I'd think that the most likely one is an Xscale chip, which is an ARM variant. They are very low power and ARM chips have been in iPods from the first model.
I would be somewhat surprised if it was something else.
seems to indicate that we indeed are talking about a ARM-based product (makes most sense as we are talking about a mobile device) and some kind of Mac OS X subset (not only by name).
There's also an article I read that says the iPhone uses one of the new Nvidia mobile chips, which basically includes everything except the actual CPU. That would make sense. The Nvidia chips are nice, and Apple may be using them if they were able to toss on a late-model ARM core for Apple.
From my point of view, the PXA270 and 290 are not well suited for the iPhone. I am not bullish on the XScale's future in this sort of device, or, for that matter, in any device. The XScale is too power-hungry to considered for true low-power devices, it doesn't have enough peripherals on-chip, and most markets applicable to the Xscale PXA270 and 290 are better off using low power PPCs or Blackfins. If Marvell wants to make money out of XScale, they should think about delivering a 65nm 255 core with oodles of on-chip peripherals. That would be something. . . .
Anyway, if I were designing the iPhone, I'd first look into using the TI OMAP 34xx. It has everything in a single package, is 65nm, packs a lot of computational power, and is very low power.
Also, ARMs are 32bit machines with little endian support and GREAT support built right in to gcc. There's absolutely no reason why Apple couldn't just compile ARM versions of their software, which are probably stripped-down variants anyway.
Don't worry Melgross, Jalkut's good people: his blog was something of a public self help group back during the Rev.A MacBook Pro hardware problems saga. I still subscribe to the RSS as he's sometimes quite interesting in his codery views. On this he's just voicing his opinion as Joe Public, like we are too.
That's fine if he came here as we do, and voiced it. But when you have a blog, you are telling the world that you have something important to say, and we should listen. The etiquette of posting a link, is to post information we don't have access to in our discussion. This was anything but that.
I'm wondering if he has two names, and posted the link to himself. Two first time posts!
There's also an article I read that says the iPhone uses one of the new Nvidia mobile chips, which basically includes everything except the actual CPU. That would make sense. The Nvidia chips are nice, and Apple may be using them if they were able to toss on a late-model ARM core for Apple.
From my point of view, the PXA270 and 290 are not well suited for the iPhone. I am not bullish on the XScale's future in this sort of device, or, for that matter, in any device. The XScale is too power-hungry to considered for true low-power devices, it doesn't have enough peripherals on-chip, and most markets applicable to the Xscale PXA270 and 290 are better off using low power PPCs or Blackfins. If Marvell wants to make money out of XScale, they should think about delivering a 65nm 255 core with oodles of on-chip peripherals. That would be something. . . .
Anyway, if I were designing the iPhone, I'd first look into using the TI OMAP 34xx. It has everything in a single package, is 65nm, packs a lot of computational power, and is very low power.
Also, ARMs are 32bit machines with little endian support and GREAT support built right in to gcc. There's absolutely no reason why Apple couldn't just compile ARM versions of their software, which are probably stripped-down variants anyway.
I would imagine that it would be best. If the OS is not too big, debugging, and optimising, shouldn't be too difficult (relatively speaking, of course.)
Hopefully, we will see third party software, and they won't have problems.
But, Jobs said to think of this as a phone, not a computer, so we shall see.
Comments
I'd think that the most likely one is an Xscale chip, which is an ARM variant. They are very low power and ARM chips have been in iPods from the first model.
I would be somewhat surprised if it was something else.
This job ad:
http://jobs.apple.com/index.ajs?BID=...&CurrentPage=1
seems to indicate that we indeed are talking about a ARM-based product (makes most sense as we are talking about a mobile device) and some kind of Mac OS X subset (not only by name).
/Jesper
From my point of view, the PXA270 and 290 are not well suited for the iPhone. I am not bullish on the XScale's future in this sort of device, or, for that matter, in any device. The XScale is too power-hungry to considered for true low-power devices, it doesn't have enough peripherals on-chip, and most markets applicable to the Xscale PXA270 and 290 are better off using low power PPCs or Blackfins. If Marvell wants to make money out of XScale, they should think about delivering a 65nm 255 core with oodles of on-chip peripherals. That would be something. . . .
Anyway, if I were designing the iPhone, I'd first look into using the TI OMAP 34xx. It has everything in a single package, is 65nm, packs a lot of computational power, and is very low power.
Also, ARMs are 32bit machines with little endian support and GREAT support built right in to gcc. There's absolutely no reason why Apple couldn't just compile ARM versions of their software, which are probably stripped-down variants anyway.
Don't worry Melgross, Jalkut's good people: his blog was something of a public self help group back during the Rev.A MacBook Pro hardware problems saga. I still subscribe to the RSS as he's sometimes quite interesting in his codery views. On this he's just voicing his opinion as Joe Public, like we are too.
That's fine if he came here as we do, and voiced it. But when you have a blog, you are telling the world that you have something important to say, and we should listen. The etiquette of posting a link, is to post information we don't have access to in our discussion. This was anything but that.
I'm wondering if he has two names, and posted the link to himself. Two first time posts!
There's also an article I read that says the iPhone uses one of the new Nvidia mobile chips, which basically includes everything except the actual CPU. That would make sense. The Nvidia chips are nice, and Apple may be using them if they were able to toss on a late-model ARM core for Apple.
From my point of view, the PXA270 and 290 are not well suited for the iPhone. I am not bullish on the XScale's future in this sort of device, or, for that matter, in any device. The XScale is too power-hungry to considered for true low-power devices, it doesn't have enough peripherals on-chip, and most markets applicable to the Xscale PXA270 and 290 are better off using low power PPCs or Blackfins. If Marvell wants to make money out of XScale, they should think about delivering a 65nm 255 core with oodles of on-chip peripherals. That would be something. . . .
Anyway, if I were designing the iPhone, I'd first look into using the TI OMAP 34xx. It has everything in a single package, is 65nm, packs a lot of computational power, and is very low power.
Also, ARMs are 32bit machines with little endian support and GREAT support built right in to gcc. There's absolutely no reason why Apple couldn't just compile ARM versions of their software, which are probably stripped-down variants anyway.
I would imagine that it would be best. If the OS is not too big, debugging, and optimising, shouldn't be too difficult (relatively speaking, of course.)
Hopefully, we will see third party software, and they won't have problems.
But, Jobs said to think of this as a phone, not a computer, so we shall see.