or Connect
AppleInsider › Forums › Mac Hardware › Future Apple Hardware › Finally an interesting G5 story
New Posts  All Forums:Forum Nav:

Finally an interesting G5 story - Page 7

post #241 of 441
[quote]Originally posted by spooky:
<strong>All I care about is that apple produces a killer system based around a killer processor. From what I'v read (and I am by no means an expert at all), the 970 sounds good but in no way does it sound like a killer processor.</strong><hr></blockquote>

The devil's in the details. For what I want, the 970 has all the details I was looking for: dramatically faster FSB, better SMP support, a second FPU, better Sprecmarks... without being out of reach for a desktop machine in price or heat.

When you look at spec chart comparisons between the IBM970, G4, AMD, Xeon, and P4, and realize just how bloody far down the list the G4 currently is, it is pretty clear that this isn't just 'keeping up with the Joneses'. It won't _pass_ anybody necessarily - but it seems like a tremendous first step. 5x speed on a particular task in a year isn't something to sneeze at.

<a href="http://www.extremetech.com/article2/0,3973,635220,00.asp" target="_blank">ExtremeTech</a> has the ppc970's estimated specs, but I don't see the chart I was looking for right this second. It showed the current Xeons & Athalons as the only things in the same ballpark with the 970. And with the list stopping at the top 20 chips or so - the G4 isn't even on the list!

The other thing to note is that there are still applications for which the G4 is one of the top chips at - even though there's a slew of benchmarks at which it sucks! AV is the only thing that has kept the G4 anywhere _near_ the pack - and the IBM970 has both raw performance in int & fp tasks - plus the full AV engine. Plus MERSI. Plus a reasonable power budget.

And Apple has been buying out companies that need a LOT more power than the G4. Either a) REALLY dumb, b) there's a plan.
post #242 of 441
b.) There's a plan.
"People don't want handouts! People want hand jobs!" ~ Connecticut governor William O'Neil at a political rally, followed by riotous applause
Reply
"People don't want handouts! People want hand jobs!" ~ Connecticut governor William O'Neil at a political rally, followed by riotous applause
Reply
post #243 of 441
[quote]Originally posted by Algol:
<strong>b.) There's a plan.</strong><hr></blockquote>

...and hopefully not Plan a)
I heard that geeks are a dime a dozen, I just want to find out who's been passin' out the dimes
----- Fred Blassie 1964
Reply
I heard that geeks are a dime a dozen, I just want to find out who's been passin' out the dimes
----- Fred Blassie 1964
Reply
post #244 of 441
[quote]Originally posted by Telomar:
<strong>I could spend an entire day in here going over any number of flaws in your posts, which are for the most part full of opinionated nonsense. You show little actual grounding in the subjects at hand or underlying knowledge of the market. In fact if I had to hazard a guess I'd say you are a student fresh out of some form of business course. You like to throw a lot of theory about but you really don't seem to have a true grasp or understanding on what it exactly means.
</strong><hr></blockquote>

Rather than attack my ideas, you attack my credibility. So how credible does that make you?
When you're ready to take me to school and point out my flaws, I'm ready. But until then, please leave my mother out of this.



[quote]<strong>
IBM's name is such {snip}

In case you aren't familiar with this market, and judging by what you say you quite clearly aren't, reliability is key. Around half the price of your payment go in support costs. People don't pick IBM in that market because they sell low quality products, they buy IBM because it is IBM and IBM has built up a name for quality and reliability in the industry. At some point you would have likely heard the old saying, "You won't get fired for suggesting IBM." and it still remains very true.

</strong><hr></blockquote>

That's some great PR.

When you use the word market, which market are you talking about? Are you talking about the server market? "High-end solutions?"

IBM has an estimated 13.9% of the worldwide server market, and an estimated 11.7% US. Obviously IBMs "name for quality and reliability in the industry" doesn't really help it to take control of DELL & HP, who are the industry leaders.

[quote]<strong>
As for IBM not being at the cutting edge of R&D that's nonsense. Every single year IBM makes more patent applications than any other company (NEC is second from memory). </strong><hr></blockquote>

This has already been disputed. Read above for clarifcation thanks.

