I can see the Mac Pro becoming immensely more powerful by 2015.
Are you suggesting they might go from 2x GPUs to 1 GPU + Xeon Phi? I'm a little picky, because at this point I'm primarily tied to OSX for an XCode machine. In the past that wasn't the case.
Are you suggesting they might go from 2x GPUs to 1 GPU + Xeon Phi? I'm a little picky, because at this point I'm primarily tied to OSX for an XCode machine. In the past that wasn't the case.
Well you know Apple, it is hard to say what they will do. I'm imagining a Mac Pro with two GPUs as today with the new XEON Phi replacing the current XEON. GPUs still have significant advantages over anything Intel has to offer so I don't see the two GPUs going away. The interesting thing here is Intel making variants of Phi that can operate as system processors. Instead of struggling to go from 8-12 cores in a Pro, Apple could easily jump to 32, 64 or any number they can come to an agreement with Intel to build.
As far as XCode goes imagine a machine with 32 or 64 cores compiling your apps.
Of course these aren't run of the mill i86 cores. But the interesting thing here is if they can do this for Phi and do it for Haswell Iris they most certainly can do it for run of the mill XEON. In other words I think thing look bright for the Mac Pro.
Well you know Apple, it is hard to say what they will do. I'm imagining a Mac Pro with two GPUs as today with the new XEON Phi replacing the current XEON. GPUs still have significant advantages over anything Intel has to offer so I don't see the two GPUs going away. The interesting thing here is Intel making variants of Phi that can operate as system processors. Instead of struggling to go from 8-12 cores in a Pro, Apple could easily jump to 32, 64 or any number they can come to an agreement with Intel to build.
As far as XCode goes imagine a machine with 32 or 64 cores compiling your apps.
Of course these aren't run of the mill i86 cores. But the interesting thing here is if they can do this for Phi and do it for Haswell Iris they most certainly can do it for run of the mill XEON. In other words I think thing look bright for the Mac Pro.
Yeah I have no idea about that. I haven't seen intel bring out anything Phi-wise that costs less than a few thousand. I will mention that chipsets will change again with Haswell E whenever that comes out, so there's nothing past this point until probably some point in 2015. Ivy Bridge E is just rolling out at the moment across oems.
Quote:
Originally Posted by wizard69
I can see FKK beaches in Relics future! After all we don't want to be seeing nasty tan lines.
I request a rendition of Rondeau within that playlist, just for the cheesy masterpiece theatre reference. Anything by Einaudi would also be cool.
Yeah I have no idea about that. I haven't seen intel bring out anything Phi-wise that costs less than a few thousand. I will mention that chipsets will change again with Haswell E whenever that comes out, so there's nothing past this point until probably some point in 2015. Ivy Bridge E is just rolling out at the moment across oems.
I suspect if Intel is to have success with Phi they will need a way to mass market it. Cost wise it isn't significantly worst than Xeon now. I think the key here is that this is an indication of the direction of technology. If they integrate RAM into the Haswell Iris packages and the Phi, it is possible for other chips focusing on performance. The trickle down theory of computer systems if you will. I
I request a rendition of Rondeau within that playlist, just for the cheesy masterpiece theatre reference. Anything by Einaudi would also be cool.
Music has never been a big part of my life, but I will listen to anything Relic wishes to play if it means getting well.
I suspect if Intel is to have success with Phi they will need a way to mass market it. Cost wise it isn't significantly worst than Xeon now. I think the key here is that this is an indication of the direction of technology. If they integrate RAM into the Haswell Iris packages and the Phi, it is possible for other chips focusing on performance. The trickle down theory of computer systems if you will.
Well it may not be much worse compared to the higher core chips, but it definitely covers different price territory unless I have missed recent developments. The Xeons that Apple has used over the past 4 years start at $300. I am curious how that thing will look in the wild, as right now the price barrier seems awfully high relative to what is included. I expected to see either a lower price than what it ended up as or the cost cutting measures (chopped down framebuffers, quad cpu) , not necessarily both.
Quote:
Music has never been a big part of my life, but I will listen to anything Relic wishes to play if it means getting well.
Bleh I like music, but I can't deal with anything lyrical while working. It's too much of a distraction.
