I've been trying to figure out a way for Apple to keep the "Mac Pro" machines in the $2000 to $3000 range, and there really isn't a good way.
Woodcrest systems are expensive. Apple can probably sell a 2S Xeon 5140 (2.33 GHz) in the neighborhood of $2500 to $2700 (maybe), and should be able to perform about the same as the G5 Quad in FPU while crushing it integer. However, a hypothetical 2.93 GHz Kentsfield system is going to smoke it, at least in Apple's market.
For Apple, the gearing for a Kentsfield system is all favorable compared to a 2S Woodcrest system:
1. A 2.66 GHz Kentsfield will be cheaper than 2 Xeon 5150 CPUs (2.66 GHz Woodcrests). It'll be 5 to 10% slower if using the same memory, but they won't be.
2. Kentsfield will use DDR2 RAM which is half as cheap as FB-DIMMs
3. DDR2 RAM is faster for Apple's "Mac Pro" market and its class of applications which should gain that 5 to 10% back.
4. Kentsfield motherboards will be half as cheap as Woodcrest boards
The only loss is that Kentsfield systems will only have 8 memory slots at most for 16 GB memory using 2 GB DIMMs, while Woodcrest system can have 16+ memory slots for 32 to 64 GB memory. Not a big loss for Apple.
Hence, Kentsfield systems can occupy the $2300 to $2700 mid-range, Conroe can occupy the sub-$2000 space, while 2S Woodcrest the $3000+ range. If not, there will be a big hole in the middle in Apple's Mac Pro lineup, or the its low-end will be outperformed by Conroe systems costing $500 less. The only advantage to such Woodcrest systems is large memory capacity. Doubt Apple has the market to support that.
I don't think there is an onsite service for consumer purchases if that's what your asking.
To be clear, melgross was comparing the PowerMac's available (optional) support with what comes standard with Dell's Precision series. Precision is not a consumer purchase by any stretch of any imagination, and I really don't think PowerMac qualifies as consumer either, so I don't see how that reasoning fits.
You're talking about a niche market within a niche market: Niche^2. It's unjustifiable...it's like me asking for a built-in NES cartridge loader so I can download the the many cartridge ROMs I collect to my computer so I can play them on an emulator.
Just buy the PCI cards that you need and be done with it. No need to put 2 ports that will be used by 1% of the already very small Mac user base.
The question is what is meant by niche. The biomedical industry is one that Apple has been pushing for years, and in which they have done fairly well. But there are catagories that they are not in, because of issues such as this. The same is true for other industries they are involved in.
It's too easy, when sitting down at a computer, to think that if we don't need something, that others don't as well. Apple wants very badly to get a big chunk of the scientific computing market. In large parts of that market, they had a bigger marketshare than they do now. One reason is the problem of control interfacing.
As I mentioned earlier, there were a number of companies that offered these serial cards to the Mac market for this very purpose. But Apple removed much of the necessary software from the OS when they went to OS X. Thinking, I suppose, the same thing that is being said here. But the markets didn't follow. They still use serial control. And Apple is lagging there. If we just had some good hi end boards it would help. But, Apple has to understand this, and they don't seem to.
We have discussed many times the fact that Apple doesn't always know what's best for them.
The same think is true for SCSI. While the use for it has diminished, it has caused problems in some areas, such as the hi end printer markets, where many machines run off SCSI. The big problem here is not the ports, or cards, it is simply that in 10.3, Apple removed some important software from the OS which has caused problems ever since. There was no need to remove it, they just wanted to push people off it, and on to Firewire.
Woodcrest systems are expensive. Apple can probably sell a 2S Xeon 5140 (2.33 GHz) in the neighborhood of $2500 to $2700 (maybe), and should be able to perform about the same as the G5 Quad in FPU while crushing it integer. However, a hypothetical 2.93 GHz Kentsfield system is going to smoke it, at least in Apple's market.
Apple can sell a 2S 2.0 Woodcrest system for $2000-$2200:
$660 for processors
$90 for HDD (250 GB maybe)
$400 MoBo
$175 Video Card (7600GT)
$150 RAM (2x512 FB-DIMM 533)
$200 Case
Sub-Total: $1675
That's paying newegg prices for the Video Card, the RAM and the HDD, and 1000-lot prices for the processors. The whole system could cost Apple $1600 when you add in BT, AirPort, FireWire, IR port, and whatever I'm forgetting if Apple gets good deals.
A quad-2.0 will hold it's own against most dual-Opteron boards, and will fare well processor-wise against 4x4. It'll avoid competing with Dell XPS's, and crush all Precision workstations on price.
Apple can sell a 2S 2.0 Woodcrest system for $2000-$2200:
$660 for processors
$90 for HDD (250 GB maybe)
$400 MoBo
$175 Video Card (7600GT)
$150 RAM (2x512 FB-DIMM 533)
$200 Case
Sub-Total: $1675
That's paying newegg prices for the Video Card, the RAM and the HDD, and 1000-lot prices for the processors. The whole system could cost Apple $1600 when you add in BT, AirPort, FireWire, IR port, and whatever I'm forgetting if Apple gets good deals.
