Well, I read the entire article and every post on the thread and I was still confused as to what was going on. I'm sure much more than half the comments here are absolute bull.
Many thanks to Commander Data for his contribution but it goes far too far into the details of processor architecture to really fill my need, which is simply to get at the basics of what this dispute actually *is*, which CPUs are affected, and how it is going to impact Apple's future products (you know, the actual premise of the article).
Fortunately, I found a comment on this same article reprinted on Dan's own site by a "davebarne"s that itself reprints a newspaper article. I found out more, and in a shorter time by reading this one article that actually *does* get into what is happening in a simple straightforward way.
For the sake of clarity (and sanity), I'm going to reprint it here:
Quote:
Originally Posted by Monica Chen and Ricky Morris, Taipei; Joseph Tsai, DIGITIMES
Nvidia CEO Speaks Out Against Intel Lawsuit [Friday 20 February 2009]
Intel recently filed a lawsuit with the Court of Chancery in the state of Delaware against Nvidia, asking the court to declare that Nvidia is not licensed to produce chipsets that are compatible with any Intel processor that has integrated memory controller functionality, such as Intel?s Nehalem processors, and that Nvidia has breached its agreement with Intel by falsely claiming that it is licensed. Jen-Hsun Huang, president and CEO of Nvidia, has exclusively sat down with Digitimes to discuss why he believes the lawsuit is groundless, and the importance of the case for the future of the industry.
If the courts rule in favor of Intel it could mean that Nvidia would not only be out of the high-end Nehalem-generation (Bloomfield) segment which utilizes Intel?s new QPI bus, which Tom Peterson, director of Technical Marketing for MCP products said in August 2008 the company had no plans to enter, it would also be officially prevented from supporting even CPUs that do not rely on QPI, including Lynnfield for mainstream/low-end desktops, Clarksfield for notebooks, and likely even Lincroft, the next-generation Atom series.
In effect the ruling could significantly reduce the scope of Nvidia?s ability to continue developing chipsets for the Intel segment, at least until a new agreement is signed.
?The disagreement is over the fact that they (Intel) don?t believe we have the right to design chipsets for CPUs with integrated memory controllers, which we do,? said Huang. ?Nvidia entered into an agreement in 2004 in order to bring platform innovations to Intel CPU based systems, and in return, Intel took a license to our rich portfolio of 3D, GPU, and other computing patents.?
Huang revealed that the agreement made with Intel is ?broad? and does not go as far as to name specific technologies. In fact, some of the technologies did not even exist at the time when the agreement was signed, Huang pointed out.
Nvidia has not yet produced or announced any disputable products, and therefore cannot understand why Intel has taken such hostile action, said Huang. Huang pointed out that the trigger point to the disagreement appears to have been Nvidia?s announcement of the Ion platform.
Huang stressed that Nvidia has been supporting Intel CPUs under the current agreement for several generations, and said that its current generation of chipsets, which support FSB-based CPUs including Atom, Celeron, Core 2 Duo and Core 2 Quad, are not affected by the dispute. Nvidia?s motherboard, notebook and system partners currently using these products will not be impacted by the case, Huang reiterated.
Huang said he is confident that the courts will ultimately find that the current license agreement does give Nvidia rights to produce chipsets that support Intel CPUs with integrated memory controllers. Huang added Nvidia will not back down because of the court filing and is not afraid of Intel. This case is about the future and Nvidia?s ability to continue to innovate and make a difference in the industry by creating its own products, not just those that Intel allows it to create, Huang said.
In other news, Huang also revealed that Nvidia will launch an Ion platform (Ion 2) supporting VIA Technologies? Nano CPUs in 2009. The platform is already in development, he added.
1. Apple has always supported multiple GPU architecture with ATI and Nvidia products.
2. How would the Mac Pro be delayed? I don't know of many dual socket Nvidia mobos.
3. OpenCL support is going to be there for Nvidia, ATI and Intel GPU
I think Nvidia support for Corei5 is important but Core i7 isn't likely going to be delivered unless Apple has an epiphany and makes a single socket Mac Pro (which I'd love love love...yes it's xMac)
3. OpenCL support is going to be there for Nvidia, ATI and Intel GPU
Nvidia and AMD GPUs already support OpenCL. As for Intel, even if they bothered to implement OpenCL on the GMA, it is likely too slow to be useful. The first Intel product capable of supporting OpenCL is Larrabee, which is not coming out until early next and will only be available as a discrete part.
So the real risk to Apple is not the iMac which wouldn't have integrated graphics anyway or core i7, [B]it is *the very next refresh of the Aluminium MacBook line... in otherwords, between now and September 2009, if Apple goes Nehalem in mobile, they might have to kiss the 9400M/Nvidia chipset goodbye, or, stick with Core 2 for the next 9 months.