Initial devices will be shipping in 2 & 4 GB offerings, not bad really. Now I haven't seen any indication that Intel or Apple are on board, but it would be hard to imagine they aren't. The thing here is what such technology could do for the likes of the Mac Pro as we dramatically increase storage density while lowering power. I could see Apple implementing this tech in a Mac Pro in a couple of years.
However it becomes even more interesting if you see this as almost the perfect technology to take iPads pass the 1GB barrier. I'm not about to discount the chips initial expense but think about what these memory cubes would allow Apple to do in an iPad. First you have a massive reduction in size. Second an estimated 70% reduction in power used per bit. Third amazing bandwidth.
On an iPad this would all work together to give us another stepped increase in performance. Apple could easily give us a 4X increase in available RAM, which Apps really need, while giving the GPU the bandwidth it needs. Bandwidth would be especially important on a larger iPad with considerably more pixels. It is the advantage to the GPU that I see as the big win here, in every APU design on the market the manufacture run up against the bandwidth problem trying to squeeze more performance out of the GPU.
Intel has gotten good results with Iris Plus using a multi chip module where the extra silicon is high speed memory. That isn't bad at all, however Memory Cube technology would offer that bandwidth to the entire memory array better support the GPU for traditional graphics and more importantly heterogenous computing.
Now some might argue that this tech would be better used in a laptop. That is possible but but the question really comes down to where do you need the extra RAM the most. I'd say iPads myself. It just seems like the perfect fit. It would likely be 2015 though before ramp up is secure enough for Apples demands.
In any event food for thought when it comes to the Mac Pro (needed to wake up the thread). I know I slipped off the path with the iPad stuff but that seems like a match made in heaven. On a Mac Pro this tech would allow for huge wads of memory soldered right onto the motherboard. In this respect Intel would have to be onboard implementing the interface into the CPU chips. I can't see Intel not being onboard though as I really don't know of a better short term solution.
Initial devices will be shipping in 2 & 4 GB offerings, not bad really. Now I haven't seen any indication that Intel or Apple are on board, but it would be hard to imagine they aren't. The thing here is what such technology could do for the likes of the Mac Pro as we dramatically increase storage density while lowering power. I could see Apple implementing this tech in a Mac Pro in a couple of years.
However it becomes even more interesting if you see this as almost the perfect technology to take iPads pass the 1GB barrier. I'm not about to discount the chips initial expense but think about what these memory cubes would allow Apple to do in an iPad. First you have a massive reduction in size. Second an estimated 70% reduction in power used per bit. Third amazing bandwidth.
On an iPad this would all work together to give us another stepped increase in performance. Apple could easily give us a 4X increase in available RAM, which Apps really need, while giving the GPU the bandwidth it needs. Bandwidth would be especially important on a larger iPad with considerably more pixels. It is the advantage to the GPU that I see as the big win here, in every APU design on the market the manufacture run up against the bandwidth problem trying to squeeze more performance out of the GPU.
Intel has gotten good results with Iris Plus using a multi chip module where the extra silicon is high speed memory. That isn't bad at all, however Memory Cube technology would offer that bandwidth to the entire memory array better support the GPU for traditional graphics and more importantly heterogenous computing.
Now some might argue that this tech would be better used in a laptop. That is possible but but the question really comes down to where do you need the extra RAM the most. I'd say iPads myself. It just seems like the perfect fit. It would likely be 2015 though before ramp up is secure enough for Apples demands.
In any event food for thought when it comes to the Mac Pro (needed to wake up the thread). I know I slipped off the path with the iPad stuff but that seems like a match made in heaven. On a Mac Pro this tech would allow for huge wads of memory soldered right onto the motherboard. In this respect Intel would have to be onboard implementing the interface into the CPU chips. I can't see Intel not being onboard though as I really don't know of a better short term solution.
This is for super computers, not desktops, laptops or tablets. Good try, but you obviously missed the target market for this. If it was designed for desktop/laptop/tablets, then they would have mentioned it.
This is for super computers, not desktops, laptops or tablets. Good try, but you obviously missed the target market for this. If it was designed for desktop/laptop/tablets, then they would have mentioned it.
When has Apple ever cared what the industry says a part is for?
This is for super computers, not desktops, laptops or tablets. Good try, but you obviously missed the target market for this. If it was designed for desktop/laptop/tablets, then they would have mentioned it.
