You?re talking to an in-and-out Mac fanatic. Where I live, 99.999999999999999 use Windows. I want to support Apple. To me, what you?re saying almost sounds treasonous. There must be a way short of abandoning the platform for someone like me to affect Apple. Am I really so exceptional, and Am I really asking for soooo much?
Three Northbridges, all with a GigaWire master controller and connection to the Southbridge.
N1:
Dual-970 Support (600/750/900MHz Bus)
Dual-channel DDR (266/333/400MHz)
Gigabit Ethernet MAC
AGP 8X
N2:
Single-970 Support (600/750/900MHz Bus)
Dual-channel DDR (266/333/400MHz)
Gigabit Ethernet MAC
AGP 8X Controller
N3:
Dual-970 Support (600/750/900MHz Bus)
Dual-channel DDR (266/333/400MHz)
Dual Gigabit Ethernet MACs
64-bit PCI Controller
A common Southbridge with GigaWire, S1:
800Mb FireWire (connected to PHY)
4-USB2 ports
Serial ATA
CardBus for AirPort, BootROM, etc.
A 24-bit Analog-out Audio Chip with GigaWire, A1:
A Line-In port
Combined Line-Out and Front Speaker port
A Center/Sub port
A Rear speaker port
A Headphone port
An Integrated Chip, the i1 without GigaWire.
Single-970 support
Single-channel DDR266/333/400 support
Gigabit Ethernet
AGP 8X
800Mb FireWire (connected to PHY)
4-USB2 ports
Serial ATA
Line Out/Front, Center/Sub, Rear, Headphone
Cardbus for Airport, BootROM etc.
---------------
GigaWire transmits USB, ATA and PCI data over 133, 266, 533 and 1066 MB/s connections. A Master Controller assigns an address to every USB, ATA and PCI port controlled by a GigaWire Secondary Controller. Data then moves in a chain up or down the addresses.
There are 8 data lines and 2 grounds. The 8 data lines can be upstream or downstream. An upstream and downstream line communicates what lines will be up and what will be down in the next transmittion simultanious with the current 2Hz transmittion. The up/down lines also communicate the address of the (up to 2-byte) packet. A packet is broadcasted up or down the chain until a device with that address recieves it, and "lifts" it out of the chain.
GigaWire is basically for transmitting different kinds of data over a chain of 12 tracers/wires. GigaWire can be used internally or externally. The northbridge and southbridge for instance appear as PCI devices, and the southbridge can send USB data to the northbridge to the AGP port.
---------------
The Power Mac is replaced by a stackable modular design (with about an 8"x8" footprint) at the introduction of the PowerPC 970 in mid-2003. The modules (except for the Power Supply) use GigaWire to communicate.
As standard the Xmac (X being "I can't think of a good name so I'll use an X") is just a Core and Power Supply.
Dual 1.2GHz is $1499, Dual 1.5GHz is $1999, Dual 1.8GHz is $2999.
A Power Supply module with 6x 12V outputs and a USB port.
A UPS module with a battery, 6x 12V outputs and a USB port.
A Core module with the motherboard (with un-upgradable dual CPUs, 4x RAM slots and an AGP 8X slot), one CD and one HDD bay. The motherboard uses the N1, S1 and A1 chips.
A PCI module with a Gigabit/PCI secondary controller and 6 or so PCI slots.
A Hard Disk module with a Gigabit/ATA secondary controller and 4 Bays/4 ATA Channels.
A PCI/Hard Disk module with a Gigabit/PCI secondary controller, 4 Bays and 2 PCI cards.
A CD Module with a Gigabit/ATA secondary controller and 2 Bays/1 ATA channel.
-------------
Apple Display Connector and Pro Speaker Minijack are thrown out, instead using the 12V connectors on the PSU or UPS for power.
-------------
In 2004 the Xserve G5, PowerBook G5 and iMac G5 are introduced, using at 90nm PowerPC 970.
The PowerBook G5 uses the N2, S1 and A1 chips.
The Xserve uses the N3 and S1 chips.
The iMac G5 uses the i1 chip.
The iBook G5, introduced in 2005, will use the i1 chip.