A quad-2.0 will hold it's own against most dual-Opteron boards, and will fare well processor-wise against 4x4. It'll avoid competing with Dell XPS's, and crush all Precision workstations on price.
When these prices are put out here, there are important parts that are missing. What about the cost to manufacture the product? What about the cost of R&D that is figured in? What about the cost of running the company, which also must be figured in? What about warranty costs, etc?
The problem with breakdowns like those done by iSupply is that all they do is to give the cost of the parts, and sometimes an undervalued assembly factor. When the breakdowns are done here, they copy those inadequate numbers. As one who did manufacturing, I can tell you that these part numbers aren't sufficient to give the true cost.
Apple can sell a 2S 2.0 Woodcrest system for $2000-$2200
They could, but who would want to buy that when you can get a 2.93 GHz Conroe system for $2000 that'll be about as fast on multithreaded apps and 50% faster for single threaded apps, which constitute a large majority of apps. A 2.67 GHz Conroe may even be a better buy than a 2S 2 GHz Woodcrest.
2S systems are still very specific in nature, specific to an application. It only pays off for heavy usage of specific apps or markets and when they are clearly better than 1S systems. That means, the floor for Apple's 2S Woodcrest systems needs to have a 2.67 GHz Woodcrest in it, maybe a 2.33 GHz; otherwise, Conroe systems will be the clear price/performance leader with 2.67 and 2.93 GHz systems selling for under $2000.
The only other big advantage is 32 to 64 GB memory support. People who can afford 16 GB, 32 GB or 64 GB of memory aren't going to be worried about the $1000 price difference between a 2S 2.66+ GHz system and a 2S <2.0 GHz system.
When these prices are put out here, there are important parts that are missing. What about the cost to manufacture the product? What about the cost of R&D that is figured in? What about the cost of running the company, which also must be figured in? What about warranty costs, etc?
The problem with breakdowns like those done by iSupply is that all they do is to give the cost of the parts, and sometimes an undervalued assembly factor. When the breakdowns are done here, they copy those inadequate numbers. As one who did manufacturing, I can tell you that these part numbers aren't sufficient to give the true cost.
That about sums up what I was going to say. There is a lot more to building, and selling a computer than it's wholesale parts price.
When these prices are put out here, there are important parts that are missing. What about the cost to manufacture the product? What about the cost of R&D that is figured in? What about the cost of running the company, which also must be figured in? What about warranty costs, etc?
The problem with breakdowns like those done by iSupply is that all they do is to give the cost of the parts, and sometimes an undervalued assembly factor. When the breakdowns are done here, they copy those inadequate numbers. As one who did manufacturing, I can tell you that these part numbers aren't sufficient to give the true cost.
I recognize that. Peg a price on that. Say $100 to cover warranty and manufacturing? R&D and "running the company" fall under "fixed costs" and they are what "margins" (Marginal Revenue - Marginal Cost) offset. I estimated the hardware at $1600 (which you didn't dispute). Let's tack on $100 to cover manufacturing and warranty (I doubt it's that high, but let's pretend).
So then Apple's marginal cost for each Mac Pro they build is $1700. Their marginal revenue is the price they sell it for. Let's call it $2100. That means they make $400 on each low-end Mac Pro. That's a margin of 19%. That $400 helps cover R&D and salaries, and whatever's left over is Apple's profit. Which is probably a few hundred dollars per machine.
My point is that a Quad-2.0 is not far-fetched at $2000-2200. I'm not aiming to argue pennies. If you want to say it'll cost Apple $1700 to make instead of $1600, OK. My point still stands - Apple can have a 3 Quad line-up.
They could, but who would want to buy that when you can get a 2.93 GHz Conroe system for $2000 that'll be about as fast on multithreaded apps and 50% faster for single threaded apps, which constitute a large majority of apps. A 2.67 GHz Conroe may even be a better buy than a 2S 2 GHz Woodcrest.
2S systems are still very specific in nature, specific to an application. It only pays off for heavy usage of specific apps or markets and when they are clearly better than 1S systems. That means, the floor for Apple's 2S Woodcrest systems needs to have a 2.67 GHz Woodcrest in it, maybe a 2.33 GHz; otherwise, Conroe systems will be the clear price/performance leader with 2.67 and 2.93 GHz systems selling for under $2000.
A Quad-2.0 system doesn't compete with Conroes. It competes will 4x4 and Precision workstations. As an entry level workstation, it destroys 4x4 on processor performance, and it destroys a Precision on price.
Another reason Apple may want an all Quad line-up is that it makes a nice delineator - a 2.67 GHz Mac Pro isn't that much ahead of a 2.4 GHz Conroe iMac. An all-quad pro line-up makes it clear that these are the workstations, especially if coupled with a "Mac" desktop.
The same think is true for SCSI. While the use for it has diminished, it has caused problems in some areas, such as the hi end printer markets, where many machines run off SCSI. The big problem here is not the ports, or cards, it is simply that in 10.3, Apple removed some important software from the OS which has caused problems ever since. There was no need to remove it, they just wanted to push people off it, and on to Firewire.
