G5 for every model ASAP would have to be the plan. I think a new imac may not be so far away, as the present one is running out of legs. Nice but over priced and cutting the price will not help enough to stop its slow death. Apple needs high margin product, not cheap shit product with high margin but quality product with high margin. The PC vedors keep lowering the price and screwing each other to death. That is not a smart long term game, because if you keep the same % margin and keep lowering the price without increasing unit sales you go broke. That was and is the gateway way of doing business and they have about 6 months to a year left. If you sell a unit for $500 and you make 15% GP on each unit thats only $75. If you sell a unit for $1200 and make 25% GP you make $300 per unit. To end up with the same $ GP you need to sell 4 times as many units. Dell is the only vendor who as been able to make this work.( Up untill now, but it must be getting harder to do} It is interesting that the clever people at HP are heading down this path, there PC business will start to loose plenty within about 6 months if they start selling lots of those cheap boxes. Apple know that they would never sell 4 times as many units if they sold a cheap box. The clones had that same effect of taking sales from high margin product and Mr Jobs understood that and good bye clones. My point you ask? High margin product has to offer more value to the customer. What do the customers think is more value? Better design, lots of free s/ware, good buying enviroment and bang for your buck. So ASAP Apple will push a G5 into everything they can, with 64 bit goodness as the next big myth, but Apple leading the charge if they can.
Matsu mentioned Gateway as one of the PC manufactures with way lower prices. I may be wrong here but I think Apple has a larger market share then Gateway and therefore larger economies of scale. The argument is constantly made for lowering prices to drive market share and the economies of scale.
Gateway can tap into massive economies of scale, because they're only one buyer of the parts they use. Intel sells zillions of P4s and motherboards, and that's what Gateway can piggyback on. Apple is the only purchaser of Apple motherboards.
They're getting slaughtered, but that's another issue.
Quote:
Apple does have nice margins on it's hardware and it seems as though they can have some flexibility in dropping prices.
Since Apple has been breaking even as an operation - at best - they need the margins they have right now, and they don't have all that much room. That's what the iPod and the iTMS are really about: Alternate sources of revenue that can tap into a much larger market. That will give Apple more flexibility as far as hardware margins go.
When discussing slower chips we could be talking about two different things. We might be running higher rated parts at minimum supply voltage, or we might be running lower clock rate parts at normal supply voltage. Let's just take IBM's early remarks about power as an example, knowing that the real numbers today will be a little different.
...
So, the way we get to the lower clock rate makes a significant difference in power dissipation, 19 Watt or 28 Watts in this example. It might be helpful to think of each 970 chip having two different ratings for clock speed, one at normal supply voltage and one at minimum supply voltage.
What are you trying to say? (In relation to my cost comment.) All bad things that happen to the 970 can also happen to the 7457. For example, Motorola can't get good enough yield on 1.3V 1.2 or 1.3 GHz parts. What then?
Which really explains why upgrade cards cost an arm and a leg since they can't buy in 10k unit quantities, in addition to the daughterboard and socket (male end) integration.
Now consider the cost of the 970. A 2 GHz 970 costs Apple around $350 to $400 based on price differences between the 1.6, 1.8 and 2 GHz PowerMac G5. Lower clocked 970 processors will be less than that by at least half, if not more. I think the costs between equivalently clocked 7457 and 970 chips are going to be very close, and as the MHz gets higher, above 1.3 GHz, the 7457 will start to become much more expensive than 970 chips.
What are you trying to say? (In relation to my cost comment.) . . .
I'm not disagreeing with you, but pointing out something that may not be well understood by everyone. Using parts that fall below 1.6 GHz might cut cost, but the power dissipation will not be reduced as much as expected. On the other hand, running a part at two thirds its clock rating lets it operate at minimum supply voltage with lower power dissipation.
So if an iMac were to have a G5 running at 1.2 GHz, the least heat would be produced by using a 1.8 GHz part and cutting the supply voltage. Just a trade off between cost and power. A part rated at 1.2 GHz would dissipate 28 Watts, while a part rated at 1.8 GHz but run at 1.2 GHz would dissipate 19 Watts.
