As long as Apple can give us a >30% performance increase at the same price as today's machine, I'll be getting a new Pro machine. Coupled with a Jaguar performance boost, these new machines should scream. Originally, I was going to hold out for the G5, but I'm getting tired of the wait.
My guess for a the top end machine at MWNY would be a dual 1.33 GHZ machine with a 166MHZ DDR using the X-Serve hack. Maybe an increase of the L3 cache to 4MB. This machine should be at least 30% faster than the current dual 1GHZ machines.
Overall, despite all the naysayers in this forum, I believe Apple is doing a surprisingly good job of competing in a very depressed tech economy. Despite the Megazhertz gap, OSX continues to mature and get faster. Apple is milking performance from their software which is better than the Wintel solution of using all the extra MHZ to run bloated code. My guess is that one of the reasons Microsoft code seems fast (ie. IE6 for Windows) is that they do not do all the bounds checking and exception handling that is needed to make a system secure. Yes, each security patch that I download for my Windog machine seems to make things a little bit slower. Security has its price. I would much rather waste a second or two for a web page to load than to re-install my OS and restore all my backed up software.
<strong>Next year is too long. MWSF? Apple could pull through by the skin of their teeth. But any later, and game over. Apple cannot sell 1 GHz Powermacs against 3.5 GHz Pentium boxes, it just ain't gonna happen.
I fear that by the time the G5 arrives, Apple's user base will have migrated to Wintel.</strong><hr></blockquote>
Actually, from my perspective, more people are moving to Apple, rather than away. Mac OS X is the sole reason.
There are probably more people out there than you realize (remember, 5% of computer users would be tremendously huge to Apple) that feel faster CPUs are not the problem with computing right now.
These are people that value reliability, stability, integration, and so on and Apple is delivering on these points. I've got a 400MHz iBook and a 450MHz Cube and while at times they seem a bit slow, it's not like 5+ years ago when I always felt that I needed a faster system.
I'm seeing a lot of switchers - people that have used unix at some time in their past, people that are just sick and tired of having virus after virus sweep through their system. There are also those that started to switch to Windows 1+ years ago that are still finishing that process up, but I haven't seen anyone switch that way in the last 6 mos, rather people starting the switch to Mac.
<strong>I don't think we'll ever seen a dual MPX bus setup -- it burdens the chipset too much with so many pins and bus snooping duties, not to mention the dual and single processor machines would need different chipsets and motherboards. And all this just for 6-12 months until the next big thing comes along. More likely we'll see exactly the Xserve solution, or a G4 w/ 166 MHz (which Motorola has mentioned).
Other than that you're probably not far off.
JYD: Hardware sales are off throughout the market, its not just Apple (nor just their Pro machines). Better machines would probably improve sales slightly, but I doubt it would be a very large effect.</strong><hr></blockquote>
Again I will fully admit that I could be just pissing in the wind so to speak. The thing that makes me think this is the same thing that is holding us back, the G4 has it's memory controller on chip. I took this quote from Ace's Hardware
[quote]For multi-processor systems with a full on-die memory controller, the aggregate memory system bandwidth increases linearly with each extra CPU, which makes designing a system with high bandwidth easier, and also more scalable - memory resources increase with memory demands. This also helps for server systems which need lots of memory as it's very hard to control multiple memory DIMMs from a single controller at high speed - having more controllers and fewer DIMMs per controller makes a higher speed design easier.
Multi-processor systems for CPUs with on-die memory controllers still need high-speed cache coherency connections, and also a way to access "non local" memory - this is a bit like having a system with multiple northbridges. Such a coherency/non-local connection could be rather like a "front side bus" in current multi-processor systems, which would mean that non-local memory requests would have similar latency and bandwidth characteristics as current systems.
<hr></blockquote>
Again from what I understand, Apple doesn't have a northbridge/southbridge combo like all the PC motherboards. They have one controller chip. (which has to already have been redesigned due to xserve)
Adding a second cpu bus wouldn't require as much trouble as you think because it isn't requiring access to a shared memory controller for both cpu's. In this regard it would be more like AMD's 760mp solution only more elegant and better.
It would be a neat engineering trick to turn a disadvantage (old on chip memory controller) into an advantage via some engineering magic. It would also allow us Mac users to tell PC users to stop comparing our Mac's with true server/workstation engineering to their PC's with an antiquated n/s bridge and all the resources associated with it.
<strong>To contribute to this thread a good friend of mine who is a former Moto employee talked to some of his old co-workers this week. The rumor inside of Moto itself is that they will no longer be producing CPUs for Apple in the future. Apple has decided to go with someone else. Whether that is IBM I don't know, but if Moto's own employees are whispering about it, there certainly is some credence to those rumors. And since this person barely knows who Apple is and would never bring the subject up in conversation I put some stock in it.</strong><hr></blockquote>
i knew there was someone on here who used to work for moto. yes, what eskimo is saying is true. i mentioned it back in january, so it's been going on for quite some time. this isn't new information. someone else mentioned the "pi" part of it (which i had heard about in jan, but didn't want to disclose because the info i got would have made the source very obvious). that was the straw that broke the camel's back. ibm's doing the next gen chip.
and from what i heard back then, they should have been in production by now for a month or two. of course, i have no idea if that means they'll show up at mwny. plus, i'm not very technical honestly, so a lot of the information he told me got lost in the mush that is my artsy brain. i would have given more information, but at the time i didn't want to risk anyones' jobs.
SiS used to make a Pentium 4 motherboard with an IC as opposed to a north/southbridge setup.
An IC is good in that there is no north/south bridge bottleneck, but means for a very large chip. There is no way you'd get decent integrated graphics, or surround sound onto an IC without making it too large to be fabbed.
