3D. Video. Production. Encoding. Handbrake. VERY FEW games. Some scientific apps. They're out there. And usually people on the edge of 'making' or 'discovering/figuring' out things. Guesses: global warming and weather modelling/predictions?
*Shrugs.
Once Open CL parcels code into parcels 'instead'(?) of threads then maybe most of everything will be able to take advantage of Quad core. You can render some 3D now. Get that going, do some Photoshop action work while browsing the Internet while playing iTunes. I'm sure that would give a Quad Core machine a work out?
Still. Apple shouldn't be waiting on quad-core chips. Use the ones that are out there now. The Conroe Penryns should be dirt cheap to use.
I get the laptops. But the Mini should be a Cube and I can't see why the iMac should wait on laptop Quad cores. To me? Flaws in their designs? Still, seeing as most apps are dual-core aware at best, Apple's technically getting away with it. But the customer will still see quad core desktops on the market 'NOW' for less than a grand with a better gpu than the iMac has. It's embarrassing to be behind in this respect.
I do see Apple pushing for more cores in future Macs to really take advantage of the HW, but only once SL gets close to a release. Otherwise it just seem to be a good marketing move if they offer a Quad-core that is slower than a Dual-core chip in most real world usage in the current Leopard OS.
While I would like to see Apple use Quad-core in the next iMacs and also offer a smaller headless Mac with one CPU, I am not convinced that either will happen. The TDP of even the new chips are still too high for the iMac (assuming nothing else changes), but I do think it would behoove Apple to offer the power and cost savings of a desktop-grade chip to the iMac if they can.
....... but I do think it would behoove Apple to offer the power and cost savings of a desktop-grade chip to the iMac if they can.
While I can agree that desktop chips offfer a performance advantage I don't buy the cost savings justification. First in the case of the iMacs we don't have any idea at all as to what Apple may be paying for the chips as they are special runs specifically for Apple. Second if you take in the total cost of ownership the lower power usage ought to pay for itself in very short order. It is an issue power usage as much as performance.
In any event it would be incredibly poor timing on Apples part to actually increase the average power usage of it's products. It is not just an issue of being green but one of overall costs when people might not be flush with cash.
While I can agree that desktop chips offfer a performance advantage I don't buy the cost savings justification.
Do we have to know how much Apple pays to know that in lots of 1000 a desktop CPU at the same GHz is cheaper and faster than a mobile CPU? Regardless of what Apple pays their can't be too much difference in their +1000 discount from using a desktop version over a mobile version.
Do we have to know how much Apple pays to know that in lots of 1000 a desktop CPU at the same GHz is cheaper and faster than a mobile CPU? Regardless of what Apple pays their can't be too much difference in their +1000 discount from using a desktop version over a mobile version.
There is if the mobile version costs significantly more to make at any given mhz. Apple may be paying the same amount for mobile chips as they would desktop chips, but they're getting a much slower product that is without affordable quad core options.
There is if the mobile version costs significantly more to make at any given mhz. Apple may be paying the same amount for mobile chips as they would desktop chips, but they're getting a much slower product that is without affordable quad core options.
Come again? If they cost more to make, then it stands to reason that intel is charging more for them. We aren't talking about low production chips for workstations that will have a considerably higher margin, we are talking about consumer-grade desktop and mobile chips. If Intel charges $300 for an Quad-core desktop chip at 2.93GHz and $300 for a Dual core mobile chip at 2.66GHz, and is still buying the same amount of chips either way, why would you think what Apple pays would be excessively different from the desktop chip at $300 per 1000 and the mobile chip at $300 per 1000. You'll have to supply some evidence to support Intel offering vastly different margins on their popular chips simply because one is mobile- and one is desktop-grade for me to swallow that.