SCSI is still popular for file servers, now moving toward the new SAS to avoid 80 pin connections. Hi end printers have Ethernet ports so they can be shared by a network. Large format inkjet printers have serial or special connectors for a hardware RIP. Old style impact printers have serial ports and or parallel ports. Big line printers for mainframes have serial ports. I think what melgross is thinking about is Adaptec dropping support for the Mac OS around Mac OS 10.3 ATTO is still going strong for SCSI cards. Again, the consumer is not doing much with SCSI nowadays, mostly servers for external tape drives and the like. It's better to go legacy free. It takes a lot longer to boot up with a SCSI card installed. As for SCSI and printers, there was a SCSI port on some laser printers for an external hard drive to store fonts. The was also one or two SCSI connected printers.
Finally, as for serial support, check out http://www.keyspan.com/products/sxpro/homepage.spml You should notice the Universal and Mac OS X support logos. Now their may not be many programs/devices that work with this because of the small market, but there is no technical reason blocking serial support on modern Macs.
I recognize that. Peg a price on that. Say $100 to cover warranty and manufacturing? R&D and "running the company" fall under "fixed costs" and they are what "margins" (Marginal Revenue - Marginal Cost) offset. I estimated the hardware at $1600 (which you didn't dispute). Let's tack on $100 to cover manufacturing and warranty (I doubt it's that high, but let's pretend).
So then Apple's marginal cost for each Mac Pro they build is $1700. Their marginal revenue is the price they sell it for. Let's call it $2100. That means they make $400 on each low-end Mac Pro. That's a margin of 19%. That $400 helps cover R&D and salaries, and whatever's left over is Apple's profit. Which is probably a few hundred dollars per machine.
My point is that a Quad-2.0 is not far-fetched at $2000-2200. I'm not aiming to argue pennies. If you want to say it'll cost Apple $1700 to make instead of $1600, OK. My point still stands - Apple can have a 3 Quad line-up.
I don't have any idea as to what these costs for Apple are. They could be higher, or they could be lower. You are just guessing. We run those numbers many times, changing factors each time to come to a final price. But fixed costs are also computed into the cost of the product. They must be. Every cost is computed in. Otherwise there are expenses that have no income base.
Capital improvements are also fiqured in. Everything is figured in. That's why margins jump around. You trust that you have allowed enough to cover costs and make that profit. Sometimes you don't, as when a manufacturing process is found to be flawed, or product recalls must be made, like with the MacBooks where they will fix the discoloring problems. Then margins drop, because this was unexpected. Next time, they might adjust some numbers to account for a similar problem.
Just I'd chime in on the video card problem. The driver problem is OpenGL on the Mac versus DirectX on the PC. There are two different interfaces to use the card and of course the relatively small Mac market which leads to a smaller still market for a Mac gamer who can afford a workstation class video card which costs almost as much as the computer itself. Of course there are relatively few games for the Mac OS X platform. Your typical user, even a Photoshop or Final Cut Pro user has little need for a fancy video card. They need lots of fast RAM, disk space and fast CPUs to process all that data. They do have a need for large displays, so a dual-link DVI card with enough VRAM should do.
[QUOTE]Originally posted by melgross
Pretty much all video cards use both. some of the biggest games have been, and still are Open/GL based.
Generally though the push in PC-gaming land is DirectX 9.0c and shaders. While some of the "biggest games" (IIRC eg. Quake4, Doom3?) are OpenGL, the general push is DirectX.
Brad, I think you are generally right. As long as Apple offers Quadros for those that need that power when doing 3D animation, and a few decent cards, say with the MacPros, a 7600GT, 7900GT, 7900GTX, things are hunky-dory.
Remember with BootCamp you pretty much have access to all the latest PC games. The X1600 in the iMac is already a nice treat.
I generally am not going to complain about the lack of video card choices, I was perhaps just wondering on issues with writing drivers for it.
Again, with the new Mac Pros, decent high 7-series nVidias or X1600, X1800, X1900 options with ATI, and decent performance in Mac general apps, Mac games, and Bootcamp PC games, I think the Mac Pros are looking solid all round.
I know gaming is not a big target market for Mac Pros, but the choices Apple is offering now with their Intel range, is really putting Apple at an unprecedented level of attractiveness to a wide target market.
Finally though, I empathise with people who have had challenges with Mac software and hardware. Knowing how to get the most out of your Mac and to have it not malfunction on you is very important for income-producing activities using the Mac.
Oh one more thing. To all those with MacBookPros, iMac Intels and those who get MacPros, even if you're not a gamer, give HalfLife2 and HalfLife2:Episode 1 a spin. It's a interactive cinematic narrative in a very interesting survival horror/ science fiction genre. Well, my usual HalfLife2 plug... Carry on...
SCSI is still popular for file servers, now moving toward the new SAS to avoid 80 pin connections. Hi end printers have Ethernet ports so they can be shared by a network. Large format inkjet printers have serial or special connectors for a hardware RIP. Old style impact printers have serial ports and or parallel ports. Big line printers for mainframes have serial ports. I think what melgross is thinking about is Adaptec dropping support for the Mac OS around Mac OS 10.3 ATTO is still going strong for SCSI cards. Again, the consumer is not doing much with SCSI nowadays, mostly servers for external tape drives and the like. It's better to go legacy free. It takes a lot longer to boot up with a SCSI card installed. As for SCSI and printers, there was a SCSI port on some laser printers for an external hard drive to store fonts. The was also one or two SCSI connected printers.