I didn't mis anything, what is super computer hardware in 2014 will be mainstream in 2015. Beyond that if you would have spent two minutes researching the tech you would see that ARM is publicly on board. More so not all of the companies involved are even in the Super Computer business.
As for desktop/laptops/tablets you need new chips designed to support the new memory interface. This isn't a big deal for Apple.
By the way I don have any illusions about the technologies cost, it won't be cheap initially. However you have to balance that against the potential here to save space and provide the performance boost to support very high res displays.
The other way to put this is what other options does Apple have to put 4 or 8 GB of RAM into an iPad?
When has Apple ever cared what the industry says a part is for?
Just look at what Apple has done with flash and PCI Express. A few months ago all you would hear is that PCI based flash storage was for data centers, now Apple leads the industry in low cost machines with high performance storage.
It is more of an issue where Apple sees a solution to a problem the industry misses completely. The biggest problem for Apple, with respect to the Mac Pro, is getting Intel onboard with XEON chips that support this technology. This is one reason why the idea of such ram being used in the iPad popped into my mind. Apple can DIY the interface themselves in an A8 or A9 chip. But back to Intel, I can't imagine them not being involved with this tech on some level if they want to remain in the high performance game. It isn't just CPUs that might leverage this tech, discrete GPU cards might be in the cards too. Again density and low power levels would be big positives for GPU cards.
Sometimes I don't think people open up their minds to the possibilities that new tech offers.
I didn't mis anything, what is super computer hardware in 2014 will be mainstream in 2015. Beyond that if you would have spent two minutes researching the tech you would see that ARM is publicly on board. More so not all of the companies involved are even in the Super Computer business.
As for desktop/laptops/tablets you need new chips designed to support the new memory interface. This isn't a big deal for Apple.
By the way I don have any illusions about the technologies cost, it won't be cheap initially. However you have to balance that against the potential here to save space and provide the performance boost to support very high res displays.
The other way to put this is what other options does Apple have to put 4 or 8 GB of RAM into an iPad?
I don't think it's going to be that quick. If they do use this, it might be another 5 years and probably not on laptops and tablets. The form factor of the memory might be too thick. Remember, this is not the same packaging as traditional RAM, it's thicker. There wasn't enough information on the site you listed for me to determine if it's even going to work anytime in the near future, but Intel probably has to come out with a PCI spec that will handle this type of memory, that's what the PC industry is currently tied to.
If this technology is even used, it will be a LOT longer than 2 years from now and it'll probably work and be used in the higher end workstation class systems first.
They have to bring the costs down to the point where it makes sense, plus there's the technically feasible, power requirements, physical size, heat, etc.
I'm all for new technology, but I think you're being a little premature on this one. Sorry.
Of course. Remember that Apple put Ethernet on its consumer-level products when no one else did. That’s just one of many examples of how they’ve bucked the assumption of the market of a piece of hardware.
Of course. Remember that Apple put Ethernet on its consumer-level products when no one else did. That’s just one of many examples of how they’ve bucked the assumption of the market of a piece of hardware.
But ethernet was already accepted in the PC market first, they just put it on the motherboard first.
I just think it's too premature to even discuss this on Apple products yet.
Let's first see it working on a Supercomputer and see what the pricing, what bus they are using it with, etc.
Of course. Remember that Apple put Ethernet on its consumer-level products when no one else did. That’s just one of many examples of how they’ve bucked the assumption of the market of a piece of hardware.
Quote:
Originally Posted by wizard69
Just look at what Apple has done with flash and PCI Express. A few months ago all you would hear is that PCI based flash storage was for data centers, now Apple leads the industry in low cost machines with high performance storage.
It is more of an issue where Apple sees a solution to a problem the industry misses completely. The biggest problem for Apple, with respect to the Mac Pro, is getting Intel onboard with XEON chips that support this technology. This is one reason why the idea of such ram being used in the iPad popped into my mind. Apple can DIY the interface themselves in an A8 or A9 chip. But back to Intel, I can't imagine them not being involved with this tech on some level if they want to remain in the high performance game. It isn't just CPUs that might leverage this tech, discrete GPU cards might be in the cards too. Again density and low power levels would be big positives for GPU cards.
Sometimes I don't think people open up their minds to the possibilities that new tech offers.