A G5 iMac, with its new approach where the CPU controls the memory, and Rapid-I/O switch(es) link main controllers like a LAN would probably look something like
<strong>Next year is too long. MWSF? Apple could pull through by the skin of their teeth. But any later, and game over. Apple cannot sell 1 GHz Powermacs against 3.5 GHz Pentium boxes, it just ain't gonna happen.
I fear that by the time the G5 arrives, Apple's user base will have migrated to Wintel.</strong><hr></blockquote>
I'll say this much:
My predictions and timetable have softened based on info in the past week. I know it sounds cheesy given that its directly following MOSR's predictions, because they are similar, but maybe we have the same source:
1.0, 1.26, 1.4 Ghz Dual G4s
133 system bus (that's right folks, no change...)
PC133 SDRAM (again, no DDR, or MAYBE the so-called "Hack")
Explanations can be provided if you want to private message me...
MOSR claims 1.0, 1.26, and 1.4 with a 166 bus. Someone want to explain how you achieve these frequencies with this bus? Sounds like a load.
Now, before you get all "Game over" Junkyard Dawg, please rest assured that the game has only begun. The bus doesn't need an update, and DDR would be there only for marketing, it wouldn't add much speed at all. (Private message me for more)
Steps have been taken to lighten the load on the Northbridge so memory can function better. I won't go into details....yet....but think about what could be unloaded from Northbridge and how.
Will be an amazing machine.
Now, updated timeline:
MWNY02:
See above
MWSF03:
G4s at or around 1.5 Ghz (maybe midrange)
166 Mhz bus (check multiplier...it works)
DDR for sure.
MWNY03:
G5.
3 of my sources have confirmed that the G5 is IBM sourced. No more talk of Moto in the G5s. I don't know yet if it will be a PowerX derivative, or what, but if I hear anything I will put it down.
Now, for some this might be a let down, but seriously folks, the 133 bus isn't our problem, no matter how much JYD (and admittedly, I, previously) has said so.
PC133 SDRAM (again, no DDR, or MAYBE the so-called "Hack")
Steps have been taken to lighten the load on the Northbridge so memory can function better. I won't go into details....yet....but think about what could be unloaded from Northbridge and how.
</strong><hr></blockquote>
So, new G4 with integrated memory controller. I'd be surprised if Xserve DDR wasn't there considering that Apple's already invested the R&D to bring it to market, and the Xserve pricing suggests that it's not overly expensive to implement. (.edu gets a $500 discount on each config, so there are either some real margins in these things, or Apple is courting higher-ed like they haven't in quite some time.)
Toss in some upgrades around the edges (Firewire, etc.) and this could be a damn nice update. I'd really like to see up to 2 of the Xserve removable drives used in a tower config - dual IDE controllers + striping could improve disk perf rather nicely for video applications, but after playing with an Xserve today, I'm not sure those drives would fit - they're damn big.
That'd just be icing on the cake. It'd help drive down costs of the drive modules due to the volume, and deliver added revenue to Apple instead of 3rd parties, but the prospect of a $350 60GB drive might cool people off a bit.
Okay, I have been lurking too long. I guess it is time for me to post.
Let me start by saying that Motorola is probably not getting complete out of the CPU business. They don't want to burn that bridge, but at the same time they don't want to invest the resources right now to get the job done. As a result, they will honor their obligations most likely by "contracting" out CPU production to a third party. Initially this is likely to be an asian semi-conductor producer, but the long term future will be dependent on how well they do.
When motorola dropped the ball on G5 development, Apple stepped in and invested further resources. As a result, Apple has certain rights to some of Motorola's technology. Apple wants the chips, and they don't care who makes them or what they are called (other than for marketing purposes) as long as they get them. Motorola is most likely going to contract out production for the time being, in the meantime, Apple is "hedging its bets" by talking with IBM and possibly AMD to produce its futrue chips as well. Understand, these things are all in the early development phases, and any the winds could shift at any moment.
If AMD or IBM was contracted to produce chips, it would most likely be the the chips that are of Apple's and Motorola's joint design. i.e the G5 in some form. Motorola will keep just enough control, to allow them to bring it back in house, if their fortunes turn in their favor, otherwise they will themselves subcontract manufacture and development out or allow Apple to do it. Either way, they dispose of a part of business that is not making money for them, while they collect royalties on their intellectual property unless Apple buys it outright.
Now what will we see? Well I don't care what anyone else says, but we will see a faster motherboard with an improved implementation DDR RAM, and it will make a "significant" performance difference. There will also be faster chips, of at least 1.3 ghz.. The 1.6 does exist, but not in sufficient quantity to meet demand. 1.3 to 1.4 would be the highest top end to hope for, but combined with the new mobo, the performance increase will be substantial.
When will we see this? Well time is a nebulous beast, and difficult to tame. Steve wants it for MWNY. There is a chance of delays, but I think still before September.
What will we most likely see? Okay, here in a fit of optimism, I will stick my neck out. If you end up with me at chopping block, be merciful and chop quickly.
I predict the top end to be around 1.4 ghz. Not exactly that, but around that. Mobo speed on the top end will by fall be 333mhz. The chips will have 512k on chip L2 cache per processor running at processor speed, backed up by 2 mb of L3 cach per process. These puppies will scream.
The new iDevice will be bit of a letdown for Apple Loyalists. It just won't be gee-whiz enough, still it will do okay in the marketplace where it will attract new customers to apple. It may be capable of working with multiple computer platforms, but of course, Apple will have the complete package that works seamlessly together.