Finally, as for serial support, check out http://www.keyspan.com/products/sxpro/homepage.spml You should notice the Universal and Mac OS X support logos. Now their may not be many programs/devices that work with this because of the small market, but there is no technical reason blocking serial support on modern Macs.
No, that's not what I'm thinking about. Adaptec has nothing to do with this. It's Apple's fault. I have a Fuji machine that stopped working with 10.3. nothing else was changes. when Fuji worked on the drivers, they found that there were API's missing. It took more than a year's work to get the machine to work with 10.3 and up. Now, there are problems with 10.5 and up. Not just a bug either.
Apple has removed many bits that don't relate to anything other than HD control.
When I say hi end printers, I'm not talking about data center laser printers, and the like. I'm talking about $500,000 graphic art printers and the like.
This is why I say that if you don't work with these machines, or aren't familiar with them, you won't know what I mean.
Generally though the push in PC-gaming land is DirectX 9.0c and shaders. While some of the "biggest games" (IIRC eg. Quake4, Doom3?) are OpenGL, the general push is DirectX.
Brad, I think you are generally right. As long as Apple offers Quadros for those that need that power when doing 3D animation, and a few decent cards, say with the MacPros, a 7600GT, 7900GT, 7900GTX, things are hunky-dory.
I agree with the other matters you mentioned.
But don't forget that there are other programs that also use Open/GL. It's not just games.
The problem is that we don't get a good selection of graphics boards. we don't have anything other than the bottom game boards, one low/middle game board, and then right up to the top pro board. That's just not good enough.
I've been trying to figure out a way for Apple to keep the "Mac Pro" machines in the $2000 to $3000 range, and there really isn't a good way.
Woodcrest systems are expensive. Apple can probably sell a 2S Xeon 5140 (2.33 GHz) in the neighborhood of $2500 to $2700 (maybe), and should be able to perform about the same as the G5 Quad in FPU while crushing it integer. However, a hypothetical 2.93 GHz Kentsfield system is going to smoke it, at least in Apple's market.
For Apple, the gearing for a Kentsfield system is all favorable compared to a 2S Woodcrest system:
1. A 2.66 GHz Kentsfield will be cheaper than 2 Xeon 5150 CPUs (2.66 GHz Woodcrests). It'll be 5 to 10% slower if using the same memory, but they won't be.
2. Kentsfield will use DDR2 RAM which is half as cheap as FB-DIMMs
3. DDR2 RAM is faster for Apple's "Mac Pro" market and its class of applications which should gain that 5 to 10% back.
4. Kentsfield motherboards will be half as cheap as Woodcrest boards
The only loss is that Kentsfield systems will only have 8 memory slots at most for 16 GB memory using 2 GB DIMMs, while Woodcrest system can have 16+ memory slots for 32 to 64 GB memory. Not a big loss for Apple.
Hence, Kentsfield systems can occupy the $2300 to $2700 mid-range, Conroe can occupy the sub-$2000 space, while 2S Woodcrest the $3000+ range. If not, there will be a big hole in the middle in Apple's Mac Pro lineup, or the its low-end will be outperformed by Conroe systems costing $500 less. The only advantage to such Woodcrest systems is large memory capacity. Doubt Apple has the market to support that.
Kentsfield isn't available now is it? Isn't it coming to market at the end of this year?
No, that's not what I'm thinking about. Adaptec has nothing to do with this. It's Apple's fault. I have a Fuji machine that stopped working with 10.3. nothing else was changes. when Fuji worked on the drivers, they found that there were API's missing. It took more than a year's work to get the machine to work with 10.3 and up. Now, there are problems with 10.5 and up. Not just a bug either.
Apple has removed many bits that don't relate to anything other than HD control.
When I say hi end printers, I'm not talking about data center laser printers, and the like. I'm talking about $500,000 graphic art printers and the like.
This is why I say that if you don't work with these machines, or aren't familiar with them, you won't know what I mean.
Sorry, I didn't know your situation. Yes, Apple has changed the printing APIs, and more than once, which has fustrated many a printer driver writer. They also tripped up scanner drivers too when 10.3 came out. The pace at which Apple released major OS upgrades also made several companies give up on the Mac. Such is life. I keep seeing this online banner ad for TypeStyler, which was a cool type effects program for OS 9. Never made it to OS X, but the ad has been saying for about two years that a OS X version is on it's way. In about 2 weeks, there won't be any Macs that can run Classic anymore. That's assuming Apple drops the PM G5. "Progress is our greatest stumbling block"
Yeah, with 10.5 coming out, drivers for scanners, printers and miscellaneous stuff is rare enough as it is, Leopard and Universal has put pressure on a lot of Mac software developers and hardware/ accessory makers. 10.5 better have some really good out-of-the-box support for tons of video and still digital cameras
Comments
Originally posted by kim kap sol
MOSR still exists?
MOSR is one issue that we can ALL agree upon.
Originally posted by onlooker
I don't think there is an onsite service for consumer purchases if that's what your asking.