And while this legal stuff goes on for a few months, again, Apple in the same position, forget the iMac, talkin' about the Aluminium MacBooks here... No NEHALEM. NO NEHALEM IN MACBOOK PRO until/ unless Apple goes back and throws out the 9400M/Nvidia-whatever chipset.
1. Intel is not introducing any Nehalem-based mobile products until Q4 of this year.
2. When they do, only Clarksfield will be based on the 45nm Nehalem. The other processor, Arrandale, will be based on the 32nm Westmere (but has only 2 cores and bundles crappy Intel GMA)
3. Even Clarksfield might be too hot for Apple notebooks. I am guessing it's only suitable for the iMac
Therefore, the next Macbook/Macbook Pro refresh (May/June?) will simply use the latest C2D-M and Nvidia 9400m.
Nvidia and AMD GPUs already support OpenCL. As for Intel, even if they bothered to implement OpenCL on the GMA, it is likely too slow to be useful. The first Intel product capable of supporting OpenCL is Larrabee, which is not coming out until early next and will only be available as a discrete part.
I hope people aren't expecting miracles in OpenCL performance for integrated GPU. Mac Pro owners will simply purchase the PCI-E card that delivers the desired performance.
I've always thought that Apple would stick with Nvidia for their consumer offerings and use Intel for their Workstations. My experience has found that at the Workstation/Server level people want genuine Intel parts. An Nvidia based Mac Pro isn't going to please everyone IMO.
Thus I have no idea why Apple has delayed the mini and iMac because they are well suited for Nvidia motherboards and graphics.
...My suggestion was that nVidia focus on fast discrete low-cost GPUs, since those will also be faster than an IGP, not necessarily that much more expensive, and will benefit from the integrated PCIe controller allowing the GPU to be closer to the CPU.
Again, brilliant point. Precisely the focus if Nvidia loses the court case or fails to hammer out a new deal.
In any case I think Nvidia will fight tooth and nail to get as much Nehalem chipset rights as possible, because this goes beyond Nehalem since Westmere has memory controller integrated into CPU as well. Nvidia chipsets is I suspect a big part of their marketing strategy and muscling into "owning" more of the computer. With Apple they're more or less marketing it as an "Nvidia platform" and "oh, BTW, it has Intel CPU". Same with Ion. It's an "Ion Platform" and "oh, BTW, has Intel Atom CPU". Turf war.
Nvidia has a lot to lose if they have to "go back" to just supplying discrete GPUs. Of course, they could counter this, like you said, good point, with really kickass low-cost, low-power, fast discrete GPUs that just destroy whatever Intel IGP comes out. Here's the fly in the ointment though: Nvidia is struggling a little bit. The 8600 M GT fiasco with solder thingys failing. Then if you look at the desktop, eg. GTX 260 or even 9800GTX+, you would really have to wonder, how the heck is Nvidia even going to have a low-power, low-cost discrete version of the GTX 260~280 etc range. Even with the 9600M GT and 9800 mobile versions, they do have a big challenge to really bring down the heat and cost.
...Thus I have no idea why Apple has delayed the mini and iMac because they are well suited for Nvidia motherboards and graphics.
This is the point I have been making all along. There is no reason why the Mac Mini and iMac can't have the Nvida 9400M chipset with fast Penryns and DDR3, etc. Nvidia discrete 9600 and 9800s on the iMac. WTF is Apple doing? That's what many people are asking. What is the hold up? What brilliance is waiting in the wings to be unleashed upon us unsuspecting peons in an explosion of technological shock and awe? A 30"+ LED backlit screen iMac or xMac with Core i7 and Nvidia GTX280? As if.
I hope people aren't expecting miracles in OpenCL performance for integrated GPU...
There is strong evidence that using just the 9400M on the Aluminium MacBook, in CUDA in Windows XP, you can encode H.264 from DVD at 3x faster than a 2.5ghz Penryn CPU (rough estimates). (Badaboom application)
The 9400M when it comes to games is really not that bad. 2007-2008 game titles played at medium to medium-high resolution.
Remember CUDA and OpenCL can give significant boosts. Look at the statistics for Folding (Stanford Uni). The CUDA version absolutely whips the CPU version.
CUDA is showing some distinct promise, and OpenCL is about a year away from really maturing. Finally with Nvidia's 9400 and above integrated or discrete series, GPGPU is starting to become a reality.
It all just shows how horrible even the Intel X3100 is. Just horrible. And what roadmap does Intel have beyond their X4500?