[quote]<strong>
As a business proposition I couldn't think of a more appealing company than IBM and the markets generally would agree with that. They're a stalwart icon of the technology and computing industry and a good option for Apple if IBM chooses.</strong><hr></blockquote>

Considering you've questioned my credibility based upon my opinions you now expect others to accept yours instead. How should I go about finding new ways to insult your intelligence as well?

post #245 of 441
[quote]Originally posted by Amorph:
<strong>Nitpick: MOTU is Mark of the Unicorn, a developer of music production hardware and software that has been a Mac stalwart for years. Motorola is Mot, or Moto.
</strong><hr></blockquote>

<img src="graemlins/lol.gif" border="0" alt="[Laughing]" />

Totally. My bad.


I use Cubase.

<img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
post #246 of 441
[quote]Originally posted by Algol:
<strong>b.) There's a plan.</strong><hr></blockquote>

Seems like it. Too many disparate pieces pointing to something unknown. But with an interesting shape. Shrug.
post #247 of 441
[quote]Originally posted by Amorph:
<strong>
As for x86 being a standard platform, you apparently haven't yet understood that I don't consider it a standard platform in any meaningful sense. I called it a common platform, which is altogether different. </strong><hr></blockquote>

That's terrific. You can call an apple an orange if you like, still doesn't change the way it tastes. If you still don't consier x86 to be the standard platorm with any amount of evidence presented before you, then there's no use continuing this conversation.
post #248 of 441
[quote]Originally posted by Amorph:
<strong>
No, but you have tellingly made no effort to find all the work they've done that is.
</strong><hr></blockquote>

BTW--It's not my job to support another person's argument. I work with what is provided to me like most people do. If people want to make their cases stronger by supporting it with evidence then by all means do so. But don't expect me to do your homework for you.

<img src="graemlins/oyvey.gif" border="0" alt="[No]" />
post #249 of 441
[quote]Originally posted by Amorph:
<strong> As I've said repeatedly, this is the future of high-performance computing according to the organizations that do high-performance computing. IBM is going that way. Intel isn't. So if you want high performance, why go Intel?
</strong><hr></blockquote>

We're talking about personal computers. That's why.

<img src="graemlins/oyvey.gif" border="0" alt="[No]" />
post #250 of 441
[quote]Originally posted by MacLuv:
<strong>

That's terrific. You can call an apple an orange if you like, still doesn't change the way it tastes. If you still don't consier x86 to be the standard platorm with any amount of evidence presented before you, then there's no use continuing this conversation.</strong><hr></blockquote>

And if you can't entertain the possibility that x86 is not a standard in any useful or meaningful way despite any amount of evidence, there's no point continuing the conversation.

(Not to mention that if you don't even know that IBM research is legendary, there's no point starting the conversation. But it's a bit late for that...).

[ 12-05-2002: Message edited by: Amorph ]</p>
"...within intervention's distance of the embassy." - CvB

Original music:
The Mayflies - Black earth Americana. Now on iTMS!
Becca Sutlive - Iowa Fried Rock 'n Roll - now on iTMS!
Reply
"...within intervention's distance of the embassy." - CvB

Original music:
The Mayflies - Black earth Americana. Now on iTMS!
Becca Sutlive - Iowa Fried Rock 'n Roll - now on iTMS!
Reply
post #251 of 441
just wait until next month.
everyone is saying that apple will use the alleged
7557 chip in new POWERMACS.
first of all we dont know that update or all new POWERMACS wont be appearing at mwsf2003.
secondly,if apple does decide to go with the 7557 it will debut at more than 1.3 ghz,which is not what most talking heads are saying.
third,motorola WILL be rolling out their embedded processors based on the book E specification.
the 8540 and 8560 both have rapid i/o and support for 333 mhz ddr ram built into the cpu!
these chips are quasi-G5's.
the only thing they lack is altivec.
how hard would it be for motorola to put altivec in these chips?
not difficult at all me thinks.
then apple could be in business.
these chips arent true g5's but would suffice for the i-book and i-mac.
AND even though im not aware of pricing on these items,motorola's own website is touting them as a price performance winner.
if you look at the literature on the 8540 which is readily available on motorolas website,these chips can adress 4Gigabytes of ram EACH,and they also have what is described as a "point to point interconnect",OCEAN.
isnt apple PI a point to point interconnect architecture?
from what ive gleaned its based on RAPID I/O and not HYPERTRANSPORT.
im not aware of the architectural particulars in terms of the differences between these 2 competing I/O archetectures,i would appreciate a little feedback.oh yeah...they consume 6.5W of power at 1GHZ!!
also multiprocessor designs will be much easier to implement using these chips.
but look out!
motorola WILL make its presence felt,and very soon! <img src="graemlins/surprised.gif" border="0" alt="[Surprised]" />
post #252 of 441
I wouldn't take the current state of the G3 as any indication of IBM being lacking in anything. What motivation have really had to improve on it in a significant fashion? It works great in Apple Portables and embeded applications such as the Game Cube, because it's a damn efficiant chip for those situations. The idea that that immediately makes the 970 a turkey with no future is rediculous.

