watchOS 3 looks to be a major improvement, and a faster second generation model would be a good move (hopefully it'll be A7 derived).
People aren't playing Infinity blade or editing video on their watch, and they probably never will, so A7 would be massive overkill and a waste of die space and battery power. I'd rather have them spend that transistor budget on LTE.
At this point, a 32 bit processor in the performance range of the A4 would be *plenty* for the Watch. Also maybe a bit more RAM and/or 3d XPoint and/or faster flash to help with speed in launching/switching apps.
The current model is A5 derived and is widely criticized for being too slow, and you want them to go even slower?
The word coming out of WWDC is that the next rev of WatchIS will be much faster! That is good news, but that doesn't mean that faster hardware isn't needed. By the time the next rev ships Apple could have 3-4 process node shrinks or steps to work with. That should give them plenty of room to lower the power profile while giving themselves a bit better performance margin. I wouldn't be surprised if they have the ability to lower the power profile to 10% of the current value through shrinks, higher integration and other tweaks.
I really hope Apple will offer upgrades for existing users, at least to Watch and Edition users (maybe not Sport). The upgrade path I have in mind would be -- take your watch into an Apple Store and have the old guts swapped out and replaced by the new guts (I'm not suggesting that users can or should do this themselves). I imagine the cost of the upgrade would be about the same as the cost of buying a new Sport.
Why do this? Four related points:
1. If they don't, Edition is dead. 2. Many people view watches as keepsakes. Apple Watch cannot compete in that market if people are throwing them out and/or recycling them every two years. 3. It reinforces the view of Apple employees as craftsmen, and Apple products as valuable things, not disposable things 4. It can be pitched as environmentally friendly -- people aren't discarding, they're re-using.
I'm not predicting they'll do this... just hoping that they will.
I'd be happy if they would just have a trade in program...where I swap my (42mm Sport) watch for a new one...and perhaps get $100-$150 off the price...I don't see why I'd want components swapped. Let them sell my old watch on eBay or in another country.
i wonder where the slowness really is? It seems from WatchOS 3 it's really fast once the app is in memory and data is loaded? Maybe there's still too much reliance on communicating with the iPhone? Maybe the storage SSD is very slow?
The slowness is likely the result of several things conspiring to impact performance. For consideration:
The proof of concept nature of Apple first new product releases. Apple would much rather get the software right than to release a brand new product on bleeding edge technology. This has certainly been the case with all recent new devices.
It looks like the next rev of Watch OS is heavily refactored. Again this relates to the "proof of concept" releases vs efforts to optimize software. Once you get the basic idea down and have customer acceptance you can ram through optimization to improve the product on existing hardware.
The silicon tech in watch could be much faster. If I remember correctly they had silicon fabricated on a 28nm process, going to 14nm would be a big jump go a step further to 10nm would be massive. Either way you get either much high performance for the same watt of power or extended run times at the same performance levels. Not to mention both options give you a huge advantage to put more stuff on the die saving even more power.
There are other things to consider but I could see a huge number of very happy customers once they see the results of refactored hardware and software.
People aren't playing Infinity blade or editing video on their watch, and they probably never will, so A7 would be massive overkill and a waste of die space and battery power. I'd rather have them spend that transistor budget on LTE.
At this point, a 32 bit processor in the performance range of the A4 would be *plenty* for the Watch. Also maybe a bit more RAM and/or 3d XPoint and/or faster flash to help with speed in launching/switching apps.
I believe when Apple introduced the Watch a while back it was considered that the S1 already had the equivalent performance of an A5!
I guess it depends upon how you measure equivalence. It might have the same cores just clocked differently. I don't really know because I never got into "Watch". My interest in Warch comes from the technology advancements we could see. For example TSMC integrated fan out tech could show up in Watch first. On 14nm that would be a massive improvement over the old Watch.
In in a nut shell the new SoC manufacturing tech could move watch farther ahead than many might suspect. This isn't so much SoC as a stacked arrangement of dies System on Cake (SoC) if you will. {you heard the term here first!!!}
Asking for a guts swap out upgrade to AW2 is like asking for a kidney transplant from the invisible man. It ain't going to happen. What Apple wants you to do is resell your AW1 and buy an AW2. No one asked you to buy a gen 1 Apple Watch. Early adopter's curse.
i wonder where the slowness really is? It seems from WatchOS 3 it's really fast once the app is in memory and data is loaded? Maybe there's still too much reliance on communicating with the iPhone? Maybe the storage SSD is very slow?
The slowness is always going back to the phone for info/processing. I already thought they were cashing things in the watch, but seems they werent'.