But ethernet was already accepted in the PC market first…
In business PCs. Not consumer. That’s the argument here. Apple will take a technology, ignore what they’re being told it’s for, and use it wherever they see it benefiting people. I just wouldn’t underestimate them.
Of course this specific tech needs more work done to it. I actually don’t think it’s going to hit user-end for a very long time, if at all. There are a ton of competing RAM technologies out there. They’re working on one that only powers the RAM in use. Now that’s darn useful for the iPad, you know? Apple could shove 8GB RAM in one and not worry about battery life.
In business PCs. Not consumer. That’s the argument here. Apple will take a technology, ignore what they’re being told it’s for, and use it wherever they see it benefiting people. I just wouldn’t underestimate them.
Of course this specific tech needs more work done to it. I actually don’t think it’s going to hit user-end for a very long time, if at all. There are a ton of competing RAM technologies out there. They’re working on one that only powers the RAM in use. Now that’s darn useful for the iPad, you know? Apple could shove 8GB RAM in one and not worry about battery life.
Huh?
I just don't think we are going to see Hybrid Memory Cube on any Apple system by 2015. They are all playing around with various things right now.
I think Apple will implement DDR4 memory in maybe the 2015 or 2016 MacPro. Remember, they are currently using DD3 just like everyone else that makes Xeon based workstations.
According to this article, it doesn't sound like Hybrid memory cubes are going to replace DDR4 memory any time soon.
It isn't that simple. DDR4 is certainly the mainstream future but it really doesn't offer the density possibilities that are being talked about here. Then you have to consider the space and power trade offs. Right now they have two sizes to go to market with, 2 & 4 GB, with definitions for 8 GB and beyond. The question is can Apple save board space using these devices to deliver more RAM in devices like the iPad or even the MB AIRs. Maybe not in the 31 mm square device but there is basically a half sized solution. So can Apple put 4GB of DDR 4 RAM into a 20 mm square plot?
For Apples "A" series processors there is also a few design considerations that might impact the selection of memory cube technology. For one it might allow them to trim the size of some of the onboard caches or redirect those arrays to other uses. Fast RAM accessed in a heterogenous manner will allow enhanced GPU capabilities including very performant OpenCL support (They could also skip OpenCL completely in a true heterogeneous computing environment). A 4GB RAM array that is very fast no matter whom or how it is accessed would be very liberating in an iPad.
In any event you need to realize the article is written by a reporter who might have limited vision as far as the possibilities. For a manufacture ramping up a brand new technology they have to be careful not to over sell the technology to the public nor investors. It is better to deliver beyond projections than to mis them completely.
It isn't that simple. DDR4 is certainly the mainstream future but it really doesn't offer the density possibilities that are being talked about here. Then you have to consider the space and power trade offs. Right now they have two sizes to go to market with, 2 & 4 GB, with definitions for 8 GB and beyond. The question is can Apple save board space using these devices to deliver more RAM in devices like the iPad or even the MB AIRs. Maybe not in the 31 mm square device but there is basically a half sized solution. So can Apple put 4GB of DDR 4 RAM into a 20 mm square plot?
For Apples "A" series processors there is also a few design considerations that might impact the selection of memory cube technology. For one it might allow them to trim the size of some of the onboard caches or redirect those arrays to other uses. Fast RAM accessed in a heterogenous manner will allow enhanced GPU capabilities including very performant OpenCL support (They could also skip OpenCL completely in a true heterogeneous computing environment). A 4GB RAM array that is very fast no matter whom or how it is accessed would be very liberating in an iPad.
In any event you need to realize the article is written by a reporter who might have limited vision as far as the possibilities. For a manufacture ramping up a brand new technology they have to be careful not to over sell the technology to the public nor investors. It is better to deliver beyond projections than to mis them completely.
I think you are being overzealous over nothing.
Currently, Apple and all of the other PC mfg are using DDR3 and how much longer do you think they are going to use DDR3? Apple still hasn't started to use the highest speed, they are currently going to be using the 1866MHz DDR ECC memory, yet it can go up to 2133MHz.
DDR4 is supposed to use either TSV or standardized 3D stacking, but I don't think it's even finalized as of yet.
Intel's 2014 Haswell E roadmap suggests DDR4 SDRAM is going to be used.
You have to look at the processors and what they are meant to work with.