And for the 3rd time, 1 to 1.4 GHz PPC 970 CPUs may be cheaper than 1 to 1.4 GHz Moto 7457 chips. If that is the situation, Apple would be fools to use the 7457 on any of their hardware. Maybe this will force Moto to sell 7457 for less $100, I don't know, but their advertized prices are non-starters for Apple low end machines.
It's not that simple. In fact, we don't know (aren't allowed to know) anything about the G5s. The big questions remain: price, power consumption, time to move to 90nm.
Price is a very tricky thing because: Apple is the largest customer and, therefore, has some discounts; Apple has been participating in the chip design, which complicates possible deal clauses; Apple may have been investing unknown sums of money in the chip production; Apple may have prepayed, etc., etc., etc. There are too many unknowns which might influence the exclusive and confidential deal on G5 offered to Apple. I am even afraid to speculate.
Power consumption is still pretty much undocumented. The only trusted(?) place is that IBM's PDF with an estimate. That estimate, being made by the manufacturer, should not miss very much. On the other hand, we all saw the new PM's innards which hint at really hot hardware. Anyway, exact figures are a mystery until IBM has much more 970s than needed by both Apple and IBM. When IBM says they are ready sell PPC970 in supermarkets, we will know the precise specs.
Time required to move production to 90nm is a big secret of IBM's. At least until they are ready to do it. Whatever current G5's power dissipation is, G5 made on a new process will be easier to stuff into PowerBooks and iPods.
Which really explains why upgrade cards cost an arm and a leg since they can't buy in 10k unit quantities, in addition to the daughterboard and socket (male end) integration.
Now consider the cost of the 970. A 2 GHz 970 costs Apple around $350 to $400 based on price differences between the 1.6, 1.8 and 2 GHz PowerMac G5. Lower clocked 970 processors will be less than that by at least half, if not more. I think the costs between equivalently clocked 7457 and 970 chips are going to be very close, and as the MHz gets higher, above 1.3 GHz, the 7457 will start to become much more expensive than 970 chips.
I don't believe that Apple pay anywhere near $300 for a 1.25 GHz G4, as their price for upgrading from a single 1.25GHz G4 with 1MB L3 to a dual with 4MB L3 is only $300. This puts quite a tight upper limit on the prices paid by Apple for both of these components.
Power consumption is still pretty much undocumented. The only trusted(?) place is that IBM's PDF with an estimate.
There are at least 2 more documents from IBM. One is Here
I can not find the other one, but it is a very large pdf file thoroughly describing the 970 and has been posted some where here and also over at ArsTechnica. I'm too busy to look it up now, maybe later.
I don't believe that Apple pay anywhere near $300 for a 1.25 GHz G4, as their price for upgrading from a single 1.25GHz G4 with 1MB L3 to a dual with 4MB L3 is only $300. This puts quite a tight upper limit on the prices paid by Apple for both of these components.
Ok, this makes me feel better at the possibility of really cheap G4 machines, in the sub-1000 dollar category. Thanks. It probably puts the 1.25 GHz 7455, and 7457, around $100. Now we really need to find out what 1 to 1.4 GHz PPC 970 costs.
Well, IBM slipped too. Gobi will not be here til later this year I believe. The following is a quote from Naked Mole Rat a while ago, saying it would be here by now.
Quote:
As has been reported in part by other Mac-o-philic worthies, the company has now navigated beyond its Sahara generation of PowerPC G3 to a remap dubbed Colorado. Next stop: Gobi, which is expected to reach full production during the first calendar quarter of 2003.
From the Appleinsider report, it looks like iMac may stay G4 for a while. However, when the G5+ arrives next year, I bet it goes G5. It also sounds like IBM may wait for the G5+ for their blade servers too, if I interpreted what I read correctly. Possibly the 130 nanometer G5 runs too hot for use in blade servers.
Why is heat and power consumption such a concern for a desktop? Granted, the iLamp is a tight litle closed box, so the next iteration of the iMac will have to account for this, but since everyone acknowledges Ives to be a 'genius', then what's the problem?
Why is heat and power consumption such a concern for a desktop? Granted, the iLamp is a tight litle closed box, so the next iteration of the iMac will have to account for this, but since everyone acknowledges Ives to be a 'genius', then what's the problem?