..Therefore, the next Macbook/Macbook Pro refresh (May/June?) will simply use the latest C2D-M and Nvidia 9400m.
Okay, good point. I didn't know mobile Nehalem was that far away.
So, actually, for the first 9 months of this year, Jan to Sep 2009, what sort of CPU gains in the mobile area are we expecting? a few 0.1 to 0.3 GHZ? Core2 Quad mobile CPUs aren't going to be in any of the "mainstream" Apple or PC laptops. So the 9400M and iLife'09, iWork'09, Aluminium was actually some kind of insurance policy for Apple to keep their mainstream laptops "fresh" throughout 2009 without having to spend on discrete GPUs. Absolutely brilliant strategy, now that it is revealed.
More and more I have to say this article written for AppleInsider is a little too forward looking, and does not really explain Mac Mini and iMac details. For 2009, Apple's Mac desktops need that level of ingenuity in strategy that Apple has shown for the notebook line. Which I really think is lacking/ hindered due to Steve's lack of everday presence.
IMHO, Apple should drop the MacBook Alu prices slightly, just do the obvious but really value-infusing Core2 and Nvidia chipset updates to the Mac Mini and iMac, and for the killer blow roll out the new iPhone 3G *AND* the "Mac Touch" (don't call it a netbook, it is a 10" multitouch e-reader/ e-app tablet/ whatever) ~ the "Mac Touch" has built-in 3G and will be sold through their telco partners so as to continue fat revenue flows and spread risk. For the Pro area, release a stunning, calibrated really Pro-suitable LED-backlit or IPS whatever 30" matte display. Mac Pros with Nehalem in a few months and 30" Pro-suitable display will hold the line in Apple's Pro offerings. WWDC June 2009 full force on Snow Leopard with public gold master late September'09. iPods and iTunes Store? Something sexy by middle of the year, slight refresh in October for holiday season. Can Tim Cook pull all this off? No disrespect, but... we'll have to see. I think Tim has big, big shoes to fill.
<okay that's the end of my rambling on for today. thanks for listening >
Well now that all the information is out in the open, this is A LOT less confusing..
1) The Nvidia-Intel dispute involves *ALL* "Nehalem" family and later Intel processors (any processor with an IMC).
2) Even if the licensing is resolved, it is unlikely nVidia will produce chipsets with Integrated graphics because the memory controller/memory are directly connected to the processor, and an IGP on the motherboard would have to connect through the southbridge DMI link. (That's how I understand it... Not sure if they could somehow rig it up for the IGP to interface directly through PCIe to the processor like a discrete part)
What does this all mean? Well, if Apple is serious about supporting OpenCL across the line, it seems unlikely future Macbooks would use an IGP considering it would most likely have to be the Intel-supplied part (which is built in to the processor package anyways). Does this mean a future "arrandale" Macbook would have a discrete GPU? perhaps..
One thing I'm still confused about is whether the quad-core "Lynnfield/Clarksfield" CPUs/platform are going to support IGPs from Intel or anyone else. Unlike the dual-core "arrandale/clarkdale", these quad-core chips don't have an IGP built in to the processor... Wouldn't these chips run into the same problem of having to have an on-motherboard integrated graphics chip go through the southbridge/DMI to the processor and SDRAM?
Having IMC on CPU doesn't really matter much for IGP since they are less sensitive to latency. And bandwidth doesn't matter since either way the IGP had never had enough bandwidth anyway.
Quote:
One thing I'm still confused about is whether the quad-core "Lynnfield/Clarksfield" CPUs/platform are going to support IGPs from Intel or anyone else. Unlike the dual-core "arrandale/clarkdale", these quad-core chips don't have an IGP built in to the processor... Wouldn't these chips run into the same problem of having to have an on-motherboard integrated graphics chip go through the southbridge/DMI to the processor and SDRAM?
And too many others to quote.
But i think everyone is mis understanding DMI 2.0 and QuickPath.
Every Nehalem or Westmere based processors will have QuickPath Inside them. The different is, for Dual Core Processors, the QuickPath will be used to connect the IGP inside the same pacakage. For Quad Core, the QuickPath is still available allowing Chipset ( Southbridge ) with IGP.
And again, once and for all, Core i7 is EXTREMELY UNLIKELY to be included inside iMac. TDP is simply too high. ( As i have already stated on 3rd or 4th reply on this thread )
...Every Nehalem or Westmere based processors will have QuickPath Inside them. The different is, for Dual Core Processors, the QuickPath will be used to connect the IGP inside the same pacakage. For Quad Core, the QuickPath is still available allowing Chipset ( Southbridge ) with IGP...
