Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Re: Re: Re: Re: Re: It's not a G5 people, this isn't SpyMac...

I believe the G4 can natively address 64 GB of memory.

Yes, the G4 can address 64GB of memory. Unfortunately 32bit addresses make this difficult. The OS would need to internally represent every memory location as 64bit, causing significant performance reduction and resident system memory bloat. The Xeon is able to accomplish 36bit addressing in a slightly more graceful way due to the x86's segmented memory model. This effectively involves bank switching - the OS running on a Xeon can only utilise 4GB of memory at a time, but can swap banks (with a performance penalty) in order to effectively increase the addressable memory size.

Are these examples of the simpler design?

Yes and no. PPC970 represents individual instructions in a simpler manner, utilising a concept similar to the P4's crack and decode method to product micro-ops. The internal state of the PPC970 is much more complex than existing G4 designs.

The PowerMac G4s that are currently on the market make very little use of the second processor due to its insanely slow system bus--167MHz, and not nearly enough.

This is not exactly true. Both CPUs can run very efficiently once code is cached in their individual L3's. What causes most of the performance problems in the duals is keeping the caches coherent between the CPUs. Memory traffic is a secondary concern. Remember that memory runs at 133MHz while the CPUs runs at 1GHz+. On a similar note: the reason why we have not seen quad G4 workstations is because the G4 IPC Bus (Inter Processor Communication) performs poorly when more than two processors are connected to it. The PPC970 fixes this problem. :)

I congratulate the Apple Engineer(s) who routed the 970's 900MHz FSB on the upcoming apple mobos.
 
MacBidouille Is Dead Wrong

I will not elaborate on the following statement. I will just make it and drift quietly away: Hon Hai (Elitegroup) is not manufacturing the upcoming 970 PowerMacs. In fact, Elitegroup has nothing to do with any portion of this product, at all, in even the most minor way.

This entire MacBidouille piece is completely fabricated from air.
 
Re: MacBidouille Is Dead Wrong

Originally posted by MacWhispers
I will not elaborate on the following statement. I will just make it and drift quietly away: Hon Hai (Elitegroup) is not manufacturing the upcoming 970 PowerMacs. In fact, Elitegroup has nothing to do with any portion of this product, at all, in even the most minor way.

This entire MacBidouille piece is completely fabricated from air.


you won't elaborate or explain youreself? then i think that most people here will just take a momentarily glance at your post...

this rumor of the 970's is a very promising one, then you just come along and say it's false? what the!!???:eek: :confused:
 
UPDATE

macbidouille just posted this correction. This sounds a bit more reasonable.
Google Translation:
A mistake in interpretation slipped into the news of yesterday.

The PPC 970 will be shown with the developers with the WWDC and the public the following semane (week?). It will be on sale immediately
But unquestionable APPLE of its planetary success will start to make large stocks and to distribute them on the planêtre as of the end May.
 
Re: Re: French Translation Req: MacBidouille

Originally posted by celaurie
Why human? Basically, both my brain and Babelfish came up with the same things... give or take some sentence structure here and there... ;)
Because a human translation is always infinitely superior. Also, if you don't understand the source language you can never be sure whether the automated translation is correct. So, yes, if you understand the text + find the machine translation +- okay, there you go. But then you have to understand the stuff in the first place. So you always have to check humanly.
BTW, I have yet to see a text that was accurately translated by a machine. As for minor details, even those can confuse. Lastly, for me, language is also about (a)esthetics.
This is what I got from Babelfish (excerpt):
"Only flat. These machines will be provided with Mac OS X 10.2 not optimized 64 Bits. For any purchase of such a machine, Mac OS X 10.3 will be offered to its exit in September. Under these conditions, has less than APPLE is not opposed to it, we will reveal the benchmarks machine at the latest on May 15."
Okay, what does "only flat" mean ? What does the last sentence mean ? I don't think one could know from this "translation". Okay, a human translation has already been given. If your brain came up with more or less the same, I pity you. Well, okay, I think you're having us on anyway...
 
Well guys,
it seems that lot of you get excited again about rumors. I just wanted to ask you: when is the last time that Apple fulfilled the expectations of the Apple community? Maybe a few decades ago. Maybe.
Do you really think that things will change? I suppose in June we will read again lot of disappointed members.