Ives isn't the only variable here: He's an industrial designer, not an engineer. Heat dissipation is pure physics, the solution is pure engineering, and if the iMac is to remain as compact as it is then heat will remain a concern. If it's to keep its price, then the difficulty with which it can be cooled will remain a concern. Power consumption only matters because it absolutely determines how hot the part is - although Energy Star compliance is always nice.
Comments
Originally posted by tink
Matsu mentioned Gateway as one of the PC manufactures with way lower prices. I may be wrong here but I think Apple has a larger market share then Gateway and therefore larger economies of scale. The argument is constantly made for lowering prices to drive market share and the economies of scale.
Gateway can tap into massive economies of scale, because they're only one buyer of the parts they use. Intel sells zillions of P4s and motherboards, and that's what Gateway can piggyback on. Apple is the only purchaser of Apple motherboards.
They're getting slaughtered, but that's another issue.
Apple does have nice margins on it's hardware and it seems as though they can have some flexibility in dropping prices.
Since Apple has been breaking even as an operation - at best - they need the margins they have right now, and they don't have all that much room. That's what the iPod and the iTMS are really about: Alternate sources of revenue that can tap into a much larger market. That will give Apple more flexibility as far as hardware margins go.
Originally posted by snoopy
When discussing slower chips we could be talking about two different things. We might be running higher rated parts at minimum supply voltage, or we might be running lower clock rate parts at normal supply voltage. Let's just take IBM's early remarks about power as an example, knowing that the real numbers today will be a little different.
...
So, the way we get to the lower clock rate makes a significant difference in power dissipation, 19 Watt or 28 Watts in this example. It might be helpful to think of each 970 chip having two different ratings for clock speed, one at normal supply voltage and one at minimum supply voltage.
What are you trying to say? (In relation to my cost comment.) All bad things that happen to the 970 can also happen to the 7457. For example, Motorola can't get good enough yield on 1.3V 1.2 or 1.3 GHz parts. What then?
Part From Price Apple price
($) (guessing)
--------------- ------ ------ ------
XPC7455RX867LC FUTURE Act. 275.00 130
XC7455ARX1000LF AVNET 325.30 160
XC7455BRX1000PF NEWARK 373.61 160
XC7455BRX1250PF NEWARK 507.74 300
XC7455BRX1400PF NEWARK 507.71 350
xxx7457xx1000xx Moto PR 189 100
Which really explains why upgrade cards cost an arm and a leg since they can't buy in 10k unit quantities, in addition to the daughterboard and socket (male end) integration.
Now consider the cost of the 970. A 2 GHz 970 costs Apple around $350 to $400 based on price differences between the 1.6, 1.8 and 2 GHz PowerMac G5. Lower clocked 970 processors will be less than that by at least half, if not more. I think the costs between equivalently clocked 7457 and 970 chips are going to be very close, and as the MHz gets higher, above 1.3 GHz, the 7457 will start to become much more expensive than 970 chips.
Originally posted by THT
What are you trying to say? (In relation to my cost comment.) . . .
I'm not disagreeing with you, but pointing out something that may not be well understood by everyone. Using parts that fall below 1.6 GHz might cut cost, but the power dissipation will not be reduced as much as expected. On the other hand, running a part at two thirds its clock rating lets it operate at minimum supply voltage with lower power dissipation.
So if an iMac were to have a G5 running at 1.2 GHz, the least heat would be produced by using a 1.8 GHz part and cutting the supply voltage. Just a trade off between cost and power. A part rated at 1.2 GHz would dissipate 28 Watts, while a part rated at 1.8 GHz but run at 1.2 GHz would dissipate 19 Watts.
Originally posted by THT
And for the 3rd time, 1 to 1.4 GHz PPC 970 CPUs may be cheaper than 1 to 1.4 GHz Moto 7457 chips. If that is the situation, Apple would be fools to use the 7457 on any of their hardware. Maybe this will force Moto to sell 7457 for less $100, I don't know, but their advertized prices are non-starters for Apple low end machines.
It's not that simple. In fact, we don't know (aren't allowed to know) anything about the G5s. The big questions remain: price, power consumption, time to move to 90nm.
Price is a very tricky thing because: Apple is the largest customer and, therefore, has some discounts; Apple has been participating in the chip design, which complicates possible deal clauses; Apple may have been investing unknown sums of money in the chip production; Apple may have prepayed, etc., etc., etc. There are too many unknowns which might influence the exclusive and confidential deal on G5 offered to Apple. I am even afraid to speculate.