So maybe what winterspan is asking is for Dual Core, who is able to put that IGP in that same package? Can it be an ATI or Nvidia IGP in the same package? Because if it is only Intel, then all graphics on Nehalem/ Westmere Dual Core that is non-Intel have to go through the PCI Express bus [DMI "southbridge"], right?
Because I think this Intel-Nvidia dispute notwithstanding, Intel AFAIK wouldn't allow anyone else to put that IGP in the CPU package.
And as lt.commander.data pointed out already, any IGP that cannot go into that same package as the CPU has to go through the southbridge, and this "IGP" doesn't have its own memory, so it's really hampered. It has to go through the PCI Express, southbridge, to the integrated "northbridge" memory controller then all the way back for all RAM access.
So was the whole thing with the IGP push like we have currently with GMAs and so on ... is it so the manufacturers can save on VRAM for the GPUs? Because they just use shared system memory. Is not having to use dedicated VRAM the big savings Intel/Nvidia/ATI make when it comes to offering integrated graphics?
So maybe what winterspan is asking is for Dual Core, who is able to put that IGP in that same package? Can it be an ATI or Nvidia IGP in the same package? Because if it is only Intel, then all graphics on Nehalem/ Westmere Dual Core that is non-Intel have to go through the PCI Express bus [DMI "southbridge"], right?
Because I think this Intel-Nvidia dispute notwithstanding, Intel AFAIK wouldn't allow anyone else to put that IGP in the CPU package.
And as lt.commander.data pointed out already, any IGP that cannot go into that same package as the CPU has to go through the southbridge, and this "IGP" doesn't have its own memory, so it's really hampered. It has to go through the PCI Express, southbridge, to the integrated "northbridge" memory controller then all the way back for all RAM access.
So was the whole thing with the IGP push like we have currently with GMAs and so on ... is it so the manufacturers can save on VRAM for the GPUs? Because they just use shared system memory. Is not having to use dedicated VRAM the big savings Intel/Nvidia/ATI make when it comes to offering integrated graphics?
1. Since Intel manufacture the whole CPU package, there is Zero chance for ATI or Nvidia's chip ( die ) to be included inside Intel CPU.
2. Since Nehalem and Westmere actually have NorthBridge intergated. Therefore there are also PCI-Express included inside CPU. So to directly answer your question.
Quote:
all graphics on Nehalem/ Westmere Dual Core that is non-Intel have to go through the PCI Express bus [DMI "southbridge"], right?
No, It will be a PCI Express directly from the CPU.
My point was never the inability to add a discrete GPU to any Nehalem or Westmere parts because as you noted they have integrated PCIe controllers. The AppleInsider article was concerned about nVidia IGPs on nVidia chipsets, and my point is that even if nVIdia got a license to produce chipsets for Nehalem, those chipsets aren't likely to have viable IGPs like the current 9400M IGP/chipset because the memory controller is no longer on the chipset and an external IGP will need to connect through DMI which as you note is designed for southbridges and doesn't have the bandwidth. My suggestion was that nVidia focus on fast discrete low-cost GPUs, since those will also be faster than an IGP, not necessarily that much more expensive, and will benefit from the integrated PCIe controller allowing the GPU to be closer to the CPU.
right on. I don't have any plans to game on my future macbook (got a nice windows 7 dual monitor desktop with a gtx 285) so I really don't care about igp power as long as it runs the ui and decodes the occasional h264 vid fine. Hopefully the westmere will allow even more battery life, the macbook is already very impressive for a 13''.
...Now the mainstream Nehalem quad core chip is going to be Lynnfield in desktop and Clarksfield in mobile. They are both quad core, 0 QPI, and 2 channel memory. Notice the key detail: 0 QPI. Mainstream desktop and all mobile chips won't have QPI and will only have DMI, because the entire northbridge, in this case memory controller and PCIe controller is integrated into Lynnfield and Clarksfield. DMI is designed as a low-bandwidth link to connect a northbridge to a southbridge. Lynnfield and Clarksfield won't have an Intel IGP on package, but you won't want to attach an IGP through the DMI link since it's low bandwidth, only 10Gb/s, and with the memory controller on the CPU, an IGP running through DMI would basically have 133MHz SDRAM like bandwidth...
OK, now the question I have is why wouldn't Nvidia try and put an IGP off the PCIe interface? Is this not possible? By integrated graphics I mean it's on the motherboard chipset and shares memory with main memory.
1. Since Intel manufacture the whole CPU package, there is Zero chance for ATI or Nvidia's chip ( die ) to be included inside Intel CPU.
