Via will live as long as Formosa Plastics Group wants to keep it alive.
http://en.wikipedia.org/wiki/Formosa_Plastics_Group
http://english.cw.com.tw/article.do?action=show&id=3450
The difference is that the metal notebooks are going to be based on mobile Nehalems, while the plastic ones will stick with Penryn Core Number Numeral CPUs. This may seem counter-intuitive, but the reason is GPUs. On the aluminum models, there is room for an external GPU, so that is exactly what you get. The plastic ones don’t have room for a separate GPU, so they are forced to stick with Penryn and the bug infested 9400.
All the information available indicates that Ivy Bridge would wipe the floor with a "performance" ARM-based processor for the foreseeable future, and we have a hard time believing users would willingly downgrade from an Ivy Bridge-based laptop to an ARM-based laptop all other things being equal.
They wanted to ditch x86 and go with Itanium at the time so they would have an architecture they wouldn't have to license to AMD.
There is an x86 cross licensing agreement between the two companies. Nobody goes begging.
Here's the odds of AMD building a CPU containing both the x86 and ARM instruction sets:
If it happens in the next several years I will buy you one
Intel would love to ditch x86. . . for their own proprietary ISA that they have complete control over. Not for ARM.
The ONLY way Intel would ever transition to ARM is if they have no choice. They would much rather stay in the x86 market, where their profit margins are MUCH higher and their only competition is AMD. Why would Intel ever want to switch to selling cheap ARM processors and competing against several other ARM licensees?
It's not impossible that Intel could get back into the ARM market, but they won't go willingly. It will only happen if ARM starts seriously eating Intel's lunch in mobile, PCs, and servers. And even then, they would still keep x86 around for legacy and niche applications, much in the way they keep making Itanium CPUs.
I think a lot of people are forgetting a lot of things. The best current ARM CPU's are on 45nm. They have 2 cores, and a die size of 100-120mm^2. AMD got its C-50 on 40nm to be about 75mm^2. Important thing though is that the A5 sips power, some 80Mw of power. Granted that can go higher, but AMD can barely get anything below 5w! Intel does even worse as far as can be seen from the likes of Atom.
is there a reason why intel can't do their own ARM socs on this new node or the next for phones/tablets? i mean no one elses SOC could come close in power usage could it? everyone else is on 45nm right?
they could make a mint and have the fab capacity to back it up.
We're all backwards hicks, what can I say? Ultimately, I don't know what AT&T is going to do. Everyone knows Verizon is the only network worth anything, and most people I know on AT&T, like myself, inherited AT&T, and will move away as soon as there is a reason to get into a contract (travel, home w/ bad service, phone that only Verizon offers while we're still network-locked, etc.).Well you know I've been in South Korea, Japan, Europe and the US in the last time and apart from the US with its atrocious networks (although as long as AT&T stays out you may still stand a chance!) I don't see much problems from that direction.
There are cell networks bringing tens to hundreds of MB/s (big B) out to the boonies, per cell user (maybe they'll only use 1MB/s right now, but when they fire up a 3D game, and you have to send all the video over the network...), in most of the world? Most places can't even do that wired, yet.You guys though? Goodness you should really hope it doesn't take off - but just looking at the US is a bit narrow for such trends.
is there a reason why intel can't do their own ARM socs on this new node or the next for phones/tablets? i mean no one elses SOC could come close in power usage could it? everyone else is on 45nm right?
they could make a mint and have the fab capacity to back it up.
Sure the A5 sips power, but it sure had better considering its comparatively anemic performance. As well, I'm pretty sure that even it ends up at more than just 80mW, considering that that figure matches up with ARMs stated dynamic power consumption for that design. Either way, the A8 definitely requires more power, and then once you get up to a dual core A9 they state 0.5W for the 800 MHz 'power optimized' core design and 1.9W for the 2GHz 'performance optimized' core design. Remember, those figures are -only- for the actual processor core, which doesn't include L2 cache, graphics implementation, memory interface, or any other I/O necessary.
As for Atom... it's nowhere near so bad as many love making it out to be. Though, of course, it could be so much better if Intel had actually put resources into improving it instead of just letting it linger pretty much unchanged since introduction. Part of the impression of poor power consumption is certainly due to the fact that its initial implementation used a horrible chipset, while the other part is thanks to the simple fact that the N and D varieties used by netbooks/nettops are 'low cost' products, not low power. I mean, they start at 6.5W and go up to 13W just for the processor with another 2.1W for the NM10 I/O hub. Meanwhile the actual low power implementation, the Z series, is at 3W or less for the processor and then 0.75W for the SM35 - that's all just on Intel's 45nm process too. I really won't find just a simple process shrink of the current design that interesting though... rather I'm anxious to see what the atom core redesign will end up looking like, not to mention when it'll arrive. It's clear that they can get down to ARM active power levels, especially if they'd leverage their process technology advantage, but compared to the A9 I'd bet that performance is merely comparable, and ARM almost surely wins at idle power.
wouldn't intel build arm SOCs with a 50% margin on them still be competitive, if not cheaper than the competition? i mean if they did this and it came down to a price war, intel could cut prices more easily than the rest due to the fab advantage right? same as they can do currently against AMD.
that's ignoring that intels chips could be faster as well as use less power than the competitions socs (assuming intel does a good job which i assume they would since they seem to be damned good in this area).
wouldn't intel build arm SOCs with a 50% margin on them still be competitive, if not cheaper than the competition? i mean if they did this and it came down to a price war, intel could cut prices more easily than the rest due to the fab advantage right? same as they can do currently against AMD.
that's ignoring that intels chips could be faster as well as use less power than the competitions socs (assuming intel does a good job which i assume they would since they seem to be damned good in this area).
This site is not loading. Didn't yesterday either. What's up?
Nvidia has an angle with their superior GPU IP, they can bootstrap their GPU to an existing ARM ISA and have a product that can be easily/readily differentiated and that differentiation can be leveraged into a gross-margin premium. Maybe.
Well, if they are projected to be that high, for one i dont believe they are, the A15 is supposed to use even less than the A5. Secondly they couldnt possibly be used in phones if their power was that high since that alone excedes the battery, and that would be horribly dumb. Id like to see where you are getting this info. If they wouldnt be used in phones, then they would obviously be faster, but seeing as how there are no numbers for that, I cant say on that one.
2. Yes I know that too, but before athalon64 released amd didn't have sse3, intel offered it for amd x64 license.
"delivers its peak performance of 4000 DMIPS while consuming less than 250mW per CPU when selected from typical silicon."Numbers are courtesy of ARM's product page for the A9, under the 'performance' tab - http://www.arm.com/products/processors/cortex-a/cortex-a9.php It's also quite important to realize the simple fact that going from the power optimized to performance optimized results in 2.5x the performance for almost 4x the power.
I do find it interesting that the A9 is the only design where they're willing to give actual core power figures. But those numbers match up quite nicely with the simple fact that there's nothing magical about ARM's design, despite the fact that many sure want to believe that such is the case. Once they start offering comparable performance to x86, their power usage goes up accordingly. Heh, and battery life tests of phones using the Tegra 2 actually match up with the 'power optimized' design figures pretty well - specifically, the LG optimus 2x with it's 5.6Whr battery manages 3.85 hours on the Anandtech 'gaming' battery life test. (It should be noted that the AP20H version of tegra 2 uses the VFPv3-D16 FPU instead of the SIMD NEON 'FPU', and hence should be a bit lower power than the default A9 ARM hard macros.)