Applecare makes no differention between professional, and consumer purchases.
So, if it doesn't have it, it's not there.
Woodcrest systems are expensive. Apple can probably sell a 2S Xeon 5140 (2.33 GHz) in the neighborhood of $2500 to $2700 (maybe), and should be able to perform about the same as the G5 Quad in FPU while crushing it integer. However, a hypothetical 2.93 GHz Kentsfield system is going to smoke it, at least in Apple's market.
For Apple, the gearing for a Kentsfield system is all favorable compared to a 2S Woodcrest system:
1. A 2.66 GHz Kentsfield will be cheaper than 2 Xeon 5150 CPUs (2.66 GHz Woodcrests). It'll be 5 to 10% slower if using the same memory, but they won't be.
2. Kentsfield will use DDR2 RAM which is half as cheap as FB-DIMMs
3. DDR2 RAM is faster for Apple's "Mac Pro" market and its class of applications which should gain that 5 to 10% back.
4. Kentsfield motherboards will be half as cheap as Woodcrest boards
The only loss is that Kentsfield systems will only have 8 memory slots at most for 16 GB memory using 2 GB DIMMs, while Woodcrest system can have 16+ memory slots for 32 to 64 GB memory. Not a big loss for Apple.
Hence, Kentsfield systems can occupy the $2300 to $2700 mid-range, Conroe can occupy the sub-$2000 space, while 2S Woodcrest the $3000+ range. If not, there will be a big hole in the middle in Apple's Mac Pro lineup, or the its low-end will be outperformed by Conroe systems costing $500 less. The only advantage to such Woodcrest systems is large memory capacity. Doubt Apple has the market to support that.
Originally posted by onlooker
I don't think there is an onsite service for consumer purchases if that's what your asking.
To be clear, melgross was comparing the PowerMac's available (optional) support with what comes standard with Dell's Precision series. Precision is not a consumer purchase by any stretch of any imagination, and I really don't think PowerMac qualifies as consumer either, so I don't see how that reasoning fits.
Originally posted by kim kap sol
You're talking about a niche market within a niche market: Niche^2. It's unjustifiable...it's like me asking for a built-in NES cartridge loader so I can download the the many cartridge ROMs I collect to my computer so I can play them on an emulator.
Just buy the PCI cards that you need and be done with it. No need to put 2 ports that will be used by 1% of the already very small Mac user base.
The question is what is meant by niche. The biomedical industry is one that Apple has been pushing for years, and in which they have done fairly well. But there are catagories that they are not in, because of issues such as this. The same is true for other industries they are involved in.
It's too easy, when sitting down at a computer, to think that if we don't need something, that others don't as well. Apple wants very badly to get a big chunk of the scientific computing market. In large parts of that market, they had a bigger marketshare than they do now. One reason is the problem of control interfacing.
As I mentioned earlier, there were a number of companies that offered these serial cards to the Mac market for this very purpose. But Apple removed much of the necessary software from the OS when they went to OS X. Thinking, I suppose, the same thing that is being said here. But the markets didn't follow. They still use serial control. And Apple is lagging there. If we just had some good hi end boards it would help. But, Apple has to understand this, and they don't seem to.
We have discussed many times the fact that Apple doesn't always know what's best for them.
The same think is true for SCSI. While the use for it has diminished, it has caused problems in some areas, such as the hi end printer markets, where many machines run off SCSI. The big problem here is not the ports, or cards, it is simply that in 10.3, Apple removed some important software from the OS which has caused problems ever since. There was no need to remove it, they just wanted to push people off it, and on to Firewire.
Originally posted by THT
Woodcrest systems are expensive. Apple can probably sell a 2S Xeon 5140 (2.33 GHz) in the neighborhood of $2500 to $2700 (maybe), and should be able to perform about the same as the G5 Quad in FPU while crushing it integer. However, a hypothetical 2.93 GHz Kentsfield system is going to smoke it, at least in Apple's market.
Apple can sell a 2S 2.0 Woodcrest system for $2000-$2200:
$660 for processors
$90 for HDD (250 GB maybe)
$400 MoBo
$175 Video Card (7600GT)
$150 RAM (2x512 FB-DIMM 533)
$200 Case
Sub-Total: $1675
That's paying newegg prices for the Video Card, the RAM and the HDD, and 1000-lot prices for the processors. The whole system could cost Apple $1600 when you add in BT, AirPort, FireWire, IR port, and whatever I'm forgetting if Apple gets good deals.
A quad-2.0 will hold it's own against most dual-Opteron boards, and will fare well processor-wise against 4x4. It'll avoid competing with Dell XPS's, and crush all Precision workstations on price.
Originally posted by ZachPruckowski
Apple can sell a 2S 2.0 Woodcrest system for $2000-$2200:
$660 for processors
$90 for HDD (250 GB maybe)
$400 MoBo
$175 Video Card (7600GT)
$150 RAM (2x512 FB-DIMM 533)
$200 Case
Sub-Total: $1675
That's paying newegg prices for the Video Card, the RAM and the HDD, and 1000-lot prices for the processors. The whole system could cost Apple $1600 when you add in BT, AirPort, FireWire, IR port, and whatever I'm forgetting if Apple gets good deals.
