spikey>>> But that is just not the point i am making. Its not the
performance gains of the G3.
>>Hrmm, well, that's good, considering there _are_ no performance gains to
be had from the G3. "
>irrelevant to the point i was making. Irrelevant to the point you were
making. thats a bo***cks reply that says nothing.
On the contrary, your initial statement ("it's not the performance gains
of the G3") presupposes that there are some sort of performance gains to
the G3, which there aren't.
>>> Its the fact that if the 7460 is closely based on the G4 it will not
scale the clock speeds well, or it will encounter a problem in its design
which will prevent iot from scaling easily.
>>You're making two points out of one. Anyway the 74xx series only needs
to scale to (max) ~2 GHz, by then Apple can switch the iMacs to the G5
while the G6 hits the market.
>I wasnt making two points out of one, i was making two points using the
same subject.
1) "if the 7460 is ... based on the G4 it will not scale ... well"
2) "or it will encounter a problem ... which will prevent it from scaling
easily."
The first statement is (1) "the 7460 won't scale if it's a G4". The second
statement is (2) "it might run into a problem preventing it from scaling".
Now what could possibly cause "the 7460" to not "scale" (statement (1))?
Little gremlins called "problems preventing it from scaling" (statement (2)).
Hence the two points are the same; your first point implies the second.
>Scale to a max of 2Ghz???
It shouldn't need to go past that; Apple should be preparing to shift the
G5 into the low-end by then and have the G6 taped out and in final testing
revisions, ready for a pro launch.
>Going by the performance of the 7450
The performance of a processor does not necessarily reflect upon its
scalability in clockspeed. In fact, it usually doesn't.
> you expect it to scale from 1.33 Ghz to 2Ghz???
Apple would want to shift the consumer lines to the G4 for reasons I have
already outlined. It thus follows that Apple want high clockspeed G4s;
hence Apple will be in pursuit of processors of a 74xx design which can be
scaled upward to high frequencies.
>I assume you are going by past record because there is nothing else to
go by.
(past record of what?) On the contrary I am basing my conclusions on the
premises outlined above.
>You didnt reply with a an argument at all to my comment on scaling clock
speeds.
There isn't an argument to reply to apart from the largely vacuous "the
7460 won't scale".
>>> IBM puts more moeny into developing chips than motorola,
>>... they put more money into developing POWER. Their PPC commitment is
way behind Moto's. "
>What??? this is madness.
Have you ever seen an IBM product which uses G3s? Does IBM develop the
G4 architecture? Are they actively developing the G5?
> They put more money and commitment into developing cpus than any other
company, its a fact.
More than Intel?
>What made you think IBM got apple out of the 500Mhz dead end that
motorola got them into?
IBM's fabs, which required no further capital input ("IBM puts more
[money]") to fab the 7410s (as motorola had already designed the 7410,
and hence IBM's labs did not have to make any input which would require
IBM's capital).
IOW IBM's money-pouring into CPU development ("They put more money and
commitment") was NOT what supplemented Apple's G4 supply, rather, it was
IBM's fabrication plants, which are technologically ahead of Motorola's.
>>> so if they do come across a problem then they should be able to solve
it faster.
>>... so given that the G4 is the problem-ridden architecture, just let
IBM fab it while Apple solves the problems. Moto's G4 design isn't
hampered by their engineers, it's their fabs, the G4 is a horrendously
complex processor as it would appear. "
>firstly you think that IBM would want to do that?
For the right price.
>they are a huge company that puts mega amounts of money into developing
CPUs, so you think they will let Motorola design a cpu that IBM are
expected to build?
If there's money in it. See above.
> Secondly it is the over complex design of the G4 that hampers its performance,
You need to look at Moto's estimated SPEC95 numbers. Try
http://e-www.motorola.com/collateral/PPCCPUSUMM.pdf
The G4's performance is not hampered compared to the G3s, as the estimated
SPEC scores will demonstrate.
> if you look at its design you will see alot of the problems encountered
have been due to altivec.
Which problems would they be and why would they be due to Altivec?
> Hence why it doesnt matter who fabs it, if the design in the first place
is troubled then you will always encounter problems. and no-one said that
the G4 is a problem ridden architecture, you cant just assume that. You
can only tell its problem ridden until you try to improve it.
Hence one redesigns it a la 7460.
>>> I would recomment the G3 because i dont think a consumer needs
altivec,
>>How many times do I need to justify Altivec's benefit to the consumer?
>Well how many to you f***ing think? this is a debate, in a debate you
try to prove yor points. you have not proven them,
On the contrary I have already justified Altivec's contribution and
possible future contribution to the consumer uses of the Mac.
> you have only stated your opinion, which conflicts with mine.
I've already given factors which correlate my opinion (`apple should
pursue a G4 imac as the g4 has altivec acceleration and stronger float
performance which will greatly benefit apps such as itunes et al'). Hence
my opinion is valid.
>>> i think it would offer great value for money, i think the 400Mhz bus
would be a great selling point,
>>Not really. How many of these `average consumers' you speak of know what
a FSB is? Do any whitebox stores you know of or brand-name vendors push
the 400 mhz FSB of the P4s they produce? "
>None, the sellers of P4 machines concentrate on Clock speed of the cpu,
if they advertised the bus speed it would only confuse the consumer
So you assert that "[advertising] the bus speed [of a P4] ... would only
confuse the consumer".