-----------
...what is that shadow? why, why, wh..... It looks (gulp!) like a really, really, huge ax..... (tremble).
a question for any of the other people with moto information: will moto continue to make "PowerPC" chips in the future? if you know what i'm talking about it makes sense.
<strong>This is the strongest case yet for Apple not having a massive update in July, and I will tell you what it will be. The fact that people are alleging insider trading on Apple due to the fact that so many execs, filed and sold their stock before their announcement about a sales shortfall.
These same execs that "didn't know" about the shortfall and selling their stock before it was announced also know if there is a real update in July that would cause the same said stock to rebound.
Based off of that I believe we will have the iMac staying with GF2mx and being bumped to 1 ghz across the line with only the drives and ram changing the price. They will also get a 133 mhz bus and a price drop. eMac will also get 1 ghz across the line, drives/ram changing price points, but will stay at 100 mhz bus.
The PowerMac line will get dual processors at 933, 1 ghz, and 1.2 ghz. Superdrives will be standard. Radeon 32 meg 7500 will remain on the low end and mid will keep 64 meg GF4mx and the top come standard with 128 meg Radeon 8500.
Motorola will not have created a true DDR solution for Apple. Apple will have engineered around this with the same solution we saw in the xserve only now they will have had a little more time to improve it a bit more. Steve will explain that the DDR implementation is DDR all the way up to the processor just like in the xserve. However on the xserve the bus to the processor is still sdr-133 that is all the G4 will support.
Apple will explain that their new dual processor machines now have two cpu buses instead of just one. The two buses are independent and can each access the DDR solution at their full sdr133 speed.
Apple will also include ATA-RAID capability in the new PowerMacs, but will not be included in on the low end to help seperate the lines a bit. (dual 933 to dual 1 gig isn't much of a jump, but dual 1 gig to 1.2 is pretty large)
Steve will announce that the prices have been dropped $300 across the board. ($200 on the low end) (Making all the folks who bought at the old price points feel like complete asses) He will also point out (just as he did before) that the previous high end Mac is now available with better throughput and drive options for $1000 less than it was just a week before.
So new pricing
$1399
dual 933
Radeon 7500w/32 megs
superdrive
256 ddr ram
60 gig HD (no raid option)
$1999
dual 1 gig
Geforce4mx w/64 megs
superdrive
512 megs ddr ram
120 gig HD (possibility of raid option by adding drive or also added via Applestore)
$2699
dual 1.2 gig
Radeon 8500 w/128 megs
512 megs ddr
superdrive 2 (just announced today in press release)
dual 120 gig HD(with 8 meg cache) preconfigured in fast array.
It will be enough for now and will raise PowerMac sales up to their old levels but Apple will sell less of the high end machine and thus make a smaller profit.
I have no insider info, I just know Apple, and have been a Mac user since 1990.
I think that you are very near from the truth. Considering the rumors and the good info from Eskimo (good : reliable) i think your prediction will be ok.
But i will put some minor restriction : i think that they will be still a single powermac computer (let's say a 933 mhz one) and two dualies the 1 ghz one and a 1,13 one or a 1,2 ghz one (it depends on the progress of mot ).
The mpx bus will be still a 133 mhz one (the 166 mhz memory module are not enough common), they will use the same memory controller than the xserve , so they will use DDR ram.
In the future i see two lines : a line of consumer chips with sahara and sahara 2 chips for the consumers and powerbook line, and the proline with a single core power4 with an altivec like unit.
In the near future perhaps Mot can do an ultimate G4 developpement a G4 blended with G5 stuff like I/O and memory support for DDR ram. 2002 seems to be the year of transition. The year of change will be 2003.
[quote]I fear that by the time the G5 arrives, Apple's user base will have migrated to Wintel.<hr></blockquote>
You're basing that on the assumption that processor speed is all anyone cares about. It would never even occur to me to switch to peecees just because they're running 3HGz processors. Okay, I'm only responsible for a dozen Macs (not including my own machines), but they will all be replaced by new Macs next year. Most of our current Macs are single G4 450s and they're already faster machines than the people using them need for the work they do. But we're on a three-year upgrade schedule and next year we get new machines.
Sure, I want a G5 for myself, but the one thing people in my office care about MOST is...the size of their monitor. Nobody (except me) could tell you what processor they have, or how fast it is, or whether it's faster than the 8500 it replaced, but man, they can tell you that they have a 17-inch monitor and it's SO much better than their old one.
There are a LOT of Mac users who don't know anything about processors or clock speed or DDR RAM. They don't compare the specs of Macs to peecees because they don't know enough about computers to understand what they're comparing. What they know is: Macs are fun and easy to use. Peecees are complicated and crash a lot. EVERYONE in my office who buys a computer to use at home buys a Mac. They always ask my advice about what model to buy, but nobody has ever said to me "Gee, this Dell runs at 2.1GHz, should I consider getting one of those instead of a new Mac?"
However, the problem is NOT existing mac users, it's professionals who have no platform loyality. These pros will use whatever gets the job done fastest, because time is money. The advantage of OS X counts for a bit, but it doesn't make up for the gross performance differential between Wintels and Powermacs.
Furthermore, Apple needs to expand their market share. They MUST win over rednecks/morons for this to happen, and these types do care about GHz, very much so. We're talking about the sorts who would just as soon overbore their small block chevy V8 on a saturday as play Quake, or drop a radical cam into their Mustang rather than burn a CD of Poison's greatest hits. Apple's got to win over these idiots and it's not going to happen with 800 MHz cutesy computers.