And another small point:
since it seems that over here messages of macbidouille are very often read, interpretated, commented from members who don't have absolutely no knowledge of french, wouldn't you think that instead of spending time on surfing around, it would be more usefull to learn a little bit of french? :D
 
Re: UPDATE

Originally posted by Shaktai
macbidouille just posted this correction. This sounds a bit more reasonable.
Google Translation:
Human translation:
A translation error has sneaked into yesterday's news.

The PPC 970 will be shown to developers at the WWDC and to the public a week later (semane is a typo in the original for "semaine").
It will be available for purchase right away.
But Apple, sure of its overwhelming (planet-wide) success, will create some stocks in order to unleash them all over the planet at the end of May.

Sorry for this error in translation.
 
Re: Re: UPDATE

Originally posted by dekator
...unleash them all over the planet at the end of May.
Sounds quite ominous...


Regarding Hon Hai:

I just did a quick google search for Hon Hai. Their registered trademark name is Foxconn. I'm not sure where 'EliteGroup' comes in. Hon Hai's primary markets, according to their website, are connectors for PCs, cables, and enclosures. One interesting note: under 'enclosures', they have an image of the original G3 Smurf case.

OK I just found a story from 2002 that reports that Hon Hai manufactures motherboards for Dell. They've been aggressively trying to expand to other markets, like assembling PCs and notebooks and manufacturing cellphones.
 
Originally posted by Titian
Well guys,
it seems that lot of you get excited again about rumors. I just wanted to ask you: when is the last time that Apple fulfilled the expectations of the Apple community? Maybe a few decades ago. Maybe.
Do you really think that things will change? I suppose in June we will read again lot of disappointed members.

Last fall, few people were seriously expecting 15" PowerBooks with 1 GHz processors and SuperDrives. Apple delivered.

January (MacWorld) was widely anticipated to be a no-show show, with little in the way of updates. Yet Steve produced 12" and 17" PowerBooks, Keynote, and (the only things that really were expected) Safari & iLife.

In April, Apple blew everyone away again with Final Cut Pro 4 and DVD Studio Pro 2.

Oh, and by the way, did you happen to hear about THE ONLY SUCCESSFUL LEGAL ONLINE MUSIC DISTRIBUTION SYSTEM?! Just a little thing.

What planet do you live on--Depression & Denial? :)
 
Originally posted by bikertwin
Last fall, few people were seriously expecting 15" PowerBooks with 1 GHz processors and SuperDrives. Apple delivered.

January (MacWorld) was widely anticipated to be a no-show show, with little in the way of updates. Yet Steve produced 12" and 17" PowerBooks, Keynote, and (the only things that really were expected) Safari & iLife.

In April, Apple blew everyone away again with Final Cut Pro 4 and DVD Studio Pro 2.

Oh, and by the way, did you happen to hear about THE ONLY SUCCESSFUL LEGAL ONLINE MUSIC DISTRIBUTION SYSTEM?! Just a little thing.

What planet do you live on--Depression & Denial? :)

If you are happy about that hardware then good for you. If didn't help much for Apple to close the gap to the PC's and no further it help to increase much the sales. These updates are just cosmetics my friend.
About the software you are right: Apple is doing great things, better than anyone else.

About music distribution: nice diversification of computer business and of course very nice way of earning money on music. I also must say that I am a music lover, for me quality is very important and at the moment the quality of digital music is far away from the quality of analogue. But this is another issue which should be put in a music forum. In other words for me music isn't a tissue to be used one day and to be thrown away after it was used 100 times.
But I suppose most of the people think differently. ;-)

cheers
 
Hardware-seeding

It might be true that some PPC 970-based systems have left the assembly lines. But that doesn't mean these are destined for the mass-market (sure, I don't have actual info on quantities).
I for one am absolutely sure that these are testing-platforms for seeding to a broader circle of developers. The new chip is code-compatible, yes, but because of the new architecture, i believe new and existing software can greatly benefit from optimization.
For me, the future scenario looks like this:
Apple shows off some PowerPC 970 demos at WWDC in june and supplies developers with public info and some with testing-platforms.
Public availability will follow in September, along with a 64-bit-OS (10.3).
This will crush sales of PM G4 for some two months. But, from what I have learnt, sales can't be much lower than they are now. This way however, Apple can eventually provide a better product to the customer.

What do you think?
 