2. Since Nehalem and Westmere actually have NorthBridge intergated. Therefore there are also PCI-Express included inside CPU. So to directly answer your question.
No, It will be a PCI Express directly from the CPU.
OK got it thanks. See my question above, are you able to shed some light on that?
Comments
Many thanks to Commander Data for his contribution but it goes far too far into the details of processor architecture to really fill my need, which is simply to get at the basics of what this dispute actually *is*, which CPUs are affected, and how it is going to impact Apple's future products (you know, the actual premise of the article).
Fortunately, I found a comment on this same article reprinted on Dan's own site by a "davebarne"s that itself reprints a newspaper article. I found out more, and in a shorter time by reading this one article that actually *does* get into what is happening in a simple straightforward way.
For the sake of clarity (and sanity), I'm going to reprint it here:
Nvidia CEO Speaks Out Against Intel Lawsuit [Friday 20 February 2009]
Intel recently filed a lawsuit with the Court of Chancery in the state of Delaware against Nvidia, asking the court to declare that Nvidia is not licensed to produce chipsets that are compatible with any Intel processor that has integrated memory controller functionality, such as Intel?s Nehalem processors, and that Nvidia has breached its agreement with Intel by falsely claiming that it is licensed. Jen-Hsun Huang, president and CEO of Nvidia, has exclusively sat down with Digitimes to discuss why he believes the lawsuit is groundless, and the importance of the case for the future of the industry.
If the courts rule in favor of Intel it could mean that Nvidia would not only be out of the high-end Nehalem-generation (Bloomfield) segment which utilizes Intel?s new QPI bus, which Tom Peterson, director of Technical Marketing for MCP products said in August 2008 the company had no plans to enter, it would also be officially prevented from supporting even CPUs that do not rely on QPI, including Lynnfield for mainstream/low-end desktops, Clarksfield for notebooks, and likely even Lincroft, the next-generation Atom series.
In effect the ruling could significantly reduce the scope of Nvidia?s ability to continue developing chipsets for the Intel segment, at least until a new agreement is signed.
?The disagreement is over the fact that they (Intel) don?t believe we have the right to design chipsets for CPUs with integrated memory controllers, which we do,? said Huang. ?Nvidia entered into an agreement in 2004 in order to bring platform innovations to Intel CPU based systems, and in return, Intel took a license to our rich portfolio of 3D, GPU, and other computing patents.?
Huang revealed that the agreement made with Intel is ?broad? and does not go as far as to name specific technologies. In fact, some of the technologies did not even exist at the time when the agreement was signed, Huang pointed out.
Nvidia has not yet produced or announced any disputable products, and therefore cannot understand why Intel has taken such hostile action, said Huang. Huang pointed out that the trigger point to the disagreement appears to have been Nvidia?s announcement of the Ion platform.
Huang stressed that Nvidia has been supporting Intel CPUs under the current agreement for several generations, and said that its current generation of chipsets, which support FSB-based CPUs including Atom, Celeron, Core 2 Duo and Core 2 Quad, are not affected by the dispute. Nvidia?s motherboard, notebook and system partners currently using these products will not be impacted by the case, Huang reiterated.
Huang said he is confident that the courts will ultimately find that the current license agreement does give Nvidia rights to produce chipsets that support Intel CPUs with integrated memory controllers. Huang added Nvidia will not back down because of the court filing and is not afraid of Intel. This case is about the future and Nvidia?s ability to continue to innovate and make a difference in the industry by creating its own products, not just those that Intel allows it to create, Huang said.
In other news, Huang also revealed that Nvidia will launch an Ion platform (Ion 2) supporting VIA Technologies? Nano CPUs in 2009. The platform is already in development, he added.
The article is incomprehensible.
1. Apple has always supported multiple GPU architecture with ATI and Nvidia products.
2. How would the Mac Pro be delayed? I don't know of many dual socket Nvidia mobos.
3. OpenCL support is going to be there for Nvidia, ATI and Intel GPU
I think Nvidia support for Corei5 is important but Core i7 isn't likely going to be delivered unless Apple has an epiphany and makes a single socket Mac Pro (which I'd love love love...yes it's xMac)
3. OpenCL support is going to be there for Nvidia, ATI and Intel GPU
Nvidia and AMD GPUs already support OpenCL. As for Intel, even if they bothered to implement OpenCL on the GMA, it is likely too slow to be useful. The first Intel product capable of supporting OpenCL is Larrabee, which is not coming out until early next and will only be available as a discrete part.