With all the phone-like additions to watchOS we're getting to point where Apple could make a "pocket watch" i.e. a small flip-phone based on watchOS as a low end Apple phone...
People aren't playing Infinity blade or editing video on their watch, and they probably never will, so A7 would be massive overkill and a waste of die space and battery power. I'd rather have them spend that transistor budget on LTE.
At this point, a 32 bit processor in the performance range of the A4 would be *plenty* for the Watch. Also maybe a bit more RAM and/or 3d XPoint and/or faster flash to help with speed in launching/switching apps.
The current model is A5 derived and is widely criticized for being too slow, and you want them to go even slower?
The slowness people complain about isn't a CPU problem, it's an app launch problem. More RAM, faster flash would address that.
I don't know what you mean by "A5 derived", but the current model is single core and < 600 MHz. The A4 was 800 MHz.
If a legitimate case can be made that something faster than an A4 is needed, then I'd suggest a single core A6 before going to the A7.
The current model is A5 derived and is widely criticized for being too slow, and you want them to go even slower?
The slowness people complain about isn't a CPU problem, it's an app launch problem. More RAM, faster flash would address that.
I don't know what you mean by "A5 derived", but the current model is single core and < 600 MHz. The A4 was 800 MHz.
If a legitimate case can be made that something faster than an A4 is needed, then I'd suggest a single core A6 before going to the A7.
Clock speed doesn't tell you how fast the CPU's going so not sure what your saying?
People that tested it said it was equivalent to the A5 in CPU speed. Could already be a downlocked A6 in that thing.
The slowness complaint was mostly for non native apps, which means the network chain to retreive the app info is a problem. The way to do it would be to prefect some info/some apps and cache it, off course for this to work the mem/storage subsystem and network should be very fast.
With all the phone-like additions to watchOS we're getting to point where Apple could make a "pocket watch" i.e. a small flip-phone based on watchOS as a low end Apple phone...
The Price of the Watch with that on would probably more than current mid level Android phones (or oldest Iphone).
The problem here is that it is a 64 bit world. Further going 64 bit doesn't add hugely to the transistor budget, most of Apple SoC only have a tiny amount of space dedicated to each CPU.
As for die space I think I'd like to see them do one of two things for RAM. Either build it on the same die, to save a huge amount of power or go to stacked technology. The thing here is that Apple has two demands from customers to contend with. One is longer run times and the other is better performance. Of course people don't play Infinity Blade on Watch (at least not yet) but they do want better performance to run apps better. In any event Apple will gain hugely by going to a latest generation semiconductor process node, there is likely to be space for a lot of features.
Looking at die shots, it seems to me that the CPU cores take up about 25% of an A7 SOC. That's not tiny, and the S1 has a smaller footprint, so the CPU is relatively larger. Also, the key thing about the A7 isn't the 64-bitness, but the fact that it is a much wider (higher IPC) core than the A4 or A5. That's what makes the A7 so big. I don't think it makes sense to put such a big core in a watch at this point. I would much rather use the extra transistor budget afforded by the next die shrink to add an LTE modem. Adding more cache / RAM could also be a good use.
I don't see the point of going to a big A7-class core until we're at, say, 7nm process tech (using TSMC's definition of that term).
The slowness people complain about isn't a CPU problem, it's an app launch problem. More RAM, faster flash would address that.
I don't know what you mean by "A5 derived", but the current model is single core and < 600 MHz. The A4 was 800 MHz.
If a legitimate case can be made that something faster than an A4 is needed, then I'd suggest a single core A6 before going to the A7.
Clock speed doesn't tell you how fast the CPU's going so not sure what your saying?
People that tested it said it was equivalent to the A5 in CPU speed. Could already be a downlocked A6 in that thing.
The slowness complaint was mostly for non native apps, which means the network chain to retreive the app info is a problem. The way to do it would be to prefect some info/some apps and cache it, off course for this to work the mem/storage subsystem and network should be very fast.
The A5 was just two A4 cores. So to say that the S1 is based on the A5 really only makes sense if the S1 was dual core (which it isn't), since being dual core is the only thing that distinguishes the A5 from the A4.
If you hold all else constant, then clock speed is absolutely a meaningful differentiation, and with the 3GS, A4, and S1, it would appear that most else is held constant. The A5 only varies core counts.
Not until the A6 did Apple really start to differentiate its core designs from ARM's in any meaningful way.
Using a 3GS-class CPU in the watch makes plenty of sense to me. The 3GS was a perfectly capable phone for the software of its time, and I doubt the watch will be running software that is any more CPU-intensive than what ran on the 3GS any time soon. I'd use process shrinks to boost the clock speed of the CPU, but not increase its transistor budget. Instead, I'd use the extra transistors to add things like LTE -- that's way more valuable.