Apple keeps coming up with wonderful things, and yes, they keep on suprising us.
But.... the ONE thing which is most rumored/talked about (the last couple of years even), has never come.... and that's the successor of the G4.
Starting just before MWSF 2002 people at Apple discussion/rumors sites have been speculating, and also expecting the "G5". Ofourse "it" (must be PPC 970) will come eventually, and the WWDC event this June sounds like the right time and place for it.
So a 1.4 GHz PPC 970-Mac is ready this May? Fine, just keep it where it is....
And let Apple surprize us once again with 10.3 (64 bits) AND the 970-Mac this June.
And just for the MHz-myth issue.... let the low-end start at 1.5 Ghz, please. :)
 
Hon Hai assembles computer boards and internals

Of course Hon Hai doesn't make the PPC 970 chip. They receive them from the vendor (and other components as well) and assemble and manufacature the various internal boards for the Macs.
 
Make me wanna!!

Apple has to make me drool over a new computer before I buy a new computer.
-
Last night at the iPod show at the local Apple Store: iPod is a kick@ss JukeBox driving JBL speakers on tripod jackstands.

The dual 1.4ghz TowerMac driving dual flatpanel displays is nice compared to my QuickSilver-733. But drooling was not initiated.
-
The QS-733 caused drooling when compared to the 6500/250 upgraded to G3-400.
-
The 6500/250 caused drooling when compared to the Performa 630-CD.
-
The Performa 630-CD caused drooling when compared to the MacSE30.
-
The MacSE30 with dual B&W monitors caused drooling when compared to the Mac Plus.
-
The Mac Plus caused drooling.
-
I want front panel access to USB and FireWire ports.
I want front panel access to a Zip Drive.
I want front panel access to dual optical drives.
-
C'mon Apple, make the new TowerMac really fast fast fast. Make the new TowerMac with useful features, like USB and FireWire plugins on the front panel so that I don't have to turn the tower around.

Make me wanna new TowerMac!! :D
-
JJ
 
Re: MacBidouille Is Dead Wrong

Originally posted by MacWhispers
I will not elaborate on the following statement. I will just make it and drift quietly away: Hon Hai (Elitegroup) is not manufacturing the upcoming 970 PowerMacs. In fact, Elitegroup has nothing to do with any portion of this product, at all, in even the most minor way.

This entire MacBidouille piece is completely fabricated from air.

Um, I thought Hon Hai == Foxconn, and Elitegroup is a separate company. Unless they recently merged? But I couldn't find any evidence of that.

Supposedly Elitegroup is no longer making Apple products, but they appear to be a separate company from Hon Hai:

http://www.digitimes.com/Backgrounders/ecs.asp?view=article&datePublish=2003/04/02&pages=02&seq=4
 
36 bit PAE addressing on PPRO+ arch

Originally posted by laodamas
Yes, the G4 can address 64GB of memory. Unfortunately 32bit addresses make this difficult. The OS would need to internally represent every memory location as 64bit, causing significant performance reduction and resident system memory bloat. The Xeon is able to accomplish 36bit addressing in a slightly more graceful way due to the x86's segmented memory model. This effectively involves bank switching - the OS running on a Xeon can only utilise 4GB of memory at a time, but can swap banks (with a performance penalty) in order to effectively increase the addressable memory size.

Just a minor nit: PAE addressing in the MMU of a PPRO or above processor doesn't really 'bank switch' in the traditional sense, since old bank switching tricks usually worked by mapping an addressable memory window across the total memory area by scribbling on a set of mapping bits at a specific memory cell (or sometimes an offset register as with the PDP-11). The processor was completely unaware of the bank switching hardware (otherwise the memory would be directly accessible) and as such block transfers, jumps, pointers, etc. were completely impossible between memory segments. Usually one would reserve a buffer window for transferring data from one segment to the next. This is how most 8 bit systems handled greater than 64KB of ROM/RAM... systems like the old Apples, C-128, a variety of CP/M boxes like the old Kaypro, etc.

The original 8088 in the IBM PC could directly address 1MB (20 bits of address lines) of RAM, but because of an internally brain damaged design, it divided the address space into 16 bit regions Intel called 'segments'. Essentially, it was a 20 bit memory space externally with a 16 bit internal path. Moving data, handing pointers, and executing conditional jumps between segments was an exercise in frustration (to put it mildly), but possible. The stack was always available across segments though.