Besides I'm just finishing up my 383 for my '69 roadrunner and I don't want a Windawg 2zigahertz box I want the torque of a big block with out all the tuning hassle of the 426 Hemi (yeah I know, should have gone with the 426 wedge)
<strong>Let me start by saying that Motorola is probably not getting complete out of the CPU business. They don't want to burn that bridge, but at the same time they don't want to invest the resources right now to get the job done. As a result, they will honor their obligations most likely by "contracting" out CPU production to a third party. Initially this is likely to be an asian semi-conductor producer, but the long term future will be dependent on how well they do......When motorola dropped the ball on G5 development, Apple stepped in and invested further resources. As a result, Apple has certain rights to some of Motorola's technology. Apple wants the chips, and they don't care who makes them or what they are called (other than for marketing purposes) as long as they get them. Motorola is most likely going to contract out production for the time being...........If AMD or IBM was contracted to produce chips, it would most likely be the the chips that are of Apple's and Motorola's joint design. i.e the G5 in some form. Motorola will keep just enough control, to allow them to bring it back in house, if their fortunes turn in their favor, otherwise they will themselves subcontract manufacture and development out or allow Apple to do it. Either way, they dispose of a part of business that is not making money for them, while they collect royalties on their intellectual property unless Apple buys it outright.
</strong><hr></blockquote>
Be a nice deal for them too if they retain rights to seel the proc in the embedded space. Apple pays for getting the chips fabbed right and Moto has a new high end chip for marketing to makers of super-routers and the ilk.
However, the problem is NOT existing mac users, it's professionals who have no platform loyality. These pros will use whatever gets the job done fastest, because time is money. The advantage of OS X counts for a bit, but it doesn't make up for the gross performance differential between Wintels and Powermacs.
Furthermore, Apple needs to expand their market share. They MUST win over rednecks/morons for this to happen, and these types do care about GHz, very much so. We're talking about the sorts who would just as soon overbore their small block chevy V8 on a saturday as play Quake, or drop a radical cam into their Mustang rather than burn a CD of Poison's greatest hits. Apple's got to win over these idiots and it's not going to happen with 800 MHz cutesy computers.</strong><hr></blockquote>
I'm sorry, why?
Surely there are enough sensible people out there that Apple don't need this particular segment of the market. The truth is that for the majority of people current machines are plenty fast enough, what they want is ease-of-use, reliability, and functionality. This is one of the reasons for the extended slow-down in computer sales, even corporations realise they don't need to upgrade their machines regularily any more. Since Apple are starting from such a small percentage of the market, they need only attack this portion of it to be able to massively increase their market share.
Personally, I want a much faster Mac, as I write scientific code that needs fast double precision floating point performance, which Apple can't yet deliver. I do, however, write the code on a Mac (B+W G3 350 O/C to 450) and run it on a PC, but I'm very much in a minority here. My field (mainly microwave/optoelectronic circuit analysis and design) is strictly PC, the only Mac programs I know of are the ones I compile for my own use. The Mac is just so much more pleasant to use, and for all my purposes bar running this code, quite adequate.
BTW, make up you mind whether Apple are bleeding market share, or failing to expand their market share.
Hey cool, a whole bunch of people who post like they really know something!
Re: 133 MHz MPX remaining unchanged and yet bandwidth problems being fixed. The problem with the fast G4 is that it can't read and write to memory fast enough (i.e. 850 MB/sec doesn't cut it). Motorola's own documentation shows that 850 is the MPX's limit, and they don't want to change it since the embedded crowd likes it the way it is. A 128-bit implementation is too expensive (and would constitute a change).
Obviously the only way that memory bandwidth could increase is by not using the MPX bus to connect to memory. Bringing the memory controller on-chip. Motorola has a DDR memory controller in the 8540, so they could use that and if the L3 cache interface is removed the pin count might not skyrocket so badly. I would have expected a RapidIO bus, but hey MPX works and its there already. This scheme would actually simplify embedded designs using the G4, so that might make Moto's other customers happy and would explain why Moto hasn't brought out a decent memory controller themselves (Apple's kick's their butt).
Another alternative would be to replace the L3 cache interface with an Intel-style FSB to an external memory controller, but this seems a little weird and considerably slower than the on-chip design.
Oh heck, maybe I should just PM you after all and put up with "ha ha, made you look".
Re: The need for fast machines. While it is true that there is a large group of Mac customers (realized and potential) for whom speed does not matter, there are also a whole bunch of markets that Apple clearly wants into that really really really care about performance. People to whom time is money. People to whom real-time is everything. Without at least competitive performance, Apple will have a very hard time competing no matter how good their software is.
If the speculation above is true, however, that ought to close the performance gap enough. A 20% MHz bump + duals + significant bandwidth improvement + I/O improvements + new GPUs on AGP 8x + Jaguar will make for a remarkable leap in performance. My money is on the table. If I were Apple my biggest concern would be how the heck to sell the existing PowerMac inventory!
Comments
My guess for a the top end machine at MWNY would be a dual 1.33 GHZ machine with a 166MHZ DDR using the X-Serve hack. Maybe an increase of the L3 cache to 4MB. This machine should be at least 30% faster than the current dual 1GHZ machines.
Overall, despite all the naysayers in this forum, I believe Apple is doing a surprisingly good job of competing in a very depressed tech economy. Despite the Megazhertz gap, OSX continues to mature and get faster. Apple is milking performance from their software which is better than the Wintel solution of using all the extra MHZ to run bloated code. My guess is that one of the reasons Microsoft code seems fast (ie. IE6 for Windows) is that they do not do all the bounds checking and exception handling that is needed to make a system secure. Yes, each security patch that I download for my Windog machine seems to make things a little bit slower. Security has its price. I would much rather waste a second or two for a web page to load than to re-install my OS and restore all my backed up software.