Come again? If they cost more to make, then it stands to reason that intel is charging more for them. We aren't talking about low production chips for workstations that will have a considerably higher margin, we are talking about consumer-grade desktop and mobile chips. If Intel charges $300 for an Quad-core desktop chip at 2.93GHz and $300 for a Dual core mobile chip at 2.66GHz, and is still buying the same amount of chips either way, why would you think what Apple pays would be excessively different from the desktop chip at $300 per 1000 and the mobile chip at $300 per 1000. You'll have to supply some evidence to support Intel offering vastly different margins on their popular chips simply because one is mobile- and one is desktop-grade for me to swallow that.
I'm not. I'm saying you're getting a lot more in that desktop quad core for your $300 than the mobile dual core.
Come again? If they cost more to make, then it stands to reason that intel is charging more for them.
We really don't know what it actually costs to make a mobile chips vs a desktop. Certainly there must be a tweak in the process but does that make up for the savings in chip real estate? I don't know but I've ling been under the impression that it is the area of the die that in the end controls the cost of the device.
By the way i would suggest that Intel charges more for them because they can. Mobile chips are seen as premium devices, so corresponding higher prices. The question is does the process Intel uses justify the higher prices. I suspect that the cost to deliver a mobile chip isn't all that much different than a desktop chip.
Quote:
We aren't talking about low production chips for workstations that will have a considerably higher margin, we are talking about consumer-grade desktop and mobile chips. If Intel charges $300 for an Quad-core desktop chip at 2.93GHz and $300 for a Dual core mobile chip at 2.66GHz, and is still buying the same amount of chips either way, why would you think what Apple pays would be excessively different from the desktop chip at $300 per 1000 and the mobile chip at $300 per 1000. You'll have to supply some evidence to support Intel offering vastly different margins on their popular chips simply because one is mobile- and one is desktop-grade for me to swallow that.
Well we all know here that figuring out Intels margins on chips is going to be difficult. It is not like they are going to publish such info. In any event in Apples case there is noting low production about mobile chips as they out pace desktop sales. I'm just not convinced that the pricing difference in mobile chips is directly related to cost of production. Throw a special part number on the chip and sell it to Apple and Intel can come to any sort of pricing arraignment they want with Apple. After all if the product is Apple specific the competition can't charge Intel with unfair business practices.
We really don't know what it actually costs to make a mobile chips vs a desktop.
[...]
I'm just not convinced that the pricing difference in mobile chips is directly related to cost of production.
[...]
Why all the emphasis on Intel's actual cost? We know what they charge for batches of 1000. Why assume that a $300 mobile chip in batches of 1000 are going to be much different in price to desktop chips selling at $300 in batches of 1000? The argument that we can't know every little detail about Intel's margins per chip model and Apple's deal with Intel means we can't get an idea of the cost to speed increase from mobile to desktop is ridiculous. We know Intel's base price so we can draw a valid comparison based on the data we do have. Their is no need to make it complex.
PS:
Quote:
But do they cost more to make?
Trying to figure that out is irreverent to the discussion, but if I were to guess I'd say mobile chips have a higher mark up, which means that that a desktop chip at the same price as a mobile chip is more expensive from R&D through production. My reasoning is, besides the higher demand for mobile chips that you mentioned, is that Intel kills AMD in this area more so than it does in the desktop arena.
I fail to see how Intel's manufacturing cost is relevant.
We are discussing the options available to Apple based on known information about Intel's chip CPU, TDP, L2 at comparable prices. Explain to me why it's relevant to know if Intel has reduced R&D for a desktop chip of reduced factory costs in China for a specific chip if a mobile chip at a lower clock speed costs the same. When you compare a Sony and LG HDTV are you comparing the specs and overall quality or are you waiting to know how much R&D went into each display before buying?
We are discussing the options available to Apple based on known information about Intel's chip CPU, TDP, L2 at comparable prices. Explain to me why it's relevant to know if Intel has reduced R&D for a desktop chip of reduced factory costs in China for a specific chip if a mobile chip at a lower clock speed costs the same. When you compare a Sony and LG HDTV are you comparing the specs and overall quality or are you waiting to know how much R&D went into each display before buying?
All 45nm Core 2 processors are manufactured using the same process at the same fabs. All things being equal (clock speed, FSB, cache size), a mobile part and a desktop part would be identical silicon. The only difference between them would be binning and the package they go on. They cost the same to make, more or less.