PAE is different. Internally a PPRO and above processor can only represent 32 bits of address space but the processor can address 36 bits worth of pages across it's pins, for a physical space of 64GB. The OS kernel and each process must still fit within a 4GB VM window. Usually this is handled by setting up a 36 bit page table in kernel space and then mapping pages per process into and out of that window. The processor can't really address more than 32 bits internally, but unlike old bank switching one needn't copy data bank and forth between banks in a common buffer since access to the total address space is simply a matter of setting a page table entry. So, unlike with bank switching the processor can represent 36 bits worth of pages for mapping into the 32 bit window directly - though for interprocess communication one must still copy data into kernel space and then back out to pages associated with each process - an expensive operation in context switching. Old bank switching could never perform these tricks since those old processors didn't even have an MMU to speak of - never mind pages, memory protection bits, etc.

As a side note, one person suggested that 36 bit architectures are inherently more performance expensive than 32 bit. This is not true. Old DEC System 10 and 20 boxes were fully 36 bit back in the late sixties through the early eighties with no performance penalty due to their memory bus sizes. These systems generally implemented 6 bit bytes, hence a 36 bit bus.

OK this is getting seriously offtopic. Getting back to message - yeah, I want a PPC970 based mac too! :) --M
 
Yeah, but did you realize

Originally posted by crapple33
wow. we just had four people answer baghdadbob's question with the exact same answer. lol.


Yeah, but did you realize all four of those replies were made within about 4 minutes time...enjoy the refurb. list while it is there...will vanish soon as it always does!
 
Re: Re: Re: Re: Re: It's not a G5 people, this isn't SpyMac...

Originally posted by ffakr
I'm confused. You are getting your info from the internet but you seem to know that the architecture of the 970 is a 'simpler design' than the G4. How so?
Did you know that the 970 is 8 way superscaler while the G4 is 5 way [i believe]? (that it can issue up to 8 instructions per clock cycle while the G4 maxes at 5) Did you know that it will have a longer pipeline than the G4? Are these examples of the simpler design?

Quite correct. The 970 design, as described on arstechnica is very complex, much more so than the current g4s.

Originally posted by ffakr
I think it's pretty safe to say that the 970 will be much faster than the G4... even when you compare a dual G4 to a 970. The SPEC scores show about twice the performance for the 970 and there is significantly more bandwidth available to the processor. The chipset of any 970 based machine will also likely use hyperthreading [probably between north & southbridges] so the entire system will be faster.

Hyperthreading, as I understand it (and certainly as Intel is advertizing it) is a processor trick rather than a northbridge/southbridge trick: allowing threads access to chip functionality that the current active thread isn't using. I have never seen HT technology mentioned with respect to the 970 except in "possible future 980 development" contexts. Furthermore, the jury is still out on the overall benefits of this technology. (IIRC, shorter pipelined designs don't get the boost that an insanely deep design like the p4 does.)

Originally posted by ffakr
I doubt we'd see something like a single 1Gig G4, a dual 1.25Gig G4, and a single 1.4 GHz 970 in that order though. It'd be awfully hard to market unless they portrayed the 970 machine as a workstation. I mean, IBM still demands $10,000 for Power based workstations that clock significantly lower than their P4 desktops.

Power chips are a different thing entirely than the 970. They are huge, with huge caches and often with multicores. Given the die size of the 970 (I've seen conflicting reports, but it seems to be a bit bigger than the g4), there is no reason to suspect that we won't be seeing it in the powermacs and powerbooks at around current pricing.

There is also the question of cache. Right now g4s are expensive because, due to the slow FSB, they have massive l3 caches. The 970 won't need this because of its fat pipes. This will further reduce cost, unless IBM decides to use this cost for other (perhaps faster) cache.

Cheers,
prat
 
Re: Quick translation by myself

Originally posted by Veldek
then the higher models (again not sure)
Seul bémol (?).

(Not sure about the next, but I think means Apple employees get 30% off of G4 Pro, which indicates that this will be the end of those machines.)

HTH,
Veldek

All in all thats a good translation... "then the higher models (again not sure) " its actualy they are saying top of the line

"Seul Bémol" means "the only dilema"

this you did all good on (Not sure about the next, but I think means Apple employees get 30% off of G4 Pro, which indicates that this will be the end of those machines.)

aethier
 
Re: Re: Re: Re: Re: Re: It's not a G5 people, this isn't SpyMac...

Originally posted by praetorian_x
I have never seen HT technology mentioned with respect to the 970 except in "possible future 980 development" contexts.

No, he meant HyperTransport.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.