<strong>Next year is too long. MWSF? Apple could pull through by the skin of their teeth. But any later, and game over. Apple cannot sell 1 GHz Powermacs against 3.5 GHz Pentium boxes, it just ain't gonna happen.
I fear that by the time the G5 arrives, Apple's user base will have migrated to Wintel.</strong><hr></blockquote>
Actually, from my perspective, more people are moving to Apple, rather than away. Mac OS X is the sole reason.
There are probably more people out there than you realize (remember, 5% of computer users would be tremendously huge to Apple) that feel faster CPUs are not the problem with computing right now.
These are people that value reliability, stability, integration, and so on and Apple is delivering on these points. I've got a 400MHz iBook and a 450MHz Cube and while at times they seem a bit slow, it's not like 5+ years ago when I always felt that I needed a faster system.
I'm seeing a lot of switchers - people that have used unix at some time in their past, people that are just sick and tired of having virus after virus sweep through their system. There are also those that started to switch to Windows 1+ years ago that are still finishing that process up, but I haven't seen anyone switch that way in the last 6 mos, rather people starting the switch to Mac.
Just my perspective.
<strong>I don't think we'll ever seen a dual MPX bus setup -- it burdens the chipset too much with so many pins and bus snooping duties, not to mention the dual and single processor machines would need different chipsets and motherboards. And all this just for 6-12 months until the next big thing comes along. More likely we'll see exactly the Xserve solution, or a G4 w/ 166 MHz (which Motorola has mentioned).
Other than that you're probably not far off.
JYD: Hardware sales are off throughout the market, its not just Apple (nor just their Pro machines). Better machines would probably improve sales slightly, but I doubt it would be a very large effect.</strong><hr></blockquote>
Again I will fully admit that I could be just pissing in the wind so to speak. The thing that makes me think this is the same thing that is holding us back, the G4 has it's memory controller on chip. I took this quote from Ace's Hardware
[quote]For multi-processor systems with a full on-die memory controller, the aggregate memory system bandwidth increases linearly with each extra CPU, which makes designing a system with high bandwidth easier, and also more scalable - memory resources increase with memory demands. This also helps for server systems which need lots of memory as it's very hard to control multiple memory DIMMs from a single controller at high speed - having more controllers and fewer DIMMs per controller makes a higher speed design easier.
Multi-processor systems for CPUs with on-die memory controllers still need high-speed cache coherency connections, and also a way to access "non local" memory - this is a bit like having a system with multiple northbridges. Such a coherency/non-local connection could be rather like a "front side bus" in current multi-processor systems, which would mean that non-local memory requests would have similar latency and bandwidth characteristics as current systems.
<hr></blockquote>
Again from what I understand, Apple doesn't have a northbridge/southbridge combo like all the PC motherboards. They have one controller chip. (which has to already have been redesigned due to xserve)
Adding a second cpu bus wouldn't require as much trouble as you think because it isn't requiring access to a shared memory controller for both cpu's. In this regard it would be more like AMD's 760mp solution only more elegant and better.
It would be a neat engineering trick to turn a disadvantage (old on chip memory controller) into an advantage via some engineering magic. It would also allow us Mac users to tell PC users to stop comparing our Mac's with true server/workstation engineering to their PC's with an antiquated n/s bridge and all the resources associated with it.
Trumptman
(wondering why his shins are wet)
[ 06-29-2002: Message edited by: trumptman ]</p>
<strong>To contribute to this thread a good friend of mine who is a former Moto employee talked to some of his old co-workers this week. The rumor inside of Moto itself is that they will no longer be producing CPUs for Apple in the future. Apple has decided to go with someone else. Whether that is IBM I don't know, but if Moto's own employees are whispering about it, there certainly is some credence to those rumors. And since this person barely knows who Apple is and would never bring the subject up in conversation I put some stock in it.</strong><hr></blockquote>
i knew there was someone on here who used to work for moto. yes, what eskimo is saying is true. i mentioned it back in january, so it's been going on for quite some time. this isn't new information. someone else mentioned the "pi" part of it (which i had heard about in jan, but didn't want to disclose because the info i got would have made the source very obvious). that was the straw that broke the camel's back. ibm's doing the next gen chip.
and from what i heard back then, they should have been in production by now for a month or two. of course, i have no idea if that means they'll show up at mwny. plus, i'm not very technical honestly, so a lot of the information he told me got lost in the mush that is my artsy brain. i would have given more information, but at the time i didn't want to risk anyones' jobs.
[ 06-29-2002: Message edited by: admactanium ]</p>
<a href="http://developer.apple.com/techpubs/hardware/Developer_Notes/Macintosh_CPUs-G4/eMac/art/nor01.gif"; target="_blank">http://developer.apple.com/techpubs/hardware/Developer_Notes/Macintosh_CPUs-G4/eMac/art/nor01.gif</a>;
PowerBook
<a href="http://developer.apple.com/techpubs/hardware/Developer_Notes/Macintosh_CPUs-G4/PowerBook_G4/art/iv01.gif"; target="_blank">http://developer.apple.com/techpubs/hardware/Developer_Notes/Macintosh_CPUs-G4/PowerBook_G4/art/iv01.gif</a>;
and Power Mac
<a href="http://developer.apple.com/techpubs/hardware/Developer_Notes/Macintosh_CPUs-G4/PowerMacG4/art/nicr01.gif"; target="_blank">http://developer.apple.com/techpubs/hardware/Developer_Notes/Macintosh_CPUs-G4/PowerMacG4/art/nicr01.gif</a>;
Use a Uni-N Northbridge and KeyLargo south bridge.
