ 05-17-2002: Message edited by: keyboardf12 ]</p>
  • Reply 140 of 238
    g-newsg-news Posts: 1,107member
    What you're forgetting about those burtst rates, is that bursts are generally a thing between the HD cache and the system RAM on an ATA bus.



    And I'm sorry to tell you, but even the now 2 year old IBM 75GXP series had internal transfer rates of up to 400MB/sec.



    That'd fill you every inch of the ATA-133 bus for every bursted data that is smaller or equal to the cache size of the HD, thus generally around 2MB today, maxing out at around 8MB.



    Problems is, only few apps burst to such an extent that they'd gain speed from faster ATA busses.

    Coz if you do a lot of bursting in a row, it's no longer bursting, it's simply reading and then you're back at the sustained rates.



    So you'd probably benefit if you had to copy a folder with many small files around 100-200KB in size, coz that'd mostly be done out of the cache.



    Denying that ATA133 is faster than ATA 66 or 100 is just stupid. The differences are small for most tasks, but it's not like Maxtor just put a "133" sticker on their ATA 66 drives.



    And yeah I think faster is better, especially because you obviously don't lose anything in the trade, you win on both size and speed.



    However, the real reason why were still stuck at ATA-66 in the towers, is because we're still using the same Southbridge since the G4 Sawtooth was released.



    And southbridges usually don't get changed without changing the whole rest of the mainboard too.



    There you go.



    G-news
Sign In or Register to comment.