Probably at a premium to the current ones since there will be less reason to upgrade in the fure. Why not just keep the 970 in a regular tower. The Power Mac G4 is already one of the most easiest to upgrade computers ever. A modular design will just add costs.
It would probably cost less. No need for PCI slots, less materials, etc.
It would cost more when you add more modules to the base core/power supply, but it gives people the freedom to buy Power Macs with the degree of expandability they want.
Also, my speculation isn't meant to be taken too seriously. It's just brainstorming the future.
Regarding the PowerMac enclosure, here is something to think about. We are in an economic down turn, and Apple just posted a loss. They will want to invest product design dollars in things that greatly increase sales, like totally-new products. When the IBM 970 comes out in the PowerMacs, these PowerMacs will sell extremely well in the existing enclosure. Making a new enclosure for it will not affect sale much, if at all. So, I'm more or less expecting the new PowerMacs to look the same as they do now, with just a few cosmetic changes.
What PC^Killa is asking for isn't too much. I think it is reasonable to ask for a moderately priced Mac, that isn't an AIO form.
Split off the monitor and add just a little upgradablity, and I think you have a pretty good starter model. This is the niche that should have been filled by the Cube. The Cube failed for a number of reasons, not the least of which were extremely over priced and non-standard upgrade path (having to buy Mac only parts is limiting enough, having to buy Cube-specific parts was crazy). Building the same machine, but in a form-factor that would allow limited upgrades, and at a reasonable price, could be done without canabalizing high end sales. Apple tried with the cube, but positioned it as a PowerMac competitor, eating some says from their high margin cashcow. Position the new Cube lower-end, allowing standard, but limited upgrades (mebbe 2 PCI, AGP, 2 or 3 RAM slots). Picing could be set at the same as an iMac minus the cost of the monitor.
The comparison to elite cars is a good one, but staying only within that niche makes it difficult to enter new markets. If Apple wants to only stay with 5% of the market, great, stay with current line up. They have sold essentially the same choices for 5 years now (excepting the Cube failure), perhaps it is tie to try something new that might attract new buyers.
Three Northbridges, all with a GigaWire master controller and connection to the Southbridge.
N1:
Dual-970 Support (600/750/900MHz Bus)
Dual-channel DDR (266/333/400MHz)
Gigabit Ethernet MAC
AGP 8X
N2:
Single-970 Support (600/750/900MHz Bus)
Dual-channel DDR (266/333/400MHz)
Gigabit Ethernet MAC
AGP 8X Controller
N3:
Dual-970 Support (600/750/900MHz Bus)
Dual-channel DDR (266/333/400MHz)
Dual Gigabit Ethernet MACs
64-bit PCI Controller
</strong><hr></blockquote>
Why produce 3 northbridges with such minor differences? Economically and logistically horrible. The northbridge will require a lot of work, and a fancy process to handle a bus, or busses, with a 450MHz clock and 900MHz data rate. The design will be difficult to get right, and may well be quite expensive to produce, you certainly want to maximise the production volume of the part.
Use a single dual 900MHz bus northbridge with memory controller, and a 16 bit 800MHz (3.2GB/s unidirectional, 6.4GB/s bidirectional) Hypertransport link, or even 8 bit 900MHz (1.8/3.6) to the southbridge which does everything else.
Calculate required bandwidth:
[code]
AGP8x 1067 MB/s
ata133 133 MB/s
ata33 33 MB/s
PCI-64/66 533 MB/s
firewire2 100 MB/s
total 1867 MB/s
</pre><hr></blockquote>
Plenty of bandwidth for that link. The southbridge is a more standard part and changed more easily, as well as being able to put it further away, I suspect the northbridge will have to be very close to the processor(s) to get good performance.
I think the 17" iMac is already being positioned as the cube replacement. If the new Cube was offered at about 1600 it would compete with the lowe end PowerMac, the midrange iMac and the high end eMac. That would be four products all for about the same price. Already the eMac seems to be the real iMac and the iMac is now really just a Cube with a monitor.