A quad-2.0 will hold it's own against most dual-Opteron boards, and will fare well processor-wise against 4x4. It'll avoid competing with Dell XPS's, and crush all Precision workstations on price.
When these prices are put out here, there are important parts that are missing. What about the cost to manufacture the product? What about the cost of R&D that is figured in? What about the cost of running the company, which also must be figured in? What about warranty costs, etc?
The problem with breakdowns like those done by iSupply is that all they do is to give the cost of the parts, and sometimes an undervalued assembly factor. When the breakdowns are done here, they copy those inadequate numbers. As one who did manufacturing, I can tell you that these part numbers aren't sufficient to give the true cost.
Originally posted by ZachPruckowski
Apple can sell a 2S 2.0 Woodcrest system for $2000-$2200
They could, but who would want to buy that when you can get a 2.93 GHz Conroe system for $2000 that'll be about as fast on multithreaded apps and 50% faster for single threaded apps, which constitute a large majority of apps. A 2.67 GHz Conroe may even be a better buy than a 2S 2 GHz Woodcrest.
2S systems are still very specific in nature, specific to an application. It only pays off for heavy usage of specific apps or markets and when they are clearly better than 1S systems. That means, the floor for Apple's 2S Woodcrest systems needs to have a 2.67 GHz Woodcrest in it, maybe a 2.33 GHz; otherwise, Conroe systems will be the clear price/performance leader with 2.67 and 2.93 GHz systems selling for under $2000.
The only other big advantage is 32 to 64 GB memory support. People who can afford 16 GB, 32 GB or 64 GB of memory aren't going to be worried about the $1000 price difference between a 2S 2.66+ GHz system and a 2S <2.0 GHz system.
Originally posted by melgross
When these prices are put out here, there are important parts that are missing. What about the cost to manufacture the product? What about the cost of R&D that is figured in? What about the cost of running the company, which also must be figured in? What about warranty costs, etc?
The problem with breakdowns like those done by iSupply is that all they do is to give the cost of the parts, and sometimes an undervalued assembly factor. When the breakdowns are done here, they copy those inadequate numbers. As one who did manufacturing, I can tell you that these part numbers aren't sufficient to give the true cost.
That about sums up what I was going to say. There is a lot more to building, and selling a computer than it's wholesale parts price.
Originally posted by onlooker
That about sums up what I was going to say. There is a lot more to building, and selling a computer than it's wholesale parts price.
Damn right there is!
Originally posted by melgross
When these prices are put out here, there are important parts that are missing. What about the cost to manufacture the product? What about the cost of R&D that is figured in? What about the cost of running the company, which also must be figured in? What about warranty costs, etc?
The problem with breakdowns like those done by iSupply is that all they do is to give the cost of the parts, and sometimes an undervalued assembly factor. When the breakdowns are done here, they copy those inadequate numbers. As one who did manufacturing, I can tell you that these part numbers aren't sufficient to give the true cost.
I recognize that. Peg a price on that. Say $100 to cover warranty and manufacturing? R&D and "running the company" fall under "fixed costs" and they are what "margins" (Marginal Revenue - Marginal Cost) offset. I estimated the hardware at $1600 (which you didn't dispute). Let's tack on $100 to cover manufacturing and warranty (I doubt it's that high, but let's pretend).
So then Apple's marginal cost for each Mac Pro they build is $1700. Their marginal revenue is the price they sell it for. Let's call it $2100. That means they make $400 on each low-end Mac Pro. That's a margin of 19%. That $400 helps cover R&D and salaries, and whatever's left over is Apple's profit. Which is probably a few hundred dollars per machine.
My point is that a Quad-2.0 is not far-fetched at $2000-2200. I'm not aiming to argue pennies. If you want to say it'll cost Apple $1700 to make instead of $1600, OK. My point still stands - Apple can have a 3 Quad line-up.
Originally posted by THT
They could, but who would want to buy that when you can get a 2.93 GHz Conroe system for $2000 that'll be about as fast on multithreaded apps and 50% faster for single threaded apps, which constitute a large majority of apps. A 2.67 GHz Conroe may even be a better buy than a 2S 2 GHz Woodcrest.
2S systems are still very specific in nature, specific to an application. It only pays off for heavy usage of specific apps or markets and when they are clearly better than 1S systems. That means, the floor for Apple's 2S Woodcrest systems needs to have a 2.67 GHz Woodcrest in it, maybe a 2.33 GHz; otherwise, Conroe systems will be the clear price/performance leader with 2.67 and 2.93 GHz systems selling for under $2000.
A Quad-2.0 system doesn't compete with Conroes. It competes will 4x4 and Precision workstations. As an entry level workstation, it destroys 4x4 on processor performance, and it destroys a Precision on price.
Another reason Apple may want an all Quad line-up is that it makes a nice delineator - a 2.67 GHz Mac Pro isn't that much ahead of a 2.4 GHz Conroe iMac. An all-quad pro line-up makes it clear that these are the workstations, especially if coupled with a "Mac" desktop.
from melgross
The same think is true for SCSI. While the use for it has diminished, it has caused problems in some areas, such as the hi end printer markets, where many machines run off SCSI. The big problem here is not the ports, or cards, it is simply that in 10.3, Apple removed some important software from the OS which has caused problems ever since. There was no need to remove it, they just wanted to push people off it, and on to Firewire.