So the real risk to Apple is not the iMac which wouldn't have integrated graphics anyway or core i7, [B]it is *the very next refresh of the Aluminium MacBook line... in otherwords, between now and September 2009, if Apple goes Nehalem in mobile, they might have to kiss the 9400M/Nvidia chipset goodbye, or, stick with Core 2 for the next 9 months.
And while this legal stuff goes on for a few months, again, Apple in the same position, forget the iMac, talkin' about the Aluminium MacBooks here... No NEHALEM. NO NEHALEM IN MACBOOK PRO until/ unless Apple goes back and throws out the 9400M/Nvidia-whatever chipset.
1. Intel is not introducing any Nehalem-based mobile products until Q4 of this year.
2. When they do, only Clarksfield will be based on the 45nm Nehalem. The other processor, Arrandale, will be based on the 32nm Westmere (but has only 2 cores and bundles crappy Intel GMA)
3. Even Clarksfield might be too hot for Apple notebooks. I am guessing it's only suitable for the iMac
Therefore, the next Macbook/Macbook Pro refresh (May/June?) will simply use the latest C2D-M and Nvidia 9400m.
Nvidia and AMD GPUs already support OpenCL. As for Intel, even if they bothered to implement OpenCL on the GMA, it is likely too slow to be useful. The first Intel product capable of supporting OpenCL is Larrabee, which is not coming out until early next and will only be available as a discrete part.
I hope people aren't expecting miracles in OpenCL performance for integrated GPU. Mac Pro owners will simply purchase the PCI-E card that delivers the desired performance.
I've always thought that Apple would stick with Nvidia for their consumer offerings and use Intel for their Workstations. My experience has found that at the Workstation/Server level people want genuine Intel parts. An Nvidia based Mac Pro isn't going to please everyone IMO.
Thus I have no idea why Apple has delayed the mini and iMac because they are well suited for Nvidia motherboards and graphics.
...My suggestion was that nVidia focus on fast discrete low-cost GPUs, since those will also be faster than an IGP, not necessarily that much more expensive, and will benefit from the integrated PCIe controller allowing the GPU to be closer to the CPU.
Again, brilliant point. Precisely the focus if Nvidia loses the court case or fails to hammer out a new deal.
In any case I think Nvidia will fight tooth and nail to get as much Nehalem chipset rights as possible, because this goes beyond Nehalem since Westmere has memory controller integrated into CPU as well. Nvidia chipsets is I suspect a big part of their marketing strategy and muscling into "owning" more of the computer. With Apple they're more or less marketing it as an "Nvidia platform" and "oh, BTW, it has Intel CPU". Same with Ion. It's an "Ion Platform" and "oh, BTW, has Intel Atom CPU". Turf war.
Nvidia has a lot to lose if they have to "go back" to just supplying discrete GPUs. Of course, they could counter this, like you said, good point, with really kickass low-cost, low-power, fast discrete GPUs that just destroy whatever Intel IGP comes out. Here's the fly in the ointment though: Nvidia is struggling a little bit. The 8600 M GT fiasco with solder thingys failing. Then if you look at the desktop, eg. GTX 260 or even 9800GTX+, you would really have to wonder, how the heck is Nvidia even going to have a low-power, low-cost discrete version of the GTX 260~280 etc range. Even with the 9600M GT and 9800 mobile versions, they do have a big challenge to really bring down the heat and cost.
...Thus I have no idea why Apple has delayed the mini and iMac because they are well suited for Nvidia motherboards and graphics.
This is the point I have been making all along. There is no reason why the Mac Mini and iMac can't have the Nvida 9400M chipset with fast Penryns and DDR3, etc. Nvidia discrete 9600 and 9800s on the iMac. WTF is Apple doing? That's what many people are asking. What is the hold up? What brilliance is waiting in the wings to be unleashed upon us unsuspecting peons in an explosion of technological shock and awe? A 30"+ LED backlit screen iMac or xMac with Core i7 and Nvidia GTX280? As if.
I hope people aren't expecting miracles in OpenCL performance for integrated GPU...
There is strong evidence that using just the 9400M on the Aluminium MacBook, in CUDA in Windows XP, you can encode H.264 from DVD at 3x faster than a 2.5ghz Penryn CPU (rough estimates). (Badaboom application)
The 9400M when it comes to games is really not that bad. 2007-2008 game titles played at medium to medium-high resolution.
Remember CUDA and OpenCL can give significant boosts. Look at the statistics for Folding (Stanford Uni). The CUDA version absolutely whips the CPU version.
CUDA is showing some distinct promise, and OpenCL is about a year away from really maturing. Finally with Nvidia's 9400 and above integrated or discrete series, GPGPU is starting to become a reality.