IBM has been pretty focused on servers and ecommerce recently and it took a while to develop the 970. The time is just about right for a mature Linux desktop enviroment to emerge whcih is something that IBM could make some good money on. I'm sure they feel that 2003 is the right time to focus on desktop machines and Apple using the chip as well is just another source of revenue.

I thought the eMac used a G4. What does that have to do with IBM?
post #253 of 441
[quote]Originally posted by geekmeat:
<strong>just wait until next month.
everyone is saying that apple will use the alleged
7557 chip in new POWERMACS.
first of all we dont know that update or all new POWERMACS wont be appearing at mwsf2003.
secondly,if apple does decide to go with the 7557 it will debut at more than 1.3 ghz,which is not what most talking heads are saying.
third,motorola WILL be rolling out their embedded processors based on the book E specification.
the 8540 and 8560 both have rapid i/o and support for 333 mhz ddr ram built into the cpu!
these chips are quasi-G5's.
the only thing they lack is altivec.
how hard would it be for motorola to put altivec in these chips?
not difficult at all me thinks.
then apple could be in business.
these chips arent true g5's but would suffice for the i-book and i-mac.
AND even though im not aware of pricing on these items,motorola's own website is touting them as a price performance winner.
if you look at the literature on the 8540 which is readily available on motorolas website,these chips can adress 4Gigabytes of ram EACH,and they also have what is described as a "point to point interconnect",OCEAN.
isnt apple PI a point to point interconnect architecture?
from what ive gleaned its based on RAPID I/O and not HYPERTRANSPORT.
im not aware of the architectural particulars in terms of the differences between these 2 competing I/O archetectures,i would appreciate a little feedback.oh yeah...they consume 6.5W of power at 1GHZ!!
also multiprocessor designs will be much easier to implement using these chips.
but look out!
motorola WILL make its presence felt,and very soon! <img src="graemlins/surprised.gif" border="0" alt="[Surprised]" /> </strong><hr></blockquote>

1) The 8540 and 8560 do not have floating point units. (The 8560 can do single precision FP in it's SIMD unit, but that's not the same at all.) Every PPC chip used by Apple has had a full double precision FPU, so a lot of programmes will have FP instructions in there without bothering to test for an FPU, and will crash on one of these processors.

2) Altivec is incompatible with BookE as it has its own set of 128 bit registers, and uses prefetch instructions which would require redesign of the load/store units of the BookE CPUs. It would be very hard to retrofit Altivec to the 8560, that's why it has it's own SIMD instruction set.

3) OCEAN is an on-chip interconnect (That's what the OC stands for). The external interface is RapidIO, which is point-to-point. RapidIO and Hypertransport are similar in concept, though RapidIO is currently a parallel bus raher than serial.

4) Noone has yet come up with any hard information on what ApplePI is. There is no certainty that it is point-to-point.

5) The 8540/8560 are barely if at all faster than an equally clocked G4 at integer processing, and are expected to not exceed 1GHz at release, hardly a G5.

6) The 8540/8560 have been promised for quite a while, with no indication that they are anywhere near being delivered.

7) As you say, there is a lot of literature at Motorola's site, try reading it.

michael
Sintoo, agora non podo falar.
Reply
Sintoo, agora non podo falar.
Reply
post #254 of 441
[quote]Originally posted by mmicist:
<strong>
4) Noone has yet come up with any hard information on what ApplePI is. There is no certainty that it is point-to-point.
</strong><hr></blockquote>

I have evidence that ApplePi is intimately related to the 970's bus.
Providing grist for the rumour mill since 2001.
Reply
Providing grist for the rumour mill since 2001.
Reply
post #255 of 441
[quote]Originally posted by MacLuv:
<strong>

We're talking about personal computers. That's why.

</strong><hr></blockquote>

Oh, that's right. I forgot. All personal computers will always use one big, hot processor. There's no way any PC maker will ever go SMP. That strategy only works when you want high performance, and nobody wants high performance in a personal computer.