The iMac
<a href="http://developer.apple.com/techpubs/hardware/Developer_Notes/Macintosh_CPUs-G4/iMac/art/ts01.gif"; target="_blank">http://developer.apple.com/techpubs/hardware/Developer_Notes/Macintosh_CPUs-G4/iMac/art/ts01.gif</a>;
and iBook
<a href="http://developer.apple.com/techpubs/hardware/Developer_Notes/Macintosh_CPUs-G3/ibook/art/dsl01.gif"; target="_blank">http://developer.apple.com/techpubs/hardware/Developer_Notes/Macintosh_CPUs-G3/ibook/art/dsl01.gif</a>;
Use a single "Integrated Controller" or IC.
SiS used to make a Pentium 4 motherboard with an IC as opposed to a north/southbridge setup.
An IC is good in that there is no north/south bridge bottleneck, but means for a very large chip. There is no way you'd get decent integrated graphics, or surround sound onto an IC without making it too large to be fabbed.
A G5 iMac, with its new approach where the CPU controls the memory, and Rapid-I/O switch(es) link main controllers like a LAN would probably look something like
Barto
[ 06-29-2002: Message edited by: Barto ]</p>
<strong>Next year is too long. MWSF? Apple could pull through by the skin of their teeth. But any later, and game over. Apple cannot sell 1 GHz Powermacs against 3.5 GHz Pentium boxes, it just ain't gonna happen.
I fear that by the time the G5 arrives, Apple's user base will have migrated to Wintel.</strong><hr></blockquote>
I'll say this much:
My predictions and timetable have softened based on info in the past week. I know it sounds cheesy given that its directly following MOSR's predictions, because they are similar, but maybe we have the same source:
1.0, 1.26, 1.4 Ghz Dual G4s
133 system bus (that's right folks, no change...)
PC133 SDRAM (again, no DDR, or MAYBE the so-called "Hack")
Explanations can be provided if you want to private message me...
MOSR claims 1.0, 1.26, and 1.4 with a 166 bus. Someone want to explain how you achieve these frequencies with this bus? Sounds like a load.
Now, before you get all "Game over" Junkyard Dawg, please rest assured that the game has only begun. The bus doesn't need an update, and DDR would be there only for marketing, it wouldn't add much speed at all. (Private message me for more)
Steps have been taken to lighten the load on the Northbridge so memory can function better. I won't go into details....yet....but think about what could be unloaded from Northbridge and how.
Will be an amazing machine.
Now, updated timeline:
MWNY02:
See above
MWSF03:
G4s at or around 1.5 Ghz (maybe midrange)
166 Mhz bus (check multiplier...it works)
DDR for sure.
MWNY03:
G5.
3 of my sources have confirmed that the G5 is IBM sourced. No more talk of Moto in the G5s. I don't know yet if it will be a PowerX derivative, or what, but if I hear anything I will put it down.
Now, for some this might be a let down, but seriously folks, the 133 bus isn't our problem, no matter how much JYD (and admittedly, I, previously) has said so.
Pvt message for more info.
<strong>
1.0, 1.26, 1.4 Ghz Dual G4s
133 system bus (that's right folks, no change...)
PC133 SDRAM (again, no DDR, or MAYBE the so-called "Hack")
Steps have been taken to lighten the load on the Northbridge so memory can function better. I won't go into details....yet....but think about what could be unloaded from Northbridge and how.
</strong><hr></blockquote>
So, new G4 with integrated memory controller. I'd be surprised if Xserve DDR wasn't there considering that Apple's already invested the R&D to bring it to market, and the Xserve pricing suggests that it's not overly expensive to implement. (.edu gets a $500 discount on each config, so there are either some real margins in these things, or Apple is courting higher-ed like they haven't in quite some time.)
Toss in some upgrades around the edges (Firewire, etc.) and this could be a damn nice update. I'd really like to see up to 2 of the Xserve removable drives used in a tower config - dual IDE controllers + striping could improve disk perf rather nicely for video applications, but after playing with an Xserve today, I'm not sure those drives would fit - they're damn big.
That'd just be icing on the cake. It'd help drive down costs of the drive modules due to the volume, and deliver added revenue to Apple instead of 3rd parties, but the prospect of a $350 60GB drive might cool people off a bit.
Let me start by saying that Motorola is probably not getting complete out of the CPU business. They don't want to burn that bridge, but at the same time they don't want to invest the resources right now to get the job done. As a result, they will honor their obligations most likely by "contracting" out CPU production to a third party. Initially this is likely to be an asian semi-conductor producer, but the long term future will be dependent on how well they do.
When motorola dropped the ball on G5 development, Apple stepped in and invested further resources. As a result, Apple has certain rights to some of Motorola's technology. Apple wants the chips, and they don't care who makes them or what they are called (other than for marketing purposes) as long as they get them. Motorola is most likely going to contract out production for the time being, in the meantime, Apple is "hedging its bets" by talking with IBM and possibly AMD to produce its futrue chips as well. Understand, these things are all in the early development phases, and any the winds could shift at any moment.
If AMD or IBM was contracted to produce chips, it would most likely be the the chips that are of Apple's and Motorola's joint design. i.e the G5 in some form. Motorola will keep just enough control, to allow them to bring it back in house, if their fortunes turn in their favor, otherwise they will themselves subcontract manufacture and development out or allow Apple to do it. Either way, they dispose of a part of business that is not making money for them, while they collect royalties on their intellectual property unless Apple buys it outright.
Now what will we see? Well I don't care what anyone else says, but we will see a faster motherboard with an improved implementation DDR RAM, and it will make a "significant" performance difference. There will also be faster chips, of at least 1.3 ghz.. The 1.6 does exist, but not in sufficient quantity to meet demand. 1.3 to 1.4 would be the highest top end to hope for, but combined with the new mobo, the performance increase will be substantial.