SCSI is still popular for file servers, now moving toward the new SAS to avoid 80 pin connections. Hi end printers have Ethernet ports so they can be shared by a network. Large format inkjet printers have serial or special connectors for a hardware RIP. Old style impact printers have serial ports and or parallel ports. Big line printers for mainframes have serial ports. I think what melgross is thinking about is Adaptec dropping support for the Mac OS around Mac OS 10.3 ATTO is still going strong for SCSI cards. Again, the consumer is not doing much with SCSI nowadays, mostly servers for external tape drives and the like. It's better to go legacy free. It takes a lot longer to boot up with a SCSI card installed. As for SCSI and printers, there was a SCSI port on some laser printers for an external hard drive to store fonts. The was also one or two SCSI connected printers.
Finally, as for serial support, check out http://www.keyspan.com/products/sxpro/homepage.spml You should notice the Universal and Mac OS X support logos. Now their may not be many programs/devices that work with this because of the small market, but there is no technical reason blocking serial support on modern Macs.
Originally posted by ZachPruckowski
I recognize that. Peg a price on that. Say $100 to cover warranty and manufacturing? R&D and "running the company" fall under "fixed costs" and they are what "margins" (Marginal Revenue - Marginal Cost) offset. I estimated the hardware at $1600 (which you didn't dispute). Let's tack on $100 to cover manufacturing and warranty (I doubt it's that high, but let's pretend).
So then Apple's marginal cost for each Mac Pro they build is $1700. Their marginal revenue is the price they sell it for. Let's call it $2100. That means they make $400 on each low-end Mac Pro. That's a margin of 19%. That $400 helps cover R&D and salaries, and whatever's left over is Apple's profit. Which is probably a few hundred dollars per machine.
My point is that a Quad-2.0 is not far-fetched at $2000-2200. I'm not aiming to argue pennies. If you want to say it'll cost Apple $1700 to make instead of $1600, OK. My point still stands - Apple can have a 3 Quad line-up.
I don't have any idea as to what these costs for Apple are. They could be higher, or they could be lower. You are just guessing. We run those numbers many times, changing factors each time to come to a final price. But fixed costs are also computed into the cost of the product. They must be. Every cost is computed in. Otherwise there are expenses that have no income base.
Capital improvements are also fiqured in. Everything is figured in. That's why margins jump around. You trust that you have allowed enough to cover costs and make that profit. Sometimes you don't, as when a manufacturing process is found to be flawed, or product recalls must be made, like with the MacBooks where they will fix the discoloring problems. Then margins drop, because this was unexpected. Next time, they might adjust some numbers to account for a similar problem.
Just I'd chime in on the video card problem. The driver problem is OpenGL on the Mac versus DirectX on the PC. There are two different interfaces to use the card and of course the relatively small Mac market which leads to a smaller still market for a Mac gamer who can afford a workstation class video card which costs almost as much as the computer itself. Of course there are relatively few games for the Mac OS X platform. Your typical user, even a Photoshop or Final Cut Pro user has little need for a fancy video card. They need lots of fast RAM, disk space and fast CPUs to process all that data. They do have a need for large displays, so a dual-link DVI card with enough VRAM should do.
[QUOTE]Originally posted by melgross
Pretty much all video cards use both. some of the biggest games have been, and still are Open/GL based.
Generally though the push in PC-gaming land is DirectX 9.0c and shaders. While some of the "biggest games" (IIRC eg. Quake4, Doom3?) are OpenGL, the general push is DirectX.
Brad, I think you are generally right. As long as Apple offers Quadros for those that need that power when doing 3D animation, and a few decent cards, say with the MacPros, a 7600GT, 7900GT, 7900GTX, things are hunky-dory.
Remember with BootCamp you pretty much have access to all the latest PC games. The X1600 in the iMac is already a nice treat.
I generally am not going to complain about the lack of video card choices, I was perhaps just wondering on issues with writing drivers for it.
Again, with the new Mac Pros, decent high 7-series nVidias or X1600, X1800, X1900 options with ATI, and decent performance in Mac general apps, Mac games, and Bootcamp PC games, I think the Mac Pros are looking solid all round.
I know gaming is not a big target market for Mac Pros, but the choices Apple is offering now with their Intel range, is really putting Apple at an unprecedented level of attractiveness to a wide target market.
Finally though, I empathise with people who have had challenges with Mac software and hardware. Knowing how to get the most out of your Mac and to have it not malfunction on you is very important for income-producing activities using the Mac.
Oh one more thing. To all those with MacBookPros, iMac Intels and those who get MacPros, even if you're not a gamer, give HalfLife2 and HalfLife2:Episode 1 a spin. It's a interactive cinematic narrative in a very interesting survival horror/ science fiction genre. Well, my usual HalfLife2 plug... Carry on...