I just think 3D memory has potential packaging issues with things like iPads, etc. The actual package can't get too thick otherwise it'll make the device too thick, so Apple has to watch that.
What type of RAM is Apple currently using with their iPad Airs? DDR3.
Comments
I can see the Mac Pro becoming immensely more powerful by 2015.
Are you suggesting they might go from 2x GPUs to 1 GPU + Xeon Phi? I'm a little picky, because at this point I'm primarily tied to OSX for an XCode machine. In the past that wasn't the case.
Well you know Apple, it is hard to say what they will do. I'm imagining a Mac Pro with two GPUs as today with the new XEON Phi replacing the current XEON. GPUs still have significant advantages over anything Intel has to offer so I don't see the two GPUs going away. The interesting thing here is Intel making variants of Phi that can operate as system processors. Instead of struggling to go from 8-12 cores in a Pro, Apple could easily jump to 32, 64 or any number they can come to an agreement with Intel to build.
As far as XCode goes imagine a machine with 32 or 64 cores compiling your apps.
Of course these aren't run of the mill i86 cores. But the interesting thing here is if they can do this for Phi and do it for Haswell Iris they most certainly can do it for run of the mill XEON. In other words I think thing look bright for the Mac Pro.
I can see FKK beaches in Relics future! After all we don't want to be seeing nasty tan lines.
Well you know Apple, it is hard to say what they will do. I'm imagining a Mac Pro with two GPUs as today with the new XEON Phi replacing the current XEON. GPUs still have significant advantages over anything Intel has to offer so I don't see the two GPUs going away. The interesting thing here is Intel making variants of Phi that can operate as system processors. Instead of struggling to go from 8-12 cores in a Pro, Apple could easily jump to 32, 64 or any number they can come to an agreement with Intel to build.
As far as XCode goes imagine a machine with 32 or 64 cores compiling your apps.
Of course these aren't run of the mill i86 cores. But the interesting thing here is if they can do this for Phi and do it for Haswell Iris they most certainly can do it for run of the mill XEON. In other words I think thing look bright for the Mac Pro.
Yeah I have no idea about that. I haven't seen intel bring out anything Phi-wise that costs less than a few thousand. I will mention that chipsets will change again with Haswell E whenever that comes out, so there's nothing past this point until probably some point in 2015. Ivy Bridge E is just rolling out at the moment across oems.
I can see FKK beaches in Relics future! After all we don't want to be seeing nasty tan lines.
I request a rendition of Rondeau within that playlist, just for the cheesy masterpiece theatre reference. Anything by Einaudi would also be cool.
Music has never been a big part of my life, but I will listen to anything Relic wishes to play if it means getting well.
I suspect if Intel is to have success with Phi they will need a way to mass market it. Cost wise it isn't significantly worst than Xeon now. I think the key here is that this is an indication of the direction of technology. If they integrate RAM into the Haswell Iris packages and the Phi, it is possible for other chips focusing on performance. The trickle down theory of computer systems if you will.
Well it may not be much worse compared to the higher core chips, but it definitely covers different price territory unless I have missed recent developments. The Xeons that Apple has used over the past 4 years start at $300. I am curious how that thing will look in the wild, as right now the price barrier seems awfully high relative to what is included. I expected to see either a lower price than what it ended up as or the cost cutting measures (chopped down framebuffers, quad cpu) , not necessarily both.
Quote:
Music has never been a big part of my life, but I will listen to anything Relic wishes to play if it means getting well.
Bleh I like music, but I can't deal with anything lyrical while working. It's too much of a distraction.
Micron has been releasing more and more info on their "Hybrid Memory Cube" technology. Some info can be found here: http://www.micron.com/products/topics/supercomputing?source=MB and here: http://www.hybridmemorycube.org. Interestingly the spec has been finalized, so this is more than vapor ware now
Initial devices will be shipping in 2 & 4 GB offerings, not bad really. Now I haven't seen any indication that Intel or Apple are on board, but it would be hard to imagine they aren't. The thing here is what such technology could do for the likes of the Mac Pro as we dramatically increase storage density while lowering power. I could see Apple implementing this tech in a Mac Pro in a couple of years.