I don't know if that answers your question or not.
The only difference between them would be binning and the package they go on. They cost the same to make, more or less.
Why go back to production and R&D costs when we have the price list that Intel sells their chips at. That should be our starting point. Take the $316 chip used in the iMac and see what the equivalent would be in performance to see the price and take the equivalent chip in price to see the performance. I don't understand why we need to complicate the issue with details we'll never know.
Why go back to production and R&D costs when we have the price list that Intel sells their chips at. That should be our starting point. Take the $316 chip used in the iMac and see what the equivalent would be in performance to see the price and take the equivalent chip in price to see the performance. I don't understand why we need to complicate the issue with details we'll never know.
One thing to take into account nevertheless, is that the cpus in the current iMac are probably cheaper than the mobile ones given that they have lesser power restrictions (higher TDP). So We must expect that those cpus would cost 10-20% less than equivalent mobile (25-35W) ones:
iMac 2.40GHz cpu for probably $222 (if it was available on Intel's price list)
iMac 2.66GHz cpu for probably $253 (if it was available on Intel's price list)
iMac 2.80GHz cpu for probably $424 (if it was available on Intel's price list)
iMac 3.06GHz cpu for probably $681 (if it was available on Intel's price list)
That doesn't change the fact that desktop dual-core cpus would be less expensive at similar speeds/cache, and that even most of the 65W quad-core cpus would also be less expensive.
Given that the new iMacs will probably get LED-BL displays (the 24" panel is already in inventory, that leaves the 20" one), it will cost more to build, but in choosing 65W desktop cpus, it could not only give a nice speedbump but also keep the prices at similar levels as today while keeping the margins the same (or better).
$1299 LED-BL 20" iMac 65W dual-core 2.66GHz, 2GB RAM, 320GB HDD, Superdrive * small price bump due to the LED-BL display and 1-?2GB of RAM
But the customer will still see quad core desktops on the market 'NOW' for less than a grand with a better gpu than the iMac has. It's embarrassing to be behind in this respect.
Never mind the iMac and multicore. The thing that's holding me back from getting a mini and may eventually drive me to getting a Linux box is the slow GPU that makes the mini inadequate for HDTV. I want a box to replace my VCR and DVD player and a consumer digital recorder is expensive for what you get.
The crazy thing is, before I realized that the GPU was the main bottleneck, I planned on buying a 1.6GHz superdrive mini off ebay and upgrading it with a 2.33GHz part also off ebay, but the used market is pricing them so high, it's not worth until (until the 2.33GHz part drops a bit more). Just as well I spotted the GPU story in time ...
Comments
3D. Video. Production. Encoding. Handbrake. VERY FEW games. Some scientific apps. They're out there. And usually people on the edge of 'making' or 'discovering/figuring' out things. Guesses: global warming and weather modelling/predictions?
*Shrugs.
Once Open CL parcels code into parcels 'instead'(?) of threads then maybe most of everything will be able to take advantage of Quad core. You can render some 3D now. Get that going, do some Photoshop action work while browsing the Internet while playing iTunes. I'm sure that would give a Quad Core machine a work out?
Still. Apple shouldn't be waiting on quad-core chips. Use the ones that are out there now. The Conroe Penryns should be dirt cheap to use.
I get the laptops. But the Mini should be a Cube and I can't see why the iMac should wait on laptop Quad cores. To me? Flaws in their designs? Still, seeing as most apps are dual-core aware at best, Apple's technically getting away with it. But the customer will still see quad core desktops on the market 'NOW' for less than a grand with a better gpu than the iMac has. It's embarrassing to be behind in this respect.
I do see Apple pushing for more cores in future Macs to really take advantage of the HW, but only once SL gets close to a release. Otherwise it just seem to be a good marketing move if they offer a Quad-core that is slower than a Dual-core chip in most real world usage in the current Leopard OS.
