I realized that the 7448 is essentially the "missing link" between the 744x series and the dual-core CPU coming down the pipe.
That is an interesting perspective as I was under the impression that 7448 was more of a process shrink. It would be nice if it picked up Book-E enhancements and in effect becomes the single core cell for their custom chips.
Quote:
As for RapidIO, wizard69 is basically right: It's not a matter of how hard it is to implement, it's a matter of whether Apple wants to commit the resources to implementing the protocol. Remember, RapidIO isn't up against HyperTransport in this particular case, it's up against Elastic Bus.
Considering some of the new chips advertised on Freescales web site I'm of the opinion that RapidI/O is more like Hypertransport than E-Bus. At least from the perspective that on the chips that support RapidI/O Freescale also has implemented and integrated memory controller. The processors remind you of AMD hardware though the currently available devices are high integration embedded chips.
Quote:
If anything, that would be more work and more complexity (and likely more $$$$) for Motorola than HT, and even more of a loss politically for RapidIO. If the rumors around the Moto G5 are correct, Moto's already refused to adopt E-Bus at least once.
The big thing with Hypertransport is that there is huge support with respect to aux. chips. For Apple to go the way of RapidI/O they would most likely need to build custom chips. It would be sounder from the engineering standpoint to simply have Freescale tack on a custom bus interface to the new processor.
Freescale is actively advertising this sort of custom chip capability and Apple has a huge investment already in Hypertransport so I see this as a real possibility. The bigger question in my mind is how long will 32 bit be around and will such an investment in hardware be practical.
Quote:
On the other hand, if Apple imagines a future for FreeScale CPUs in their portables, and in similarly constrained devices, they might find it worthwhile to build RapidIO into their controller. Really, it depends on where Apple's going more than anything.
I'm of the opinion that building Hypertransport into the processor is a better path for Apple. How doable this is is not entirely clear from Freescales web site. The problem is that Apple doesn't need another bus to support.
Quote:
Or, for at least one more iteration, they'll stick with MaxBus...
Yeah I can see Apple, with their apparent departure from innovation, going this route. How usefull of an upgrade this would be is very dependant on just what is the 7448. Personally I'd rather see them get something new on the market if it can attack battery life and performance issues.
Fat Freddy: IBM is 0.2v away from having a mobile chip, and Freescale knows it.
Quote:
Originally posted by wizard69
That is an interesting perspective as I was under the impression that 7448 was more of a process shrink. It would be nice if it picked up Book-E enhancements and in effect becomes the single core cell for their custom chips.
I was as well, but if you look at the chip from that perspective, the Register's claim makes sense.
Of course, the source of the claim has to be considered here.
Quote:
Considering some of the new chips advertised on Freescales web site I'm of the opinion that RapidI/O is more like Hypertransport than E-Bus.
True. But from Apple's perspective that doesn't matter, because what RapidIO would be up against on an Apple motherboard would be Elastic Bus. IBM doesn't use HyperTransport as a CPU bus, and it doesn't look like they will any time soon.
So from Apple's perspective, they'd have to support two significantly different bus technologies. This would also be true if Freescale used HyperTransport instead of RapidIO. What Apple would be pushing for would be Elastic Bus, which is trickier than either of the other two busses, and (apparently) owned by IBM.
Quote:
Yeah I can see Apple, with their apparent departure from innovation, going this route. How usefull of an upgrade this would be is very dependant on just what is the 7448. Personally I'd rather see them get something new on the market if it can attack battery life and performance issues.
There's no reason to be this gloomy. The only reason to stick with MaxBus for one more iteration would be to roll out a bump to the line in the fall with minimal work, so that the line doesn't stagnate while Apple works on a significant revision. It wouldn't be a signal of defeat, just a little bump to the line for the holidays that Apple can roll out if they won't have the Next Big Thing™ ready for some months yet.
Don't the actual chips have to be released first? That's the only reason why I waited for my 3rd-generation (Rev. C is a misnomer, IMO) -- knowing full well that the lifecycle updates was in sync with Motorola's release of the 7447A.
How does freescale taking over moto's ppc department effect the Moto, IBM, Apple treaty?
AIM is dead, long since.
The current arrangements between Apple, IBM, and Motorola/Freescale, such as they are, are built around specific products and applications, rather than describing a general strategy and joint R&D. It's not significantly different than Apple partnering with Hitachi, IBM partnering with NVIDIA, Motorola partnering with AMD, etc.
Comments
Originally posted by Amorph
I realized that the 7448 is essentially the "missing link" between the 744x series and the dual-core CPU coming down the pipe.