It all just shows how horrible even the Intel X3100 is. Just horrible. And what roadmap does Intel have beyond their X4500?
..Therefore, the next Macbook/Macbook Pro refresh (May/June?) will simply use the latest C2D-M and Nvidia 9400m.
Okay, good point. I didn't know mobile Nehalem was that far away.
So, actually, for the first 9 months of this year, Jan to Sep 2009, what sort of CPU gains in the mobile area are we expecting? a few 0.1 to 0.3 GHZ? Core2 Quad mobile CPUs aren't going to be in any of the "mainstream" Apple or PC laptops. So the 9400M and iLife'09, iWork'09, Aluminium was actually some kind of insurance policy for Apple to keep their mainstream laptops "fresh" throughout 2009 without having to spend on discrete GPUs. Absolutely brilliant strategy, now that it is revealed.
More and more I have to say this article written for AppleInsider is a little too forward looking, and does not really explain Mac Mini and iMac details. For 2009, Apple's Mac desktops need that level of ingenuity in strategy that Apple has shown for the notebook line. Which I really think is lacking/ hindered due to Steve's lack of everday presence.
IMHO, Apple should drop the MacBook Alu prices slightly, just do the obvious but really value-infusing Core2 and Nvidia chipset updates to the Mac Mini and iMac, and for the killer blow roll out the new iPhone 3G *AND* the "Mac Touch" (don't call it a netbook, it is a 10" multitouch e-reader/ e-app tablet/ whatever) ~ the "Mac Touch" has built-in 3G and will be sold through their telco partners so as to continue fat revenue flows and spread risk. For the Pro area, release a stunning, calibrated really Pro-suitable LED-backlit or IPS whatever 30" matte display. Mac Pros with Nehalem in a few months and 30" Pro-suitable display will hold the line in Apple's Pro offerings. WWDC June 2009 full force on Snow Leopard with public gold master late September'09. iPods and iTunes Store? Something sexy by middle of the year, slight refresh in October for holiday season. Can Tim Cook pull all this off? No disrespect, but... we'll have to see. I think Tim has big, big shoes to fill.
<okay that's the end of my rambling on for today. thanks for listening
1) The Nvidia-Intel dispute involves *ALL* "Nehalem" family and later Intel processors (any processor with an IMC).
2) Even if the licensing is resolved, it is unlikely nVidia will produce chipsets with Integrated graphics because the memory controller/memory are directly connected to the processor, and an IGP on the motherboard would have to connect through the southbridge DMI link. (That's how I understand it... Not sure if they could somehow rig it up for the IGP to interface directly through PCIe to the processor like a discrete part)
What does this all mean? Well, if Apple is serious about supporting OpenCL across the line, it seems unlikely future Macbooks would use an IGP considering it would most likely have to be the Intel-supplied part (which is built in to the processor package anyways). Does this mean a future "arrandale" Macbook would have a discrete GPU? perhaps..
One thing I'm still confused about is whether the quad-core "Lynnfield/Clarksfield" CPUs/platform are going to support IGPs from Intel or anyone else. Unlike the dual-core "arrandale/clarkdale", these quad-core chips don't have an IGP built in to the processor... Wouldn't these chips run into the same problem of having to have an on-motherboard integrated graphics chip go through the southbridge/DMI to the processor and SDRAM?
One thing I'm still confused about is whether the quad-core "Lynnfield/Clarksfield" CPUs/platform are going to support IGPs from Intel or anyone else. Unlike the dual-core "arrandale/clarkdale", these quad-core chips don't have an IGP built in to the processor... Wouldn't these chips run into the same problem of having to have an on-motherboard integrated graphics chip go through the southbridge/DMI to the processor and SDRAM?
And too many others to quote.
But i think everyone is mis understanding DMI 2.0 and QuickPath.
Every Nehalem or Westmere based processors will have QuickPath Inside them. The different is, for Dual Core Processors, the QuickPath will be used to connect the IGP inside the same pacakage. For Quad Core, the QuickPath is still available allowing Chipset ( Southbridge ) with IGP.
And again, once and for all, Core i7 is EXTREMELY UNLIKELY to be included inside iMac. TDP is simply too high. ( As i have already stated on 3rd or 4th reply on this thread )
...Every Nehalem or Westmere based processors will have QuickPath Inside them. The different is, for Dual Core Processors, the QuickPath will be used to connect the IGP inside the same pacakage. For Quad Core, the QuickPath is still available allowing Chipset ( Southbridge ) with IGP...
So maybe what winterspan is asking is for Dual Core, who is able to put that IGP in that same package? Can it be an ATI or Nvidia IGP in the same package? Because if it is only Intel, then all graphics on Nehalem/ Westmere Dual Core that is non-Intel have to go through the PCI Express bus [DMI "southbridge"], right?