Do you work for Intel?
"...within intervention's distance of the embassy." - CvB

Original music:
The Mayflies - Black earth Americana. Now on iTMS!
Becca Sutlive - Iowa Fried Rock 'n Roll - now on iTMS!
Reply
"...within intervention's distance of the embassy." - CvB

Original music:
The Mayflies - Black earth Americana. Now on iTMS!
Becca Sutlive - Iowa Fried Rock 'n Roll - now on iTMS!
Reply
post #256 of 441
from Cube Owner:
[quote]PowerLogix has sent an update regarding their previously reported plans for a more robust DC/DC board replacement. Here's what Robert jagitsch, President of PowerLogix had to day:

"We have come up with an optimized method that allows the use of the standard Apple DC/DC card, so that card will not need replacing. We expect the duals for Cube to be available by the end of January. We intend to offer dual 800 and faster for Cube, including 1.2GHz." <hr></blockquote>

It sounds like Moto has moved the G4 to a smaller manufacturing process, and that they will have sufficient 1.2 Ghz chips sometime after January to supply Apple, and the upgrade manufacturers....Good news for me and my Cube.
post #257 of 441
A couple of these threads seem to prove what a strong hold marketing can have over the human mind. I'm sure many are persuaded that a processor's core performance is the most important factor, especially clock rate in GHz. It is no longer the MHz race. Well, it has produced a lot of good replies and information, and caused me to consider a couple things.

There really is a limit to how fast a single core can go, and and how much performance can be squeezed out with techniques like SMT. It looks like Intel is determined to discover just where that limit is and get as close to it as possible. That's one approach. It appears that nothing else will satisfy those who are completely sold on the current marketing hype.

However, it looks like IBM has a different approach. Single core performance is just one factor in computer performance. Why push the core beyond the point of diminishing returns, where single core performance begins to cost more and more for less and less improvement? IBM appears to be saying it is better to get good performance from a core, while keeping cost and power at lower levels. In this approach, increased computer performance comes from having more cores.

If we wish to argue, let it be about which limit will be reached first? Will single core performance hit the wall before adding more cores becomes impractical, or will it be the other way around? Which approach will be most economical for getting high end computers? I do believe some of this has been addressed already, but it will never satisfy those who are completely sold on core specs.

Excuse the double posting. It looks like my post is more appropriate in this thread. Also to clear up one issue -- to have more than one core, I also include dual, quad and more single core processors. I am not just referring to multi-core chips.
post #258 of 441
[quote]Originally posted by Bigc:
<strong>

...and hopefully not Plan a)</strong><hr></blockquote>

<img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
go AAPL, go to $70 !!! © 2004
Reply
go AAPL, go to $70 !!! © 2004
Reply
post #259 of 441
[quote]Originally posted by Programmer:
<strong>

I have evidence that ApplePi is intimately related to the 970's bus.</strong><hr></blockquote>
where from?
go AAPL, go to $70 !!! © 2004
Reply
go AAPL, go to $70 !!! © 2004
Reply
post #260 of 441
<a href="http://www.siliconstrategies.com/story/OEG20021204S0041" target="_blank">Very interesting</a>, especially the 3rd paragraph.
post #261 of 441
hi MacLuv.

intel is moving from x86 to itanium 2.
amd is using a RISC-based chip which cracks the x86 instructions before processing.
IBM will build the worlds fastest computing-station ever with 120000 Power5 cpus (10 times faster than todays fastest).
the 970 is based on the Power4.
970s-follow-ups will be based on the Power5.
Mac OS X is smp-capable.....
the G4 isn't THAT much slower than a 3Ghz P4 or AMD 2.6+ (or is there a 2.8+ already?).
the G4 shows up in SPEC-marks up to 5 times slower than the 970.
SPEC is NOT a good benchmark to compare PPC and x86.
and please stop using cubase and start using Logic Audio - it's so much better!
AMD is having problems to stay in market and looks for finding new markets (so they're not basing all their future-plans on their x86 cpus).
apple bought some high-performance-killer-apps of the movie-and-music-market.
amd says: currently no chips for apple.
intel P4 is slower on the same clock than a P3 - the 970 is faster at the same clock than a G4 - the G4 is faster at the same clock than a P4 - who made progress?
ibms roadmap shows the G3-successor with RapidIO and SIMD and SMP-capable and other goodies...