The comparison to elite cars is a good one, but staying only within that niche makes it difficult to enter new markets. If Apple wants to only stay with 5% of the market, great, stay with current line up. They have sold essentially the same choices for 5 years now (excepting the Cube failure), perhaps it is tie to try something new that might attract new buyers.</strong><hr></blockquote>
Significantly, though, both Porsche and BMW are now making and selling SUVs. They're breaking out of their niche markets and going more mainstream. If people (Americans, specifically) are going to shell out money for these things, both of these companies have decided they want a piece of the pie. I know a lot of car enthusiasts groaned about how they were "selling out", but there's a lot of people who wouldn't touch a Lincoln Navigator that would buy a BMW X5 in a heartbeat. The "Apple" brand carries a lot of weight in the marketplace - they need to start cashing in on it, IMO.
The comparison to elite cars is a good one, but staying only within that niche makes it difficult to enter new markets. If Apple wants to only stay with 5% of the market, great, stay with current line up.
</strong><hr></blockquote>
I agree that Apple needs to get into more markets to increase market share. Some Mac users say Apple can do well with low market share. They typically cite how well elite autos do with low market share. The problem is software development. The Mac is more than just another make of computer. It is a whole platform that needs its own type of software, different from that used in over 90 percent of all personal computers. To succeed as a platform, the Mac must be in many markets, and have high enough sales to attract and keep developers. So yes, they should definitely begin to move into new areas, but selectively and with good strategy.
Significantly, though, both Porsche and BMW are now making and selling SUVs.</strong><hr></blockquote>
Amusingly the BMW approach was to BUY Land Rover. I don't see a competent-yet-reasonably-priced PC company outside of Apple's niche that might be worth buying.
Also, the problem with 'the car analogy' is the software developer business (as mentioned above). The lack of software developers always loomed under Mac OS 9... But I don't think it looms so much anymore. NeXt survived for years with a marketshare of what, 0.00%? Cocoa really is that easy to use for one. Second is the open underpinnings. Third is the fact that the underpinnings are unix-based.
Those three pieces together are pretty darn useful. It is easy to prove to yourself that they _are_ that useful also. There appears to be a LOT of interest from niches that have historically ranked Apple somewhere below maggot-ridden-meat. Even if the speed isn't there in the 'Power' line.
A 5x increase in SpecFP a full year from now should be worth a substantial increase in sales. Ignoring Quads or dual cores or exotic memory topologies. Just a straight dual-chip 970 @ 1.8 GHz should rock. At 50 million transistors the price should be right. Spend an extra little bit on going to interleaved memory and you are right in where the dual-Xeon range will be - and there's no way this will cost as much as a Xeon.
I wouldn't want another Cube priced at $1600. That puts it out of the entry level position, just as the high price did before. A replacement for the Cube, price at ~$1000, with similar specs to the entry level iMac sans monitor, and with some limited, standard size upgradability would only eat into the iMac line. But with no monitor and approximately the same price, the margins would probably be a lot higher, so losing a few imac sales would be offset by higher margins and hopefully, higher volume. I know everyone who pushes for a entry level tower says "I know someone who's buy one today", but I really do know people who would. My dad is looking to possibly replace his Sawtooth 450 in the next year, but doesn't want to drop $3000 Canadian. He is also put off by the AIO concept. A striped down tower, that he could have atleast 1 PCI, an extra drive and added memory would be all he would need, and would be enough for me to recommend it to him. Right now, he is considering a CPU upgrade instead, but that doesn't really help Apple. My friend is also wanting a new computer and has shown interest in a Mac, but can't afford a PowerMac and doesn't want an iMac...to limiting in his mind.
[quote]Originally posted by jante99:
<strong>I think the 17" iMac is already being positioned as the cube replacement. If the new Cube was offered at about 1600 it would compete with the lowe end PowerMac, the midrange iMac and the high end eMac. That would be four products all for about the same price. Already the eMac seems to be the real iMac and the iMac is now really just a Cube with a monitor.</strong><hr></blockquote>
. . . A striped down tower, that he could have at least 1 PCI, an extra drive and added memory would be all he would need . . .
</strong><hr></blockquote>
I do believe those would sell. Much smaller case, let's say just two PCI slots and room for one extra drive. Yes. Put in an IBM 970 and it could be Apple's top selling model.