Because I think this Intel-Nvidia dispute notwithstanding, Intel AFAIK wouldn't allow anyone else to put that IGP in the CPU package.
And as lt.commander.data pointed out already, any IGP that cannot go into that same package as the CPU has to go through the southbridge, and this "IGP" doesn't have its own memory, so it's really hampered. It has to go through the PCI Express, southbridge, to the integrated "northbridge" memory controller then all the way back for all RAM access.
So was the whole thing with the IGP push like we have currently with GMAs and so on ... is it so the manufacturers can save on VRAM for the GPUs? Because they just use shared system memory. Is not having to use dedicated VRAM the big savings Intel/Nvidia/ATI make when it comes to offering integrated graphics?
So maybe what winterspan is asking is for Dual Core, who is able to put that IGP in that same package? Can it be an ATI or Nvidia IGP in the same package? Because if it is only Intel, then all graphics on Nehalem/ Westmere Dual Core that is non-Intel have to go through the PCI Express bus [DMI "southbridge"], right?
Because I think this Intel-Nvidia dispute notwithstanding, Intel AFAIK wouldn't allow anyone else to put that IGP in the CPU package.
And as lt.commander.data pointed out already, any IGP that cannot go into that same package as the CPU has to go through the southbridge, and this "IGP" doesn't have its own memory, so it's really hampered. It has to go through the PCI Express, southbridge, to the integrated "northbridge" memory controller then all the way back for all RAM access.
So was the whole thing with the IGP push like we have currently with GMAs and so on ... is it so the manufacturers can save on VRAM for the GPUs? Because they just use shared system memory. Is not having to use dedicated VRAM the big savings Intel/Nvidia/ATI make when it comes to offering integrated graphics?
1. Since Intel manufacture the whole CPU package, there is Zero chance for ATI or Nvidia's chip ( die ) to be included inside Intel CPU.
2. Since Nehalem and Westmere actually have NorthBridge intergated. Therefore there are also PCI-Express included inside CPU. So to directly answer your question.
all graphics on Nehalem/ Westmere Dual Core that is non-Intel have to go through the PCI Express bus [DMI "southbridge"], right?
No, It will be a PCI Express directly from the CPU.
My point was never the inability to add a discrete GPU to any Nehalem or Westmere parts because as you noted they have integrated PCIe controllers. The AppleInsider article was concerned about nVidia IGPs on nVidia chipsets, and my point is that even if nVIdia got a license to produce chipsets for Nehalem, those chipsets aren't likely to have viable IGPs like the current 9400M IGP/chipset because the memory controller is no longer on the chipset and an external IGP will need to connect through DMI which as you note is designed for southbridges and doesn't have the bandwidth. My suggestion was that nVidia focus on fast discrete low-cost GPUs, since those will also be faster than an IGP, not necessarily that much more expensive, and will benefit from the integrated PCIe controller allowing the GPU to be closer to the CPU.
right on. I don't have any plans to game on my future macbook (got a nice windows 7 dual monitor desktop with a gtx 285) so I really don't care about igp power as long as it runs the ui and decodes the occasional h264 vid fine. Hopefully the westmere will allow even more battery life, the macbook is already very impressive for a 13''.
...Now the mainstream Nehalem quad core chip is going to be Lynnfield in desktop and Clarksfield in mobile. They are both quad core, 0 QPI, and 2 channel memory. Notice the key detail: 0 QPI. Mainstream desktop and all mobile chips won't have QPI and will only have DMI, because the entire northbridge, in this case memory controller and PCIe controller is integrated into Lynnfield and Clarksfield. DMI is designed as a low-bandwidth link to connect a northbridge to a southbridge. Lynnfield and Clarksfield won't have an Intel IGP on package, but you won't want to attach an IGP through the DMI link since it's low bandwidth, only 10Gb/s, and with the memory controller on the CPU, an IGP running through DMI would basically have 133MHz SDRAM like bandwidth...
OK, now the question I have is why wouldn't Nvidia try and put an IGP off the PCIe interface? Is this not possible? By integrated graphics I mean it's on the motherboard chipset and shares memory with main memory.
1. Since Intel manufacture the whole CPU package, there is Zero chance for ATI or Nvidia's chip ( die ) to be included inside Intel CPU.
2. Since Nehalem and Westmere actually have NorthBridge intergated. Therefore there are also PCI-Express included inside CPU. So to directly answer your question.
No, It will be a PCI Express directly from the CPU.
OK got it thanks. See my question above, are you able to shed some light on that?