me thinks an IBM-Apple-machine is the future! and a good future!
go AAPL, go to $70 !!! © 2004
Reply
go AAPL, go to $70 !!! © 2004
Reply
post #262 of 441
[quote]Originally posted by Krassy:
<strong>
where from?</strong><hr></blockquote>

I second that request.
just waiting to be included in one of Apple's target markets.
Don't get me wrong, I like the flat panel iMac, actually own an iMac, and I like the Mac mini, but...........
Reply
just waiting to be included in one of Apple's target markets.
Don't get me wrong, I like the flat panel iMac, actually own an iMac, and I like the Mac mini, but...........
Reply
post #263 of 441
[quote]Originally posted by Programmer:
<strong>

I have evidence that ApplePi is intimately related to the 970's bus.</strong><hr></blockquote>

Is that intimately in terms of being screwed together or just huggy-kissy.
<img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
I heard that geeks are a dime a dozen, I just want to find out who's been passin' out the dimes
----- Fred Blassie 1964
Reply
I heard that geeks are a dime a dozen, I just want to find out who's been passin' out the dimes
----- Fred Blassie 1964
Reply
post #264 of 441
[quote]Originally posted by MacLuv:
<strong>That IBM is cutting edge is opinion, not fact. (That would hardly get one a job if put on a resume.)</strong><hr></blockquote>

I am not applying for a job, you dick, and who else is spending good money on R&D for molecular switching in CPUs? AMD? Intel? Hell no!

[quote]<strong>That IBM has millions to invest in R&D is opinion, not fact.</strong><hr></blockquote>

Sorry, billions, I made a mistake - look at their <a href="http://biz.yahoo.com/fin/l/i/ibm.html" target="_blank">financial statement</a> before blatantly spewing forth such crap! They are investing over 1.5 billion per quarter. AMD....170 million. Hmmmm.

[quote]<strong>Your opinion of their workstatios and servers is irrelevant in a conversation about consumer PCs and CPUs</strong><hr></blockquote>

Why? That's the CPU they're putting in the next Powermac's! PowerPC 970's are based on a Power4 CPU that exists, wait for it....in their servers and workstations!!! Hey, I guess your statement shows how deaf, dumb and blind you really are. Goodbye, Tommy!

[ 12-05-2002: Message edited by: Rhumgod ]</p>
...we have assumed control
Reply
...we have assumed control
Reply
post #265 of 441
[[And if you can't entertain the possibility that x86 is not a standard in any useful or meaningful way despite any amount of evidence, there's no point continuing the conversation.]]]

Exactly, Amorph...

I pointed that out in another post. It's called "Slothful Induction". Just look at how he words things and just look at how often he posts rebuttals. He certainly has a lot of time on his hands.

MacLuv says that x86 is the industry standard... but saying it a million times doesn't make it so.

He asks for substantial proof that x86 ISN'T the industry standard, yet the he hasn't provided the proof that it *is*...

He made the statement that x86 is the industry standard -- The burden of proof lies with him.

He likes to make comments and statements and then "shifts the burden of proof." It's typical language acrobatics like this that he believes he's good at.

That said, we all know that he takes pride in slothfully inducing the material at hand. As for the other fallacies that he employs, let's expose this individual and maybe get a clearer understanding of his character and purpose on these forums...

As mentioned earlier, "Shifting the burden of proof" is a way of demanding that the person denying an assertion prove his/her case, however the burden of proof is upon the person who argues (or is arguing) the position.

Let me explain... MacLuv stated that the x86 is the Industry Standard platform, processor, ISA, whatever... He staked the claim, he made/passed the comment. The rest of the forum contributors are the ones denying that assertion of x86 being the standard. Therefore it is *he* who has to supply *us* with all the concrete proof and evidence supporting *his* claim, which is an "accident" (another fallacy) to begin with. Once everyone picks up on this I think he has little chance of being taken seriously on these boards.

Programmer is probably the only one wise enough to ignore him.

Anyway, another tactic that MacLuv likes to employ is that of "Special Pleading". He expects *us* to track and make note of our own *supposed* errors in logic while at the same time shifting the burden of proof onto us, when he has yet to do so for himself. In short, Special Pleading is refusing to apply the same principles to oneself that one applies to others.

Given his previous posts, it's likely that he doesn't have the ability to even realize that he's committing fallacy after fallacy. The very things he's attempting to pelt *us* with. It's another fallacy in fact; It's called the Fallacy of opposition -- since we disagree with him, we must not be thinking straight or reasoning things out properly.