> and it would detract from the appeal of high clock speed.
I hardly see how or why, even if it did, lower multipliers aid performance.
>On a mac though it would not detract, it would be a bragging point.
So fresh from asserting that "[advertising] the bus speed [of a P4] ...
would only confuse the consumer", you assert that pushing the Mac's FSB
speed "would be a bragging point." Precisely how is advertising of FSB
speed going to aid the Mac and detract from PCs as far as advertising
value is concerned?
>>> and i would rather have IBM behind apple than motorola going by the
recent past.
>>Apple needs to design the PPC by itself. Optimally they should buy the
IP from Moto and spin it off into a joint venture with IBM (IBM could
contribute their POWER IP and engineers). "
>But the fact is no-one knows if this is going to happen, so right now its
all make believe. The best option would be that, but of the options
available rioght now IBM is the way to go as i have stated.
Apple could always contract someone else to fabricate the various G4
designs for them. You presuppose that the G4 is flawed such that no
company can replicate it en masse. On the contrary IBM reportedly had
great success with the 7410 at high clockspeeds whereas Motorola didn't.
This suggests to me that quality of facilities, rather than architectural
design, is at fault. Hence get a better fab onto the job when it comes to
G4s.
>>> and the fact that while the G4 might be at 1.33 Ghz, the G3 would have
a 400Mhz bus.
>>... if it's such a big deal, then Apple would use a 400 mhz fsb on the
G4s. Bus multipliers don't hamper performance until we get past 6x or so.
>you think IBM would let out a technology like that
What technology would this be?
>on a motorola product??? Madness, it aint gonna happen.
And precisely what technology will IBM stand in the way of? I can't see
anything in the above which IBM has an interest against.
>>> And the fact that bus speed is one of the biggest
bottlenecks in a system would give it quite a big advantage/selling factor.
>>All you're suggesting is that Apple shuffle the bottleneck somewhere
else.
>What the hell??? The bottleneck isnt a constant throughout the system.
I didn't say bottlenecks were constant; I suggested that bottlenecks would
be present, but moving the bottleneck somewhere else isn't going to solve
speed problems.
>There are several of them, getting rid of one of the biggest would
increase performance alot.
You presuppose that the FSB is the greatest bottleneck.
>>> Macs need to come down in price, price is a huge selling point of PCs
over macs.
>>Apple sells their machines on capability. A person who cares more about
price doesn't care about the capabilities of their box, they already know
what they want to run on it. Telling them what they _can_ do on an imac
won't change them.
>wrong. The biggest and best way to tap into the Pc market is to give
valuye for money.
... and one of the best ways to do that is to bundle and develop cool
software, funded by sales of hardware. How is that `wrong'?
> That is the while point of capability of a machine, you get more
capability for less price.
The reason why PCs are so cheap is because a breakdown of the typical
price tag looks much like:
tangibles (parts, windows licence) + labour + skinny margin
Apple's equation looks more like:
tangibles (parts) + labour + bigger margin
The bigger margin supports R&D of the OS (presumably much more than the $x
OEMs pay for a 9x/XP licence) and other software. Also, Apple develop a
lot of their own chipsets and ICs (re their purchase of raycer graphics)
and contribute to the PPC (more in the recent past). Hence Apple's
products have a higher price tag. Critically though, it can't be lowered.
Apple _needs_ the 30% gross margin to fund R&D. Particularly, with the OS,
Apple doesn't have economies of scale working in its favour. Apple and MS
both develop full-featured consumer OSes and hence spend (at least
roughly) similar amounts on them. MS can spread the cost over 95% of the
PC market; Apple has only 5%.
> Hence why value for money is the way to tap into the PC market.
If Apple could reduce the price of Macs further they would. If they
haven't, it's for a reason (outlined above).
>One of the most important facts about this debate is not what a cpu can
do, but what future it has.
True, there's no point using a high performance architecture only to bang
one's head into the wall eventually.
> about 2 years ago apple went with the G4. And it was a great processor
of its time, one of the best if not the best. But it had no future and
ultimately has put apple in the s**t. The mistake they made then is not
looking at the long term effects. If they go with a motorola design again
which is based on the original G4 then they are asking for the same kind
of trouble.
I can't see your point. Perhaps IBM should mint G4s for Apple?
> If they go with IBM then they are going with a design that could kill
many PCs, but more importantly will give apple a future.
Using a processor that debuted before 98 is hardly going to `give apple a
future'.
>IBM put money into developing for the future, motorola have not and do
not. infact they havent commited to the powerpc nearly as much as IBM
have offered to.
Since when has IBM committed to the PPC?
>Oh, and this 750FX is meant to have altivec like accelaration.
Perhaps. IBM licenced Altivec a while ago and are rumored to be in
negotiations to purchase it.
> so it ouwldnt be that far off the new G4. Or not nearly as much as the
current 750cx is.
Apple could conceivably settle for that, but as the G3 still needs better
floating point performance the G3s would need one or both of 1) higher
clockspeeds or 2) stronger float (perhaps the FX is a serious redesign of
the G3).