When will we see this? Well time is a nebulous beast, and difficult to tame. Steve wants it for MWNY. There is a chance of delays, but I think still before September.
What will we most likely see? Okay, here in a fit of optimism, I will stick my neck out. If you end up with me at chopping block, be merciful and chop quickly.
I predict the top end to be around 1.4 ghz. Not exactly that, but around that. Mobo speed on the top end will by fall be 333mhz. The chips will have 512k on chip L2 cache per processor running at processor speed, backed up by 2 mb of L3 cach per process. These puppies will scream.
The new iDevice will be bit of a letdown for Apple Loyalists. It just won't be gee-whiz enough, still it will do okay in the marketplace where it will attract new customers to apple. It may be capable of working with multiple computer platforms, but of course, Apple will have the complete package that works seamlessly together.
-----------
...what is that shadow? why, why, wh..... It looks (gulp!) like a really, really, huge ax..... (tremble).
<strong>This is the strongest case yet for Apple not having a massive update in July, and I will tell you what it will be. The fact that people are alleging insider trading on Apple due to the fact that so many execs, filed and sold their stock before their announcement about a sales shortfall.
These same execs that "didn't know" about the shortfall and selling their stock before it was announced also know if there is a real update in July that would cause the same said stock to rebound.
Based off of that I believe we will have the iMac staying with GF2mx and being bumped to 1 ghz across the line with only the drives and ram changing the price. They will also get a 133 mhz bus and a price drop. eMac will also get 1 ghz across the line, drives/ram changing price points, but will stay at 100 mhz bus.
The PowerMac line will get dual processors at 933, 1 ghz, and 1.2 ghz. Superdrives will be standard. Radeon 32 meg 7500 will remain on the low end and mid will keep 64 meg GF4mx and the top come standard with 128 meg Radeon 8500.
Motorola will not have created a true DDR solution for Apple. Apple will have engineered around this with the same solution we saw in the xserve only now they will have had a little more time to improve it a bit more. Steve will explain that the DDR implementation is DDR all the way up to the processor just like in the xserve. However on the xserve the bus to the processor is still sdr-133 that is all the G4 will support.
Apple will explain that their new dual processor machines now have two cpu buses instead of just one. The two buses are independent and can each access the DDR solution at their full sdr133 speed.
Apple will also include ATA-RAID capability in the new PowerMacs, but will not be included in on the low end to help seperate the lines a bit. (dual 933 to dual 1 gig isn't much of a jump, but dual 1 gig to 1.2 is pretty large)
Steve will announce that the prices have been dropped $300 across the board. ($200 on the low end) (Making all the folks who bought at the old price points feel like complete asses) He will also point out (just as he did before) that the previous high end Mac is now available with better throughput and drive options for $1000 less than it was just a week before.
So new pricing
$1399
dual 933
Radeon 7500w/32 megs
superdrive
256 ddr ram
60 gig HD (no raid option)
$1999
dual 1 gig
Geforce4mx w/64 megs
superdrive
512 megs ddr ram
120 gig HD (possibility of raid option by adding drive or also added via Applestore)
$2699
dual 1.2 gig
Radeon 8500 w/128 megs
512 megs ddr
superdrive 2 (just announced today in press release)
dual 120 gig HD(with 8 meg cache) preconfigured in fast array.
It will be enough for now and will raise PowerMac sales up to their old levels but Apple will sell less of the high end machine and thus make a smaller profit.
I have no insider info, I just know Apple, and have been a Mac user since 1990.
Trumptman
[ 06-28-2002: Message edited by: trumptman ]</strong><hr></blockquote>
I think that you are very near from the truth. Considering the rumors and the good info from Eskimo (good : reliable) i think your prediction will be ok.
But i will put some minor restriction : i think that they will be still a single powermac computer (let's say a 933 mhz one) and two dualies the 1 ghz one and a 1,13 one or a 1,2 ghz one (it depends on the progress of mot ).
The mpx bus will be still a 133 mhz one (the 166 mhz memory module are not enough common), they will use the same memory controller than the xserve , so they will use DDR ram.
In the future i see two lines : a line of consumer chips with sahara and sahara 2 chips for the consumers and powerbook line, and the proline with a single core power4 with an altivec like unit.
In the near future perhaps Mot can do an ultimate G4 developpement a G4 blended with G5 stuff like I/O and memory support for DDR ram. 2002 seems to be the year of transition. The year of change will be 2003.
You're basing that on the assumption that processor speed is all anyone cares about. It would never even occur to me to switch to peecees just because they're running 3HGz processors. Okay, I'm only responsible for a dozen Macs (not including my own machines), but they will all be replaced by new Macs next year. Most of our current Macs are single G4 450s and they're already faster machines than the people using them need for the work they do. But we're on a three-year upgrade schedule and next year we get new machines.
Sure, I want a G5 for myself, but the one thing people in my office care about MOST is...the size of their monitor. Nobody (except me) could tell you what processor they have, or how fast it is, or whether it's faster than the 8500 it replaced, but man, they can tell you that they have a 17-inch monitor and it's SO much better than their old one.
There are a LOT of Mac users who don't know anything about processors or clock speed or DDR RAM. They don't compare the specs of Macs to peecees because they don't know enough about computers to understand what they're comparing. What they know is: Macs are fun and easy to use. Peecees are complicated and crash a lot. EVERYONE in my office who buys a computer to use at home buys a Mac. They always ask my advice about what model to buy, but nobody has ever said to me "Gee, this Dell runs at 2.1GHz, should I consider getting one of those instead of a new Mac?"