He also like to employ what essentially is the "Appeal to numbers" or majority, or popularity: asserting that the acceptance of an idea by a majority, or by a large number of people, is reason to believe it.

In this case his claims about x86. After all, x86 is only a small piece of the equation. The embedded market for example, is significantly larger than the *desktop* computer market. He ignores all other processors in all other markets other than x86 and the *desktop*. It's called "cherry picking" or more correctly, he's making broad/sweeping generalizations and presenting it as the *rule*.

He also likes to employ "Humor and ridicule" to support his baseless arguments. Just look at all the sarcasm, all those "emotion faces" and other colorful remarks he posted throughout these forums. He uses these to avoid the issue at hand and to cast unwarranted aspersions and to deflect attention away from the discussion. Then there is his use of "rationalization" and the tactic of spewing "Nothing but objections".

At nearly every instance, he's continually raising objections as a means of avoiding the issue. -- which is to provide us with the absolute, indisputable proof that x86 is the Industry standard. With these thing in mind, all of you can now draw your own conclusions about MacLuv, his character and his true intentions.

Perhaps we should all follow Programmer's example and ignore this individual, simply because he isn't bringing anything constructive to the conversation and only causing a massive irritation which is only resulting in a thread-war.

--
Ed M.
post #266 of 441
[quote]Originally posted by Ed M.:
<strong>[[And if you can't entertain the possibility that x86 is not a standard in any useful or meaningful way despite any amount of evidence, there's no point continuing the conversation.]]]

Exactly, Amorph...

[...]

Programmer is probably the only one wise enough to ignore him. </strong><hr></blockquote>

Perhaps. But I appreciated the opportunity to think through the argument anyway. That's why all my posts on this topic are so damned long.

[ 12-05-2002: Message edited by: Amorph ]</p>
"...within intervention's distance of the embassy." - CvB

Original music:
The Mayflies - Black earth Americana. Now on iTMS!
Becca Sutlive - Iowa Fried Rock 'n Roll - now on iTMS!
Reply
"...within intervention's distance of the embassy." - CvB

Original music:
The Mayflies - Black earth Americana. Now on iTMS!
Becca Sutlive - Iowa Fried Rock 'n Roll - now on iTMS!
Reply
post #267 of 441
[quote]Originally posted by Amorph:
<strong>

Perhaps. But I appreciated the opportunity to think through the argument anyway. That's why all my posts on this topic are so damned long.

[ 12-05-2002: Message edited by: Amorph ]</strong><hr></blockquote>

and for me the only thing that counts is to have a 970-based mac os x-machine on my desk as early as possible instead of this x86-horror-imagination-mac-combination-or-whatever-this-will-lead-us-to *ergs* <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
go AAPL, go to $70 !!! © 2004
Reply
go AAPL, go to $70 !!! © 2004
Reply
post #268 of 441
[quote]Originally posted by snoopy:
<strong>

. . . There really is a limit to how fast a single core can go, and and how much performance can be squeezed out with techniques like SMT. It looks like Intel is determined to discover just where that limit is and get as close to it as possible. . .

</strong><hr></blockquote>

Hey, I think I'll reply to my own post with another thought. Maybe Intel is near the limit already, no? I should have also stated that the limit will automatically go higher as process size goes down, but the benefit of smaller process size is seen by everyone, even if they follow a different strategy. To get top performance, IBM is pursuing ease of adding more cores or processors, and Intel is trying to push the performance of a single core, at any cost it seems.

Anyway, Intel may have nearly maxed out, where the only way to increase performance is to shrink the process. IBM and Apple can increase performance a long way by adding more processors. Of course shrinking the process size takes both sides to a higher level. The only question is the economics for a near-high-end computer. Will an expensive single processor, or cheaper dual processors be less costly for the same performance? The IBM strategy can get to the very highest performance computer, but it is at the cost of using many processors.

The tradeoffs are interesting in this contest. IBM seems to be on a middle of the road approach. The core has very good performance, but it is not pushed to an extreme. Another approach would have been to use still lower power, lower performance cores and use even more of them. They likely worked through the options, however, and chose an optimum strategy.

[ 12-05-2002: Message edited by: snoopy ]</p>
post #269 of 441
[quote]Originally posted by Ed M.:
<strong>{...}</strong><hr></blockquote>

Actually, Ed, you've done nothing but attack me with your bullsh*t and not my points. You seem to be the one pleading for the masses to ignore me, based on your testimony. I think you're just stalking me.