Power consumption is still pretty much undocumented. The only trusted(?) place is that IBM's PDF with an estimate. That estimate, being made by the manufacturer, should not miss very much. On the other hand, we all saw the new PM's innards which hint at really hot hardware. Anyway, exact figures are a mystery until IBM has much more 970s than needed by both Apple and IBM. When IBM says they are ready sell PPC970 in supermarkets, we will know the precise specs.
Time required to move production to 90nm is a big secret of IBM's. At least until they are ready to do it. Whatever current G5's power dissipation is, G5 made on a new process will be easier to stuff into PowerBooks and iPods.
So, who knows?
Originally posted by THT
I've found some 7455 CPU prices:
Code:
Part From Price Apple price
($) (guessing)
--------------- ------ ------ ------
XPC7455RX867LC FUTURE Act. 275.00 130
XC7455ARX1000LF AVNET 325.30 160
XC7455BRX1000PF NEWARK 373.61 160
XC7455BRX1250PF NEWARK 507.74 300
XC7455BRX1400PF NEWARK 507.71 350
xxx7457xx1000xx Moto PR 189 100
Which really explains why upgrade cards cost an arm and a leg since they can't buy in 10k unit quantities, in addition to the daughterboard and socket (male end) integration.
Now consider the cost of the 970. A 2 GHz 970 costs Apple around $350 to $400 based on price differences between the 1.6, 1.8 and 2 GHz PowerMac G5. Lower clocked 970 processors will be less than that by at least half, if not more. I think the costs between equivalently clocked 7457 and 970 chips are going to be very close, and as the MHz gets higher, above 1.3 GHz, the 7457 will start to become much more expensive than 970 chips.
I don't believe that Apple pay anywhere near $300 for a 1.25 GHz G4, as their price for upgrading from a single 1.25GHz G4 with 1MB L3 to a dual with 4MB L3 is only $300. This puts quite a tight upper limit on the prices paid by Apple for both of these components.
michael
Originally posted by costique
Power consumption is still pretty much undocumented. The only trusted(?) place is that IBM's PDF with an estimate.
There are at least 2 more documents from IBM. One is Here
I can not find the other one, but it is a very large pdf file thoroughly describing the 970 and has been posted some where here and also over at ArsTechnica. I'm too busy to look it up now, maybe later.
Originally posted by mmicist
I don't believe that Apple pay anywhere near $300 for a 1.25 GHz G4, as their price for upgrading from a single 1.25GHz G4 with 1MB L3 to a dual with 4MB L3 is only $300. This puts quite a tight upper limit on the prices paid by Apple for both of these components.
Ok, this makes me feel better at the possibility of really cheap G4 machines, in the sub-1000 dollar category. Thanks.
Originally posted by I, Fred
Of course, instead of redesigning the iMac to use the G5, we could just wait, and wait, and wait, and wait for Motorola......
Well, IBM slipped too. Gobi will not be here til later this year I believe. The following is a quote from Naked Mole Rat a while ago, saying it would be here by now.
As has been reported in part by other Mac-o-philic worthies, the company has now navigated beyond its Sahara generation of PowerPC G3 to a remap dubbed Colorado. Next stop: Gobi, which is expected to reach full production during the first calendar quarter of 2003.
From the Appleinsider report, it looks like iMac may stay G4 for a while. However, when the G5+ arrives next year, I bet it goes G5. It also sounds like IBM may wait for the G5+ for their blade servers too, if I interpreted what I read correctly. Possibly the 130 nanometer G5 runs too hot for use in blade servers.
Originally posted by I, Fred
Why is heat and power consumption such a concern for a desktop? Granted, the iLamp is a tight litle closed box, so the next iteration of the iMac will have to account for this, but since everyone acknowledges Ives to be a 'genius', then what's the problem?
Ives isn't the only variable here: He's an industrial designer, not an engineer. Heat dissipation is pure physics, the solution is pure engineering, and if the iMac is to remain as compact as it is then heat will remain a concern. If it's to keep its price, then the difficulty with which it can be cooled will remain a concern. Power consumption only matters because it absolutely determines how hot the part is - although Energy Star compliance is always nice.