However, the problem is NOT existing mac users, it's professionals who have no platform loyality. These pros will use whatever gets the job done fastest, because time is money. The advantage of OS X counts for a bit, but it doesn't make up for the gross performance differential between Wintels and Powermacs.
Furthermore, Apple needs to expand their market share. They MUST win over rednecks/morons for this to happen, and these types do care about GHz, very much so. We're talking about the sorts who would just as soon overbore their small block chevy V8 on a saturday as play Quake, or drop a radical cam into their Mustang rather than burn a CD of Poison's greatest hits. Apple's got to win over these idiots and it's not going to happen with 800 MHz cutesy computers.
<strong>Let me start by saying that Motorola is probably not getting complete out of the CPU business. They don't want to burn that bridge, but at the same time they don't want to invest the resources right now to get the job done. As a result, they will honor their obligations most likely by "contracting" out CPU production to a third party. Initially this is likely to be an asian semi-conductor producer, but the long term future will be dependent on how well they do......When motorola dropped the ball on G5 development, Apple stepped in and invested further resources. As a result, Apple has certain rights to some of Motorola's technology. Apple wants the chips, and they don't care who makes them or what they are called (other than for marketing purposes) as long as they get them. Motorola is most likely going to contract out production for the time being...........If AMD or IBM was contracted to produce chips, it would most likely be the the chips that are of Apple's and Motorola's joint design. i.e the G5 in some form. Motorola will keep just enough control, to allow them to bring it back in house, if their fortunes turn in their favor, otherwise they will themselves subcontract manufacture and development out or allow Apple to do it. Either way, they dispose of a part of business that is not making money for them, while they collect royalties on their intellectual property unless Apple buys it outright.
</strong><hr></blockquote>
Be a nice deal for them too if they retain rights to seel the proc in the embedded space. Apple pays for getting the chips fabbed right and Moto has a new high end chip for marketing to makers of super-routers and the ilk.
<strong>That's an interesting point.
However, the problem is NOT existing mac users, it's professionals who have no platform loyality. These pros will use whatever gets the job done fastest, because time is money. The advantage of OS X counts for a bit, but it doesn't make up for the gross performance differential between Wintels and Powermacs.
Furthermore, Apple needs to expand their market share. They MUST win over rednecks/morons for this to happen, and these types do care about GHz, very much so. We're talking about the sorts who would just as soon overbore their small block chevy V8 on a saturday as play Quake, or drop a radical cam into their Mustang rather than burn a CD of Poison's greatest hits. Apple's got to win over these idiots and it's not going to happen with 800 MHz cutesy computers.</strong><hr></blockquote>
I'm sorry, why?
Surely there are enough sensible people out there that Apple don't need this particular segment of the market. The truth is that for the majority of people current machines are plenty fast enough, what they want is ease-of-use, reliability, and functionality. This is one of the reasons for the extended slow-down in computer sales, even corporations realise they don't need to upgrade their machines regularily any more. Since Apple are starting from such a small percentage of the market, they need only attack this portion of it to be able to massively increase their market share.
Personally, I want a much faster Mac, as I write scientific code that needs fast double precision floating point performance, which Apple can't yet deliver. I do, however, write the code on a Mac (B+W G3 350 O/C to 450) and run it on a PC, but I'm very much in a minority here. My field (mainly microwave/optoelectronic circuit analysis and design) is strictly PC, the only Mac programs I know of are the ones I compile for my own use. The Mac is just so much more pleasant to use, and for all my purposes bar running this code, quite adequate.
BTW, make up you mind whether Apple are bleeding market share, or failing to expand their market share.
Michael
<strong>Douglass Adams was a Mac freak. Now he's dead.</strong><hr></blockquote>
What does that have to do with future Mac hardware? <img src="confused.gif" border="0">
Re: 133 MHz MPX remaining unchanged and yet bandwidth problems being fixed. The problem with the fast G4 is that it can't read and write to memory fast enough (i.e. 850 MB/sec doesn't cut it). Motorola's own documentation shows that 850 is the MPX's limit, and they don't want to change it since the embedded crowd likes it the way it is. A 128-bit implementation is too expensive (and would constitute a change).
Obviously the only way that memory bandwidth could increase is by not using the MPX bus to connect to memory. Bringing the memory controller on-chip. Motorola has a DDR memory controller in the 8540, so they could use that and if the L3 cache interface is removed the pin count might not skyrocket so badly. I would have expected a RapidIO bus, but hey MPX works and its there already. This scheme would actually simplify embedded designs using the G4, so that might make Moto's other customers happy and would explain why Moto hasn't brought out a decent memory controller themselves (Apple's kick's their butt).
Another alternative would be to replace the L3 cache interface with an Intel-style FSB to an external memory controller, but this seems a little weird and considerably slower than the on-chip design.
Oh heck, maybe I should just PM you after all and put up with "ha ha, made you look".
Re: The need for fast machines. While it is true that there is a large group of Mac customers (realized and potential) for whom speed does not matter, there are also a whole bunch of markets that Apple clearly wants into that really really really care about performance. People to whom time is money. People to whom real-time is everything. Without at least competitive performance, Apple will have a very hard time competing no matter how good their software is.
If the speculation above is true, however, that ought to close the performance gap enough. A 20% MHz bump + duals + significant bandwidth improvement + I/O improvements + new GPUs on AGP 8x + Jaguar will make for a remarkable leap in performance. My money is on the table. If I were Apple my biggest concern would be how the heck to sell the existing PowerMac inventory!
[ 06-29-2002: Message edited by: Programmer ]</p>