Perhaps you should be in the disgruntled postal workers forum.

Have a nice day

PS. I'm glad you're learning all about debate. Now you should practice focusing on ideas, not the person. Unless, of course, you want to work for Kenneth Star or commit "hate" crimes.

D'OH!

<img src="graemlins/lol.gif" border="0" alt="[Laughing]" />

[ 12-05-2002: Message edited by: MacLuv ]</p>
post #270 of 441
[quote]Originally posted by Ed M.:
<strong>

In this case his claims about x86. After all, x86 is only a small piece of the equation. The embedded market for example, is significantly larger than the *desktop* computer market. He ignores all other processors in all other markets other than x86 and the *desktop*. It's called "cherry picking" or more correctly, he's making broad/sweeping generalizations and presenting it as the *rule*.

{...}
Perhaps we should all follow Programmer's example {...}

</strong><hr></blockquote>

Fine, perhaps you should follow programmer's example.

Originally posted by programmer:

[quote] Well technically the x86 ISA (as opposed to "the x86 processor") is the de facto standard, although there are enough variations to cause a fair bit of chaos (MMX, SSE, SSE2, 3DNow!). <hr></blockquote>

If you want to split hairs with me to make basis for your own arguments, fine. I presented the Gates to Sculley memo as further evidence to support what exactly "industry standard" means to the PC industry, but for some reason it was bounced because Amorph doesn't like Bill Gates or whatever. Frankly it seems like I'd have to somehow go through the hassle of proving that the WTC did indeed go down with some of you people.



Give it up already. I'm saying a car has four wheels, you're asking me to prove it because some may have three. I'm also talking about Apple Computer and future business strategy. Contrary to popular belief, technology is not a business strategy.

As far as anyone in here insisting that the 970 will gain momentum in sales not only through Apple but through high-end server sales as well, I'd like to examine this contradiction:

If the logic behind an x86 migration is that Apple would not be able to compete with Windows based on a loss of proprietary system design, what's going to happen to Xserve when Apple throws the 970 in there? It's captured a whopping 1.2% of the server market, and that's only in the US. Based on the same logic that x86 migration would kill off Apple, how can one apply the same logic to Xserve and say it will fly?

<img src="confused.gif" border="0">

[ 12-09-2002: Message edited by: MacLuv ]</p>
post #271 of 441
[quote]Originally posted by Outsider:
<strong><a href="http://www.siliconstrategies.com/story/OEG20021204S0041" target="_blank">Very interesting</a>, especially the 3rd paragraph.</strong><hr></blockquote>


Go Outsider! Thanks for the link. I find it interesting that IBM is on the forefront of SiGe in CPU fabrication. It makes complete sense, considering that their "Pixie Dust" is exactly that... Germanium, currently used on their winchester drives. The downside to SiGe is that Germanium is *very* *very* rare [but can be synth'd] and would make for higher costs.

For those who aren't clued in to IBM's Pixie Dust, it is a way to treat the winchester platters so you can achieve a higher arial density. I believe it also allows for lower voltage to flip a bit.

This obviously has *direct* implications for CPUs. Suddenly, you can achieve a greater density of transisters using the normal CMOS process *AND* operate at a lower core voltage. This means smaller, cooler, faster chips. yea!!


The future's so bright... I've gotta wear shades :cool:
http://blocklevel.com
Practical Information Architecture

The Visigothe

Ideas that change the way you think are the most powerful ideas of them all
Reply
http://blocklevel.com
Practical Information Architecture

The Visigothe

Ideas that change the way you think are the most powerful ideas of them all
Reply
post #272 of 441
[quote]If the logic behind an x86 migration is that Apple would not be able to compete with Windows based on a loss of proprietary system design, what's going to happen to Xserve when Apple throws the 970 in there? <hr></blockquote>

Huh? I can't see the timely addition of a PowerPC 970 hurting the XServe. 1.2% is quite impressive for a newcomer's first real product for that market.
Stoo
Reply
Stoo
Reply
post #273 of 441
[quote]Originally posted by Stoo:
<strong>
1.2% is quite impressive for a newcomer's first real product for that market.</strong><hr></blockquote>

Stoo--

<img src="graemlins/lol.gif" border="0" alt="[Laughing]" />

Sorry, you hit a funny bone... the next time I write a business proposal I'll make sure to mention that a 1% market penetration is worthy of investment...