Originally posted by BradMacPro
SCSI is still popular for file servers, now moving toward the new SAS to avoid 80 pin connections. Hi end printers have Ethernet ports so they can be shared by a network. Large format inkjet printers have serial or special connectors for a hardware RIP. Old style impact printers have serial ports and or parallel ports. Big line printers for mainframes have serial ports. I think what melgross is thinking about is Adaptec dropping support for the Mac OS around Mac OS 10.3 ATTO is still going strong for SCSI cards. Again, the consumer is not doing much with SCSI nowadays, mostly servers for external tape drives and the like. It's better to go legacy free. It takes a lot longer to boot up with a SCSI card installed. As for SCSI and printers, there was a SCSI port on some laser printers for an external hard drive to store fonts. The was also one or two SCSI connected printers.
Finally, as for serial support, check out http://www.keyspan.com/products/sxpro/homepage.spml You should notice the Universal and Mac OS X support logos. Now their may not be many programs/devices that work with this because of the small market, but there is no technical reason blocking serial support on modern Macs.
No, that's not what I'm thinking about. Adaptec has nothing to do with this. It's Apple's fault. I have a Fuji machine that stopped working with 10.3. nothing else was changes. when Fuji worked on the drivers, they found that there were API's missing. It took more than a year's work to get the machine to work with 10.3 and up. Now, there are problems with 10.5 and up. Not just a bug either.
Apple has removed many bits that don't relate to anything other than HD control.
When I say hi end printers, I'm not talking about data center laser printers, and the like. I'm talking about $500,000 graphic art printers and the like.
This is why I say that if you don't work with these machines, or aren't familiar with them, you won't know what I mean.
Originally posted by sunilraman
Generally though the push in PC-gaming land is DirectX 9.0c and shaders. While some of the "biggest games" (IIRC eg. Quake4, Doom3?) are OpenGL, the general push is DirectX.
Brad, I think you are generally right. As long as Apple offers Quadros for those that need that power when doing 3D animation, and a few decent cards, say with the MacPros, a 7600GT, 7900GT, 7900GTX, things are hunky-dory.
I agree with the other matters you mentioned.
But don't forget that there are other programs that also use Open/GL. It's not just games.
The problem is that we don't get a good selection of graphics boards. we don't have anything other than the bottom game boards, one low/middle game board, and then right up to the top pro board. That's just not good enough.
Originally posted by THT
I've been trying to figure out a way for Apple to keep the "Mac Pro" machines in the $2000 to $3000 range, and there really isn't a good way.
Woodcrest systems are expensive. Apple can probably sell a 2S Xeon 5140 (2.33 GHz) in the neighborhood of $2500 to $2700 (maybe), and should be able to perform about the same as the G5 Quad in FPU while crushing it integer. However, a hypothetical 2.93 GHz Kentsfield system is going to smoke it, at least in Apple's market.
For Apple, the gearing for a Kentsfield system is all favorable compared to a 2S Woodcrest system:
1. A 2.66 GHz Kentsfield will be cheaper than 2 Xeon 5150 CPUs (2.66 GHz Woodcrests). It'll be 5 to 10% slower if using the same memory, but they won't be.
2. Kentsfield will use DDR2 RAM which is half as cheap as FB-DIMMs
3. DDR2 RAM is faster for Apple's "Mac Pro" market and its class of applications which should gain that 5 to 10% back.
4. Kentsfield motherboards will be half as cheap as Woodcrest boards
The only loss is that Kentsfield systems will only have 8 memory slots at most for 16 GB memory using 2 GB DIMMs, while Woodcrest system can have 16+ memory slots for 32 to 64 GB memory. Not a big loss for Apple.
Hence, Kentsfield systems can occupy the $2300 to $2700 mid-range, Conroe can occupy the sub-$2000 space, while 2S Woodcrest the $3000+ range. If not, there will be a big hole in the middle in Apple's Mac Pro lineup, or the its low-end will be outperformed by Conroe systems costing $500 less. The only advantage to such Woodcrest systems is large memory capacity. Doubt Apple has the market to support that.
Kentsfield isn't available now is it? Isn't it coming to market at the end of this year?
Originally posted by melgross
No, that's not what I'm thinking about. Adaptec has nothing to do with this. It's Apple's fault. I have a Fuji machine that stopped working with 10.3. nothing else was changes. when Fuji worked on the drivers, they found that there were API's missing. It took more than a year's work to get the machine to work with 10.3 and up. Now, there are problems with 10.5 and up. Not just a bug either.
Apple has removed many bits that don't relate to anything other than HD control.
When I say hi end printers, I'm not talking about data center laser printers, and the like. I'm talking about $500,000 graphic art printers and the like.
This is why I say that if you don't work with these machines, or aren't familiar with them, you won't know what I mean.
Sorry, I didn't know your situation. Yes, Apple has changed the printing APIs, and more than once, which has fustrated many a printer driver writer. They also tripped up scanner drivers too when 10.3 came out. The pace at which Apple released major OS upgrades also made several companies give up on the Mac. Such is life. I keep seeing this online banner ad for TypeStyler, which was a cool type effects program for OS 9. Never made it to OS X, but the ad has been saying for about two years that a OS X version is on it's way. In about 2 weeks, there won't be any Macs that can run Classic anymore. That's assuming Apple drops the PM G5. "Progress is our greatest stumbling block"