That is an interesting perspective as I was under the impression that 7448 was more of a process shrink. It would be nice if it picked up Book-E enhancements and in effect becomes the single core cell for their custom chips.
As for RapidIO, wizard69 is basically right: It's not a matter of how hard it is to implement, it's a matter of whether Apple wants to commit the resources to implementing the protocol. Remember, RapidIO isn't up against HyperTransport in this particular case, it's up against Elastic Bus.
Considering some of the new chips advertised on Freescales web site I'm of the opinion that RapidI/O is more like Hypertransport than E-Bus. At least from the perspective that on the chips that support RapidI/O Freescale also has implemented and integrated memory controller. The processors remind you of AMD hardware though the currently available devices are high integration embedded chips.
If anything, that would be more work and more complexity (and likely more $$$$) for Motorola than HT, and even more of a loss politically for RapidIO. If the rumors around the Moto G5 are correct, Moto's already refused to adopt E-Bus at least once.
The big thing with Hypertransport is that there is huge support with respect to aux. chips. For Apple to go the way of RapidI/O they would most likely need to build custom chips. It would be sounder from the engineering standpoint to simply have Freescale tack on a custom bus interface to the new processor.
Freescale is actively advertising this sort of custom chip capability and Apple has a huge investment already in Hypertransport so I see this as a real possibility. The bigger question in my mind is how long will 32 bit be around and will such an investment in hardware be practical.
On the other hand, if Apple imagines a future for FreeScale CPUs in their portables, and in similarly constrained devices, they might find it worthwhile to build RapidIO into their controller. Really, it depends on where Apple's going more than anything.
I'm of the opinion that building Hypertransport into the processor is a better path for Apple. How doable this is is not entirely clear from Freescales web site. The problem is that Apple doesn't need another bus to support.
Or, for at least one more iteration, they'll stick with MaxBus...
Yeah I can see Apple, with their apparent departure from innovation, going this route. How usefull of an upgrade this would be is very dependant on just what is the 7448. Personally I'd rather see them get something new on the market if it can attack battery life and performance issues.
Dave
Originally posted by wizard69
That is an interesting perspective as I was under the impression that 7448 was more of a process shrink. It would be nice if it picked up Book-E enhancements and in effect becomes the single core cell for their custom chips.
I was as well, but if you look at the chip from that perspective, the Register's claim makes sense.
Of course, the source of the claim has to be considered here.
Considering some of the new chips advertised on Freescales web site I'm of the opinion that RapidI/O is more like Hypertransport than E-Bus.
True. But from Apple's perspective that doesn't matter, because what RapidIO would be up against on an Apple motherboard would be Elastic Bus. IBM doesn't use HyperTransport as a CPU bus, and it doesn't look like they will any time soon.
So from Apple's perspective, they'd have to support two significantly different bus technologies. This would also be true if Freescale used HyperTransport instead of RapidIO. What Apple would be pushing for would be Elastic Bus, which is trickier than either of the other two busses, and (apparently) owned by IBM.
Yeah I can see Apple, with their apparent departure from innovation, going this route. How usefull of an upgrade this would be is very dependant on just what is the 7448. Personally I'd rather see them get something new on the market if it can attack battery life and performance issues.
There's no reason to be this gloomy. The only reason to stick with MaxBus for one more iteration would be to roll out a bump to the line in the fall with minimal work, so that the line doesn't stagnate while Apple works on a significant revision. It wouldn't be a signal of defeat, just a little bump to the line for the holidays that Apple can roll out if they won't have the Next Big Thing™ ready for some months yet.
How does freescale taking over moto's ppc department effect the Moto, IBM, Apple treaty?
Originally posted by dennis88
...
The new pb's will look like this:
12" pb: 1.6 ghz G4, a Nvidia low/mid-end card 64 mb, 256mb ddr 333mhz.
15" pb: 1.8 ghz G4, Ati 9800 mobility 128 mb, 512 mb ddr 400 mhz.
17" pb: 1.8/1.93 ghz G4, Ati 9800 mobility 128 mb, 512 mb ddr 400 mhz.
All with 1024k chache.
...
Originally posted by emig647
Just out of curiousity....
How does freescale taking over moto's ppc department effect the Moto, IBM, Apple treaty?
AIM is dead, long since.
The current arrangements between Apple, IBM, and Motorola/Freescale, such as they are, are built around specific products and applications, rather than describing a general strategy and joint R&D. It's not significantly different than Apple partnering with Hitachi, IBM partnering with NVIDIA, Motorola partnering with AMD, etc.