However it becomes even more interesting if you see this as almost the perfect technology to take iPads pass the 1GB barrier. I'm not about to discount the chips initial expense but think about what these memory cubes would allow Apple to do in an iPad. First you have a massive reduction in size. Second an estimated 70% reduction in power used per bit. Third amazing bandwidth.
On an iPad this would all work together to give us another stepped increase in performance. Apple could easily give us a 4X increase in available RAM, which Apps really need, while giving the GPU the bandwidth it needs. Bandwidth would be especially important on a larger iPad with considerably more pixels. It is the advantage to the GPU that I see as the big win here, in every APU design on the market the manufacture run up against the bandwidth problem trying to squeeze more performance out of the GPU.
Intel has gotten good results with Iris Plus using a multi chip module where the extra silicon is high speed memory. That isn't bad at all, however Memory Cube technology would offer that bandwidth to the entire memory array better support the GPU for traditional graphics and more importantly heterogenous computing.
Now some might argue that this tech would be better used in a laptop. That is possible but but the question really comes down to where do you need the extra RAM the most. I'd say iPads myself. It just seems like the perfect fit. It would likely be 2015 though before ramp up is secure enough for Apples demands.
In any event food for thought when it comes to the Mac Pro (needed to wake up the thread). I know I slipped off the path with the iPad stuff but that seems like a match made in heaven. On a Mac Pro this tech would allow for huge wads of memory soldered right onto the motherboard. In this respect Intel would have to be onboard implementing the interface into the CPU chips. I can't see Intel not being onboard though as I really don't know of a better short term solution.
Near Future Apple Technology
Micron has been releasing more and more info on their "Hybrid Memory Cube" technology. Some info can be found here: http://www.micron.com/products/topics/supercomputing?source=MB and here: http://www.hybridmemorycube.org. Interestingly the spec has been finalized, so this is more than vapor ware now
Initial devices will be shipping in 2 & 4 GB offerings, not bad really. Now I haven't seen any indication that Intel or Apple are on board, but it would be hard to imagine they aren't. The thing here is what such technology could do for the likes of the Mac Pro as we dramatically increase storage density while lowering power. I could see Apple implementing this tech in a Mac Pro in a couple of years.
However it becomes even more interesting if you see this as almost the perfect technology to take iPads pass the 1GB barrier. I'm not about to discount the chips initial expense but think about what these memory cubes would allow Apple to do in an iPad. First you have a massive reduction in size. Second an estimated 70% reduction in power used per bit. Third amazing bandwidth.
On an iPad this would all work together to give us another stepped increase in performance. Apple could easily give us a 4X increase in available RAM, which Apps really need, while giving the GPU the bandwidth it needs. Bandwidth would be especially important on a larger iPad with considerably more pixels. It is the advantage to the GPU that I see as the big win here, in every APU design on the market the manufacture run up against the bandwidth problem trying to squeeze more performance out of the GPU.
Intel has gotten good results with Iris Plus using a multi chip module where the extra silicon is high speed memory. That isn't bad at all, however Memory Cube technology would offer that bandwidth to the entire memory array better support the GPU for traditional graphics and more importantly heterogenous computing.
Now some might argue that this tech would be better used in a laptop. That is possible but but the question really comes down to where do you need the extra RAM the most. I'd say iPads myself. It just seems like the perfect fit. It would likely be 2015 though before ramp up is secure enough for Apples demands.
In any event food for thought when it comes to the Mac Pro (needed to wake up the thread). I know I slipped off the path with the iPad stuff but that seems like a match made in heaven. On a Mac Pro this tech would allow for huge wads of memory soldered right onto the motherboard. In this respect Intel would have to be onboard implementing the interface into the CPU chips. I can't see Intel not being onboard though as I really don't know of a better short term solution.
This is for super computers, not desktops, laptops or tablets. Good try, but you obviously missed the target market for this. If it was designed for desktop/laptop/tablets, then they would have mentioned it.
When has Apple ever cared what the industry says a part is for?
When has Apple ever cared what the industry says a part is for
TS, really?
I didn't mis anything, what is super computer hardware in 2014 will be mainstream in 2015. Beyond that if you would have spent two minutes researching the tech you would see that ARM is publicly on board. More so not all of the companies involved are even in the Super Computer business.
As for desktop/laptops/tablets you need new chips designed to support the new memory interface. This isn't a big deal for Apple.