While I would like to see Apple use Quad-core in the next iMacs and also offer a smaller headless Mac with one CPU, I am not convinced that either will happen. The TDP of even the new chips are still too high for the iMac (assuming nothing else changes), but I do think it would behoove Apple to offer the power and cost savings of a desktop-grade chip to the iMac if they can.
ffer the power and cost savings of a desktop-grade chip to the iMac if they can.
We'll see, I guess.
Lemon Bon Bon.
....... but I do think it would behoove Apple to offer the power and cost savings of a desktop-grade chip to the iMac if they can.
While I can agree that desktop chips offfer a performance advantage I don't buy the cost savings justification. First in the case of the iMacs we don't have any idea at all as to what Apple may be paying for the chips as they are special runs specifically for Apple. Second if you take in the total cost of ownership the lower power usage ought to pay for itself in very short order. It is an issue power usage as much as performance.
In any event it would be incredibly poor timing on Apples part to actually increase the average power usage of it's products. It is not just an issue of being green but one of overall costs when people might not be flush with cash.
Dave
Is there a real cost savings?
While I can agree that desktop chips offfer a performance advantage I don't buy the cost savings justification.
Do we have to know how much Apple pays to know that in lots of 1000 a desktop CPU at the same GHz is cheaper and faster than a mobile CPU? Regardless of what Apple pays their can't be too much difference in their +1000 discount from using a desktop version over a mobile version.
Do we have to know how much Apple pays to know that in lots of 1000 a desktop CPU at the same GHz is cheaper and faster than a mobile CPU? Regardless of what Apple pays their can't be too much difference in their +1000 discount from using a desktop version over a mobile version.
There is if the mobile version costs significantly more to make at any given mhz. Apple may be paying the same amount for mobile chips as they would desktop chips, but they're getting a much slower product that is without affordable quad core options.
There is if the mobile version costs significantly more to make at any given mhz. Apple may be paying the same amount for mobile chips as they would desktop chips, but they're getting a much slower product that is without affordable quad core options.
Come again? If they cost more to make, then it stands to reason that intel is charging more for them. We aren't talking about low production chips for workstations that will have a considerably higher margin, we are talking about consumer-grade desktop and mobile chips. If Intel charges $300 for an Quad-core desktop chip at 2.93GHz and $300 for a Dual core mobile chip at 2.66GHz, and is still buying the same amount of chips either way, why would you think what Apple pays would be excessively different from the desktop chip at $300 per 1000 and the mobile chip at $300 per 1000. You'll have to supply some evidence to support Intel offering vastly different margins on their popular chips simply because one is mobile- and one is desktop-grade for me to swallow that.
Come again? If they cost more to make, then it stands to reason that intel is charging more for them. We aren't talking about low production chips for workstations that will have a considerably higher margin, we are talking about consumer-grade desktop and mobile chips. If Intel charges $300 for an Quad-core desktop chip at 2.93GHz and $300 for a Dual core mobile chip at 2.66GHz, and is still buying the same amount of chips either way, why would you think what Apple pays would be excessively different from the desktop chip at $300 per 1000 and the mobile chip at $300 per 1000. You'll have to supply some evidence to support Intel offering vastly different margins on their popular chips simply because one is mobile- and one is desktop-grade for me to swallow that.
I'm not. I'm saying you're getting a lot more in that desktop quad core for your $300 than the mobile dual core.
Come again? If they cost more to make, then it stands to reason that intel is charging more for them.
We really don't know what it actually costs to make a mobile chips vs a desktop. Certainly there must be a tweak in the process but does that make up for the savings in chip real estate? I don't know but I've ling been under the impression that it is the area of the die that in the end controls the cost of the device.
By the way i would suggest that Intel charges more for them because they can. Mobile chips are seen as premium devices, so corresponding higher prices. The question is does the process Intel uses justify the higher prices. I suspect that the cost to deliver a mobile chip isn't all that much different than a desktop chip.