Anyone ever consider that Rambus may be used in the PowerMacs that would use the 970 chips? It would make perfect sense seeing as the bandwidth would be enough to sustain the giant bandwidth at their disposal. But one thing in my mind precludes this: Rambus is power hungry in comparison to DDR. Therefore a laptop implementation would be difficult at best and the cost of designing a separate controller chip may force the 970 to be connected to DDR or a dual channel DDR set up. Unless something has changed with Rambus and it is much more conservative with power.
<strong>I do believe those would sell. Much smaller case, let's say just two PCI slots and room for one extra drive. Yes. Put in an IBM 970 and it could be Apple's top selling model.</strong><hr></blockquote>
I think that would sell too - but _two_ slots & a drive bay -> cannibalized "Power" line.
There's a LOT of towers out there with just one or two cards. If we're making a box to 1) sell well and 2) sell to people that would otherwise not buy a Mac.
The second internal drive bay is also one of the main 'features' of the towers. And it is one that even the iMac folks can get around with a FW enclosure -> skip it for the pizzabox so they lust after something better
Comments
----------------
Three PowerPC 970s
1.2 GHz, 1.5GHz and 1.8GHz
Three Northbridges, all with a GigaWire master controller and connection to the Southbridge.
N1:
Dual-970 Support (600/750/900MHz Bus)
Dual-channel DDR (266/333/400MHz)
Gigabit Ethernet MAC
AGP 8X
N2:
Single-970 Support (600/750/900MHz Bus)
Dual-channel DDR (266/333/400MHz)
Gigabit Ethernet MAC
AGP 8X Controller
N3:
Dual-970 Support (600/750/900MHz Bus)
Dual-channel DDR (266/333/400MHz)
Dual Gigabit Ethernet MACs
64-bit PCI Controller
A common Southbridge with GigaWire, S1:
800Mb FireWire (connected to PHY)
4-USB2 ports
Serial ATA
CardBus for AirPort, BootROM, etc.
A 24-bit Analog-out Audio Chip with GigaWire, A1:
A Line-In port
Combined Line-Out and Front Speaker port
A Center/Sub port
A Rear speaker port
A Headphone port
An Integrated Chip, the i1 without GigaWire.
Single-970 support
Single-channel DDR266/333/400 support
Gigabit Ethernet
AGP 8X
800Mb FireWire (connected to PHY)
4-USB2 ports
Serial ATA
Line Out/Front, Center/Sub, Rear, Headphone
Cardbus for Airport, BootROM etc.
---------------
GigaWire transmits USB, ATA and PCI data over 133, 266, 533 and 1066 MB/s connections. A Master Controller assigns an address to every USB, ATA and PCI port controlled by a GigaWire Secondary Controller. Data then moves in a chain up or down the addresses.
There are 8 data lines and 2 grounds. The 8 data lines can be upstream or downstream. An upstream and downstream line communicates what lines will be up and what will be down in the next transmittion simultanious with the current 2Hz transmittion. The up/down lines also communicate the address of the (up to 2-byte) packet. A packet is broadcasted up or down the chain until a device with that address recieves it, and "lifts" it out of the chain.
GigaWire is basically for transmitting different kinds of data over a chain of 12 tracers/wires. GigaWire can be used internally or externally. The northbridge and southbridge for instance appear as PCI devices, and the southbridge can send USB data to the northbridge to the AGP port.
---------------
The Power Mac is replaced by a stackable modular design (with about an 8"x8" footprint) at the introduction of the PowerPC 970 in mid-2003. The modules (except for the Power Supply) use GigaWire to communicate.
As standard the Xmac (X being "I can't think of a good name so I'll use an X") is just a Core and Power Supply.
Dual 1.2GHz is $1499, Dual 1.5GHz is $1999, Dual 1.8GHz is $2999.
A Power Supply module with 6x 12V outputs and a USB port.
A UPS module with a battery, 6x 12V outputs and a USB port.
A Core module with the motherboard (with un-upgradable dual CPUs, 4x RAM slots and an AGP 8X slot), one CD and one HDD bay. The motherboard uses the N1, S1 and A1 chips.
A PCI module with a Gigabit/PCI secondary controller and 6 or so PCI slots.
A Hard Disk module with a Gigabit/ATA secondary controller and 4 Bays/4 ATA Channels.
A PCI/Hard Disk module with a Gigabit/PCI secondary controller, 4 Bays and 2 PCI cards.