By the way I don have any illusions about the technologies cost, it won't be cheap initially. However you have to balance that against the potential here to save space and provide the performance boost to support very high res displays.
The other way to put this is what other options does Apple have to put 4 or 8 GB of RAM into an iPad?
Just look at what Apple has done with flash and PCI Express. A few months ago all you would hear is that PCI based flash storage was for data centers, now Apple leads the industry in low cost machines with high performance storage.
It is more of an issue where Apple sees a solution to a problem the industry misses completely. The biggest problem for Apple, with respect to the Mac Pro, is getting Intel onboard with XEON chips that support this technology. This is one reason why the idea of such ram being used in the iPad popped into my mind. Apple can DIY the interface themselves in an A8 or A9 chip. But back to Intel, I can't imagine them not being involved with this tech on some level if they want to remain in the high performance game. It isn't just CPUs that might leverage this tech, discrete GPU cards might be in the cards too. Again density and low power levels would be big positives for GPU cards.
Sometimes I don't think people open up their minds to the possibilities that new tech offers.
I didn't mis anything, what is super computer hardware in 2014 will be mainstream in 2015. Beyond that if you would have spent two minutes researching the tech you would see that ARM is publicly on board. More so not all of the companies involved are even in the Super Computer business.
As for desktop/laptops/tablets you need new chips designed to support the new memory interface. This isn't a big deal for Apple.
By the way I don have any illusions about the technologies cost, it won't be cheap initially. However you have to balance that against the potential here to save space and provide the performance boost to support very high res displays.
The other way to put this is what other options does Apple have to put 4 or 8 GB of RAM into an iPad?
I don't think it's going to be that quick. If they do use this, it might be another 5 years and probably not on laptops and tablets. The form factor of the memory might be too thick. Remember, this is not the same packaging as traditional RAM, it's thicker. There wasn't enough information on the site you listed for me to determine if it's even going to work anytime in the near future, but Intel probably has to come out with a PCI spec that will handle this type of memory, that's what the PC industry is currently tied to.
If this technology is even used, it will be a LOT longer than 2 years from now and it'll probably work and be used in the higher end workstation class systems first.
They have to bring the costs down to the point where it makes sense, plus there's the technically feasible, power requirements, physical size, heat, etc.
I'm all for new technology, but I think you're being a little premature on this one. Sorry.
Of course. Remember that Apple put Ethernet on its consumer-level products when no one else did. That’s just one of many examples of how they’ve bucked the assumption of the market of a piece of hardware.
Of course. Remember that Apple put Ethernet on its consumer-level products when no one else did. That’s just one of many examples of how they’ve bucked the assumption of the market of a piece of hardware.
But ethernet was already accepted in the PC market first, they just put it on the motherboard first.
I just think it's too premature to even discuss this on Apple products yet.
Let's first see it working on a Supercomputer and see what the pricing, what bus they are using it with, etc.
Of course. Remember that Apple put Ethernet on its consumer-level products when no one else did. That’s just one of many examples of how they’ve bucked the assumption of the market of a piece of hardware.
Just look at what Apple has done with flash and PCI Express. A few months ago all you would hear is that PCI based flash storage was for data centers, now Apple leads the industry in low cost machines with high performance storage.
It is more of an issue where Apple sees a solution to a problem the industry misses completely. The biggest problem for Apple, with respect to the Mac Pro, is getting Intel onboard with XEON chips that support this technology. This is one reason why the idea of such ram being used in the iPad popped into my mind. Apple can DIY the interface themselves in an A8 or A9 chip. But back to Intel, I can't imagine them not being involved with this tech on some level if they want to remain in the high performance game. It isn't just CPUs that might leverage this tech, discrete GPU cards might be in the cards too. Again density and low power levels would be big positives for GPU cards.
Sometimes I don't think people open up their minds to the possibilities that new tech offers.
Read this article.
http://www.extremetech.com/computing/167368-hybrid-memory-cube-160gbsec-ram-starts-shipping-is-this-the-technology-that-finally-kills-ddr-ram
According to this article, it doesn't sound like Hybrid memory cubes are going to replace DDR4 memory any time soon.
But ethernet was already accepted in the PC market first…
In business PCs. Not consumer. That’s the argument here. Apple will take a technology, ignore what they’re being told it’s for, and use it wherever they see it benefiting people. I just wouldn’t underestimate them.