We aren't talking about low production chips for workstations that will have a considerably higher margin, we are talking about consumer-grade desktop and mobile chips. If Intel charges $300 for an Quad-core desktop chip at 2.93GHz and $300 for a Dual core mobile chip at 2.66GHz, and is still buying the same amount of chips either way, why would you think what Apple pays would be excessively different from the desktop chip at $300 per 1000 and the mobile chip at $300 per 1000. You'll have to supply some evidence to support Intel offering vastly different margins on their popular chips simply because one is mobile- and one is desktop-grade for me to swallow that.
Well we all know here that figuring out Intels margins on chips is going to be difficult. It is not like they are going to publish such info. In any event in Apples case there is noting low production about mobile chips as they out pace desktop sales. I'm just not convinced that the pricing difference in mobile chips is directly related to cost of production. Throw a special part number on the chip and sell it to Apple and Intel can come to any sort of pricing arraignment they want with Apple. After all if the product is Apple specific the competition can't charge Intel with unfair business practices.
Dave
We really don't know what it actually costs to make a mobile chips vs a desktop.
[...]
I'm just not convinced that the pricing difference in mobile chips is directly related to cost of production.
[...]
Why all the emphasis on Intel's actual cost? We know what they charge for batches of 1000. Why assume that a $300 mobile chip in batches of 1000 are going to be much different in price to desktop chips selling at $300 in batches of 1000? The argument that we can't know every little detail about Intel's margins per chip model and Apple's deal with Intel means we can't get an idea of the cost to speed increase from mobile to desktop is ridiculous. We know Intel's base price so we can draw a valid comparison based on the data we do have. Their is no need to make it complex.
PS:
But do they cost more to make?
Trying to figure that out is irreverent to the discussion, but if I were to guess I'd say mobile chips have a higher mark up, which means that that a desktop chip at the same price as a mobile chip is more expensive from R&D through production. My reasoning is, besides the higher demand for mobile chips that you mentioned, is that Intel kills AMD in this area more so than it does in the desktop arena.
I fail to see how Intel's manufacturing cost is relevant.
We are discussing the options available to Apple based on known information about Intel's chip CPU, TDP, L2 at comparable prices. Explain to me why it's relevant to know if Intel has reduced R&D for a desktop chip of reduced factory costs in China for a specific chip if a mobile chip at a lower clock speed costs the same. When you compare a Sony and LG HDTV are you comparing the specs and overall quality or are you waiting to know how much R&D went into each display before buying?
We are discussing the options available to Apple based on known information about Intel's chip CPU, TDP, L2 at comparable prices. Explain to me why it's relevant to know if Intel has reduced R&D for a desktop chip of reduced factory costs in China for a specific chip if a mobile chip at a lower clock speed costs the same. When you compare a Sony and LG HDTV are you comparing the specs and overall quality or are you waiting to know how much R&D went into each display before buying?
All 45nm Core 2 processors are manufactured using the same process at the same fabs. All things being equal (clock speed, FSB, cache size), a mobile part and a desktop part would be identical silicon. The only difference between them would be binning and the package they go on. They cost the same to make, more or less.
I don't know if that answers your question or not.
The only difference between them would be binning and the package they go on. They cost the same to make, more or less.
Why go back to production and R&D costs when we have the price list that Intel sells their chips at. That should be our starting point. Take the $316 chip used in the iMac and see what the equivalent would be in performance to see the price and take the equivalent chip in price to see the performance. I don't understand why we need to complicate the issue with details we'll never know.
Why go back to production and R&D costs when we have the price list that Intel sells their chips at. That should be our starting point. Take the $316 chip used in the iMac and see what the equivalent would be in performance to see the price and take the equivalent chip in price to see the performance. I don't understand why we need to complicate the issue with details we'll never know.