A CD Module with a Gigabit/ATA secondary controller and 2 Bays/1 ATA channel.
-------------
Apple Display Connector and Pro Speaker Minijack are thrown out, instead using the 12V connectors on the PSU or UPS for power.
-------------
In 2004 the Xserve G5, PowerBook G5 and iMac G5 are introduced, using at 90nm PowerPC 970.
The PowerBook G5 uses the N2, S1 and A1 chips.
The Xserve uses the N3 and S1 chips.
The iMac G5 uses the i1 chip.
The iBook G5, introduced in 2005, will use the i1 chip.
--------------
Barto
[ 10-19-2002: Message edited by: Barto ]</p>
Probably at a premium to the current ones since there will be less reason to upgrade in the fure. Why not just keep the 970 in a regular tower. The Power Mac G4 is already one of the most easiest to upgrade computers ever. A modular design will just add costs.
It would cost more when you add more modules to the base core/power supply, but it gives people the freedom to buy Power Macs with the degree of expandability they want.
Also, my speculation isn't meant to be taken too seriously. It's just brainstorming the future.
Barto
[ 10-18-2002: Message edited by: Barto ]</p>
<strong>I wish the Mac community didn?t behave like such a bunch of sheep.</strong><hr></blockquote>
It ain't sheep, Mika, it's hope.
Split off the monitor and add just a little upgradablity, and I think you have a pretty good starter model. This is the niche that should have been filled by the Cube. The Cube failed for a number of reasons, not the least of which were extremely over priced and non-standard upgrade path (having to buy Mac only parts is limiting enough, having to buy Cube-specific parts was crazy). Building the same machine, but in a form-factor that would allow limited upgrades, and at a reasonable price, could be done without canabalizing high end sales. Apple tried with the cube, but positioned it as a PowerMac competitor, eating some says from their high margin cashcow. Position the new Cube lower-end, allowing standard, but limited upgrades (mebbe 2 PCI, AGP, 2 or 3 RAM slots). Picing could be set at the same as an iMac minus the cost of the monitor.
The comparison to elite cars is a good one, but staying only within that niche makes it difficult to enter new markets. If Apple wants to only stay with 5% of the market, great, stay with current line up. They have sold essentially the same choices for 5 years now (excepting the Cube failure), perhaps it is tie to try something new that might attract new buyers.
<strong>This is pure speculation.
----------------
Three PowerPC 970s
1.2 GHz, 1.5GHz and 1.8GHz
Three Northbridges, all with a GigaWire master controller and connection to the Southbridge.
N1:
Dual-970 Support (600/750/900MHz Bus)
Dual-channel DDR (266/333/400MHz)
Gigabit Ethernet MAC
AGP 8X
N2:
Single-970 Support (600/750/900MHz Bus)
Dual-channel DDR (266/333/400MHz)
Gigabit Ethernet MAC
AGP 8X Controller
N3:
Dual-970 Support (600/750/900MHz Bus)
Dual-channel DDR (266/333/400MHz)
Dual Gigabit Ethernet MACs
64-bit PCI Controller
</strong><hr></blockquote>
Why produce 3 northbridges with such minor differences? Economically and logistically horrible. The northbridge will require a lot of work, and a fancy process to handle a bus, or busses, with a 450MHz clock and 900MHz data rate. The design will be difficult to get right, and may well be quite expensive to produce, you certainly want to maximise the production volume of the part.
Use a single dual 900MHz bus northbridge with memory controller, and a 16 bit 800MHz (3.2GB/s unidirectional, 6.4GB/s bidirectional) Hypertransport link, or even 8 bit 900MHz (1.8/3.6) to the southbridge which does everything else.
Calculate required bandwidth:
[code]
AGP8x 1067 MB/s
ata133 133 MB/s
ata33 33 MB/s
PCI-64/66 533 MB/s
firewire2 100 MB/s
total 1867 MB/s
</pre><hr></blockquote>
Plenty of bandwidth for that link. The southbridge is a more standard part and changed more easily, as well as being able to put it further away, I suspect the northbridge will have to be very close to the processor(s) to get good performance.
michael
<strong>
It ain't sheep, Mika, it's hope.</strong><hr></blockquote>
Hope?!!
What are you hoping for? A winning lottery ticket?