[cartman voice]
Hippies... they want to save the world but all they do is smoke pot and smell bad... hippies...
[/cartman voice]

(PS... not calling you a hippie!)

<img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
post #274 of 441
[quote]Originally posted by MacLuv:
<strong>

Stoo--

<img src="graemlins/lol.gif" border="0" alt="[Laughing]" />

Sorry, you hit a funny bone... the next time I write a business proposal I'll make sure to mention that a 1% market penetration is worthy of investment...



[cartman voice]
Hippies... they want to save the world but all they do is smoke pot and smell bad... hippies...
[/cartman voice]

(PS... not calling you a hippie!)

<img src="graemlins/smokin.gif" border="0" alt="[Chilling]" /> </strong><hr></blockquote>

1% is often considered an excellent first move into a mature marketplace...I guess when you are focused on c-stores it would be bad but in beer 1% share is significant and "worthy of investment"

get back to Econ 201 hoser

[ 12-05-2002: Message edited by: Little Newton ]</p>
semper ubi sub ubi
Reply
semper ubi sub ubi
Reply
post #275 of 441
[quote]Originally posted by MacLuv:
<strong>

Stoo--

<img src="graemlins/lol.gif" border="0" alt="[Laughing]" />

Sorry, you hit a funny bone... the next time I write a business proposal I'll make sure to mention that a 1% market penetration is worthy of investment...

</strong><hr></blockquote>

You'll probably want to go to business school before you try it too! What were you expecting? Apple to have 50% of a decades old server server market in one quarter?
post #276 of 441
[quote]Originally posted by Programmer:
<strong>I have evidence that ApplePi is intimately related to the 970's bus.</strong><hr></blockquote>

Well, um, yeah.
Now, is it RapidIO-like or Hypertransport-like?
If neither, would it be simple to bridge between them?

The wildest thing about the RapidIO parts I was reading about was that they were implemented in a FPGA - it sort of seemed eminently reprogrammable on the fly.
post #277 of 441
[quote]Originally posted by Programmer:
<strong>

I have evidence that ApplePi is intimately related to the 970's bus.</strong><hr></blockquote>

I understand if you do not wish to say more. But could you not at least let us know what ApplePI really is?
"People don't want handouts! People want hand jobs!" ~ Connecticut governor William O'Neil at a political rally, followed by riotous applause
Reply
"People don't want handouts! People want hand jobs!" ~ Connecticut governor William O'Neil at a political rally, followed by riotous applause
Reply
post #278 of 441
[quote]Originally posted by visigothe:
<strong>

The downside to SiGe is that Germanium is *very* *very* rare [but can be synth'd] and would make for higher costs.

</strong><hr></blockquote>

I didn't think germanium was that rare. Originally, transistors were all germanium, but later went to silicon because it tolerates heat much better. Germanium still has the edge for higher frequencies if the power can be kept low.
post #279 of 441
Wow... this guy is truly amazing. At one point he claims that Intel is bigger than IBM, and then the next he's criticizing a 1% market share entry into a well-developed market. I wish to God my company had a 1% market share in legal publishing. With 10% commission I'd be a rich guy (literally millionaire in months).

He can't be real. Maybe he's another one of 68k's "experiments".
post #280 of 441
[quote]Originally posted by Algol:
<strong>But could you not at least let us know what ApplePI really is?</strong><hr></blockquote>

I don't know anything about ApplePI, but there are some specs for the 970's bus on IBM's website.

There's a PDF to download, this is from page 12:
[quote]
Features
- Two unidirectional buses
- 32-bit read, 32-bit write
- Point-to-point
- Source syncronous
Elastic Interface
- Allows multiple cycle wire delays between chips
- Hardware deskew
Bus Protocol
- Address, control, and data multiplexing
- Sideband signals
- Pipelined transactions
- Out of order data
- Coherency and sharing via snooping
- Processor synchronization for SMP
Up to 900 MHz bit rate achieves up to 6.4 GB/s useable bandwidth<hr></blockquote>

But I can't decipher whether that is one of the other emerging 'standards' in disguise. The rates and a slew of the features line up with both RapidIO and HT as far as I can tell. Then there's bi-directional v dual uni-directional. Shrug.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Future Apple Hardware
AppleInsider › Forums › Mac Hardware › Future Apple Hardware › Finally an interesting G5 story