Of course this specific tech needs more work done to it. I actually don’t think it’s going to hit user-end for a very long time, if at all. There are a ton of competing RAM technologies out there. They’re working on one that only powers the RAM in use. Now that’s darn useful for the iPad, you know? Apple could shove 8GB RAM in one and not worry about battery life.
In business PCs. Not consumer. That’s the argument here. Apple will take a technology, ignore what they’re being told it’s for, and use it wherever they see it benefiting people. I just wouldn’t underestimate them.
Of course this specific tech needs more work done to it. I actually don’t think it’s going to hit user-end for a very long time, if at all. There are a ton of competing RAM technologies out there. They’re working on one that only powers the RAM in use. Now that’s darn useful for the iPad, you know? Apple could shove 8GB RAM in one and not worry about battery life.
Huh?
I just don't think we are going to see Hybrid Memory Cube on any Apple system by 2015. They are all playing around with various things right now.
I think Apple will implement DDR4 memory in maybe the 2015 or 2016 MacPro. Remember, they are currently using DD3 just like everyone else that makes Xeon based workstations.
It isn't that simple. DDR4 is certainly the mainstream future but it really doesn't offer the density possibilities that are being talked about here. Then you have to consider the space and power trade offs. Right now they have two sizes to go to market with, 2 & 4 GB, with definitions for 8 GB and beyond. The question is can Apple save board space using these devices to deliver more RAM in devices like the iPad or even the MB AIRs. Maybe not in the 31 mm square device but there is basically a half sized solution. So can Apple put 4GB of DDR 4 RAM into a 20 mm square plot?
For Apples "A" series processors there is also a few design considerations that might impact the selection of memory cube technology. For one it might allow them to trim the size of some of the onboard caches or redirect those arrays to other uses. Fast RAM accessed in a heterogenous manner will allow enhanced GPU capabilities including very performant OpenCL support (They could also skip OpenCL completely in a true heterogeneous computing environment). A 4GB RAM array that is very fast no matter whom or how it is accessed would be very liberating in an iPad.
In any event you need to realize the article is written by a reporter who might have limited vision as far as the possibilities. For a manufacture ramping up a brand new technology they have to be careful not to over sell the technology to the public nor investors. It is better to deliver beyond projections than to mis them completely.
It isn't that simple. DDR4 is certainly the mainstream future but it really doesn't offer the density possibilities that are being talked about here. Then you have to consider the space and power trade offs. Right now they have two sizes to go to market with, 2 & 4 GB, with definitions for 8 GB and beyond. The question is can Apple save board space using these devices to deliver more RAM in devices like the iPad or even the MB AIRs. Maybe not in the 31 mm square device but there is basically a half sized solution. So can Apple put 4GB of DDR 4 RAM into a 20 mm square plot?
For Apples "A" series processors there is also a few design considerations that might impact the selection of memory cube technology. For one it might allow them to trim the size of some of the onboard caches or redirect those arrays to other uses. Fast RAM accessed in a heterogenous manner will allow enhanced GPU capabilities including very performant OpenCL support (They could also skip OpenCL completely in a true heterogeneous computing environment). A 4GB RAM array that is very fast no matter whom or how it is accessed would be very liberating in an iPad.
In any event you need to realize the article is written by a reporter who might have limited vision as far as the possibilities. For a manufacture ramping up a brand new technology they have to be careful not to over sell the technology to the public nor investors. It is better to deliver beyond projections than to mis them completely.
I think you are being overzealous over nothing.
Currently, Apple and all of the other PC mfg are using DDR3 and how much longer do you think they are going to use DDR3? Apple still hasn't started to use the highest speed, they are currently going to be using the 1866MHz DDR ECC memory, yet it can go up to 2133MHz.
DDR4 is supposed to use either TSV or standardized 3D stacking, but I don't think it's even finalized as of yet.
Intel's 2014 Haswell E roadmap suggests DDR4 SDRAM is going to be used.
You have to look at the processors and what they are meant to work with.
I just think 3D memory has potential packaging issues with things like iPads, etc. The actual package can't get too thick otherwise it'll make the device too thick, so Apple has to watch that.
What type of RAM is Apple currently using with their iPad Airs? DDR3.