<strong>
The comparison to elite cars is a good one, but staying only within that niche makes it difficult to enter new markets. If Apple wants to only stay with 5% of the market, great, stay with current line up. They have sold essentially the same choices for 5 years now (excepting the Cube failure), perhaps it is tie to try something new that might attract new buyers.</strong><hr></blockquote>
Significantly, though, both Porsche and BMW are now making and selling SUVs. They're breaking out of their niche markets and going more mainstream. If people (Americans, specifically) are going to shell out money for these things, both of these companies have decided they want a piece of the pie. I know a lot of car enthusiasts groaned about how they were "selling out", but there's a lot of people who wouldn't touch a Lincoln Navigator that would buy a BMW X5 in a heartbeat. The "Apple" brand carries a lot of weight in the marketplace - they need to start cashing in on it, IMO.
<strong>
The comparison to elite cars is a good one, but staying only within that niche makes it difficult to enter new markets. If Apple wants to only stay with 5% of the market, great, stay with current line up.
</strong><hr></blockquote>
I agree that Apple needs to get into more markets to increase market share. Some Mac users say Apple can do well with low market share. They typically cite how well elite autos do with low market share. The problem is software development. The Mac is more than just another make of computer. It is a whole platform that needs its own type of software, different from that used in over 90 percent of all personal computers. To succeed as a platform, the Mac must be in many markets, and have high enough sales to attract and keep developers. So yes, they should definitely begin to move into new areas, but selectively and with good strategy.
<strong>
Hope?!!
What are you hoping for?</strong><hr></blockquote>
I'm just hoping a decent alternative to Windows can survive.
<strong>
Significantly, though, both Porsche and BMW are now making and selling SUVs.</strong><hr></blockquote>
Amusingly the BMW approach was to BUY Land Rover. I don't see a competent-yet-reasonably-priced PC company outside of Apple's niche that might be worth buying.
Also, the problem with 'the car analogy' is the software developer business (as mentioned above). The lack of software developers always loomed under Mac OS 9... But I don't think it looms so much anymore. NeXt survived for years with a marketshare of what, 0.00%? Cocoa really is that easy to use for one. Second is the open underpinnings. Third is the fact that the underpinnings are unix-based.
Those three pieces together are pretty darn useful. It is easy to prove to yourself that they _are_ that useful also. There appears to be a LOT of interest from niches that have historically ranked Apple somewhere below maggot-ridden-meat. Even if the speed isn't there in the 'Power' line.
A 5x increase in SpecFP a full year from now should be worth a substantial increase in sales. Ignoring Quads or dual cores or exotic memory topologies. Just a straight dual-chip 970 @ 1.8 GHz should rock. At 50 million transistors the price should be right. Spend an extra little bit on going to interleaved memory and you are right in where the dual-Xeon range will be - and there's no way this will cost as much as a Xeon.
[quote]Originally posted by jante99:
<strong>I think the 17" iMac is already being positioned as the cube replacement. If the new Cube was offered at about 1600 it would compete with the lowe end PowerMac, the midrange iMac and the high end eMac. That would be four products all for about the same price. Already the eMac seems to be the real iMac and the iMac is now really just a Cube with a monitor.</strong><hr></blockquote>
<strong>
. . . A striped down tower, that he could have at least 1 PCI, an extra drive and added memory would be all he would need . . .
</strong><hr></blockquote>
I do believe those would sell. Much smaller case, let's say just two PCI slots and room for one extra drive. Yes. Put in an IBM 970 and it could be Apple's top selling model.
<strong>I do believe those would sell. Much smaller case, let's say just two PCI slots and room for one extra drive. Yes. Put in an IBM 970 and it could be Apple's top selling model.</strong><hr></blockquote>
I think that would sell too - but _two_ slots & a drive bay -> cannibalized "Power" line.
The ranking of 'how many cards do I use':
0 >> 1 >> 2 >> 3 >> 4 >> expansion chassis..
There's a LOT of towers out there with just one or two cards. If we're making a box to 1) sell well and 2) sell to people that would otherwise not buy a Mac.
The second internal drive bay is also one of the main 'features' of the towers. And it is one that even the iMac folks can get around with a FW enclosure -> skip it for the pizzabox so they lust after something better