I agree, so let me try:
iMac (current generation): 2.40GHz 1066FSB 6MB cache
mobile equivalent none, closest to it:
P8600 2.40GHz 1066FSB 3MB cache $241
SP9400 2.40GHz 1066FSB 6MB cache $316
desktop equivalent: E8200 2.66GHz 1066FSB 6MB cache $163
At a similar price, the upcoming 65W 2.33 quad-core $245
iMac (current generation): 2.66GHz 1066FSB 6MB cache
mobile equivalent none, closest to it:
T9400 2.53GHz 1066FSB 6MB cache 35W $316
upcoming (dec 28) T9500 2.66GHz 1066FSB 6MB cache 35W $316 probably
desktop equivalent: E8200 2.66GHz 1066FSB 6MB cache $163
At a similar price, the upcoming 65W 2.66 quad-core $320
iMac (current generation): 2.80GHz 1066FSB 6MB cache
mobile equivalent: T9600 2.80GHz 1066FSB 6MB cache 35W $530
desktop equivalent: E8300 2.83GHz 1066FSB 6MB cache $163
Even more affordable, the upcoming 65W 2.93 quad-core $369
iMac (current generation): 3.06GHz 1066FSB 6MB cache
mobile equivalent: X9100 3.06GHz 1066FSB 6MB cache 45W $851
desktop equivalent: E8400 3.00GHz 1066FSB 6MB cache $163
or E8500 3.16GHz 1066FSB 6MB cache $183
One thing to take into account nevertheless, is that the cpus in the current iMac are probably cheaper than the mobile ones given that they have lesser power restrictions (higher TDP). So We must expect that those cpus would cost 10-20% less than equivalent mobile (25-35W) ones:
iMac 2.40GHz cpu for probably $222 (if it was available on Intel's price list)
iMac 2.66GHz cpu for probably $253 (if it was available on Intel's price list)
iMac 2.80GHz cpu for probably $424 (if it was available on Intel's price list)
iMac 3.06GHz cpu for probably $681 (if it was available on Intel's price list)
That doesn't change the fact that desktop dual-core cpus would be less expensive at similar speeds/cache, and that even most of the 65W quad-core cpus would also be less expensive.
Given that the new iMacs will probably get LED-BL displays (the 24" panel is already in inventory, that leaves the 20" one), it will cost more to build, but in choosing 65W desktop cpus, it could not only give a nice speedbump but also keep the prices at similar levels as today while keeping the margins the same (or better).
$1299 LED-BL 20" iMac 65W dual-core 2.66GHz, 2GB RAM, 320GB HDD, Superdrive * small price bump due to the LED-BL display and 1-?2GB of RAM
$1499 LED-BL 20" iMac 65W quad-core 2.33GHz, 2GB RAM, 500GB HDD, Superdrive * price unchanged
$1799 LED-BL 24" iMac 65W quad-core 2.66GHz, 2GB RAM, 500GB HDD, Superdrive * price unchanged
$1999 LED-BL 24" iMac 65W quad-core 2.93GHz, 2GB RAM, 750GB HDD, Superdrive * price cut due to the price of the cpu ($369 vs about $681)
Stretching it a little, we could also see an all quad-core line-up:
$1399 LED-BL 20" iMac quad-core 2.33GHz, 2GB RAM, 320GB HDD, Superdrive
$1599 LED-BL 20" iMac quad-core 2.66GHz, 2GB RAM, 500GB HDD, Superdrive
$1799 LED-BL 24" iMac quad-core 2.66GHz, 2GB RAM, 500GB HDD, Superdrive
$1999 LED-BL 24" iMac quad-core 2.93GHz, 2GB RAM, 750GB HDD, Superdrive
This would give more room for the Mac "mini" to be revised/replaced ($799/999/1199)... I let your imagination think about that one!
But the customer will still see quad core desktops on the market 'NOW' for less than a grand with a better gpu than the iMac has. It's embarrassing to be behind in this respect.
Never mind the iMac and multicore. The thing that's holding me back from getting a mini and may eventually drive me to getting a Linux box is the slow GPU that makes the mini inadequate for HDTV. I want a box to replace my VCR and DVD player and a consumer digital recorder is expensive for what you get.
The crazy thing is, before I realized that the GPU was the main bottleneck, I planned on buying a 1.6GHz superdrive mini off ebay and upgrading it with a 2.33GHz part also off ebay, but the used market is pricing them so high, it's not worth until (until the 2.33GHz part drops a bit more). Just as well I spotted the GPU story in time ...