I really hope Apple will offer upgrades for existing users, at least to Watch and Edition users (maybe not Sport). The upgrade path I have in mind would be -- take your watch into an Apple Store and have the old guts swapped out and replaced by the new guts (I'm not suggesting that users can or should do this themselves). I imagine the cost of the upgrade would be about the same as the cost of buying a new Sport.
Why do this? Four related points:
1. If they don't, Edition is dead. 2. Many people view watches as keepsakes. Apple Watch cannot compete in that market if people are throwing them out and/or recycling them every two years. 3. It reinforces the view of Apple employees as craftsmen, and Apple products as valuable things, not disposable things 4. It can be pitched as environmentally friendly -- people aren't discarding, they're re-using.
I'm not predicting they'll do this... just hoping that they will.
I'd be happy if they would just have a trade in program...where I swap my (42mm Sport) watch for a new one...and perhaps get $100-$150 off the price...I don't see why I'd want components swapped. Let them sell my old watch on eBay or in another country.
Yeah... a trade-in program might be a more realistic way to do it.
I have to point to those of you naive enough to believe this that: 1. This is DigiTimes we're talking about. Check their track record. Also 2. Unless this "new" Apple Watch has **no hardware changes,** Apple is going to have to tell devs about that in the sessions this week, and that tidbit WILL leak out. So, yeah, that would seem to suggest that new hardware (beyond a minor spec bump, maybe) is out for this year.
To me, it seemed very very obvious from Lynch's demo that there is still a lot Apple can do with the existing Apple Watch before a real 2.0 comes out -- look at how much better the software is, think about the case materials they haven't used yet, and of course they could do much more with bands (I'm talking tech and materials here, not colors) as well. Something possibly a little thinner and faster for Xmas? Maybe. A totally new design? I don't see it this year. Just my $0.02.
"Reports" have been that sales of the Apple Watch jumped between 30-40% after Apple dropped the price, and put a lot of pressure on Android wear and Tizen based watches, as many cost between $250-$300.
my concern is that Apple has done that for the first gen model, but will go back to the old pricing for the 2nd model. It would likely be best if they retain the new pricing. I'm seeing more of them as time goes on, and most all are the sports models. Just yesterday, on the subway here in NYC I saw a young woman, across from me, wearing the smaller model.
im really interested in the 2nd gen model. I want the black SS version. I'm wondering whether we will see an additional case in Liquidmetal at a slightly higher price than the SS versions. While Liquidmetal is too expensive for most of their products, it's perfect for a watch for which they can charge more, as that's a fairly small amount of material. If they do that, I could be interested, depending on the price, including metal band.
I really doubt this will happen.
In in my mind this version is a proof of concept much like the first iPads and iPhones. The fact that the watch was implemented on somewhat older technology processes highlights this.
As as for LiquidMetal that is likely a very cheap way to make a watch. Machining SS is expensive. Injection molded processes are pretty cheap once perfected.
The cost of Liquidmetal is so much higher than SS that the cost is very high. In that, molds need to be made. They cost tens of thousands each, and Apple would need quite a few. Molds wear out, particularly for high precision, and parts would need to be machined anyway. A Liquidmetal watch would sell at a premium. The question is how much of a premium.
The current model is A5 derived and is widely criticized for being too slow, and you want them to go even slower?
The slowness people complain about isn't a CPU problem, it's an app launch problem. More RAM, faster flash would address that.
I don't know what you mean by "A5 derived", but the current model is single core and < 600 MHz. The A4 was 800 MHz.
If a legitimate case can be made that something faster than an A4 is needed, then I'd suggest a single core A6 before going to the A7.
To me the watch always seemed like the prefect device for Hybrid RAM like Intel's new 3D xpoint. After all it's a device for quick interaction with live data. There isn't much on the store that can't be called up a moments notice. Slower RAM that contains everything could be a lot faster overall in that situation than managing the split fast and slower stores.
Edit: Oops the watch already uses SRAM not DRAM for memory so that would make the gap in speed from SRAM to XPoint even larger than I thought it would be.
No if anything Apple needs to go further in not providing sales figures. Providing sales figures for iPhone and iPad is only hurting Apple right now.
They don't have to provide sales numbers for anything. But, if a product, or product line, comprises a "significant" percentage of their business, they must provide the dollar amount of sales. Given what we know about pricing, that would mean that a good guess as to numbers would be possible anyway. Apple rarely provides a breakdown of numbers of models sold within a line.
Truth is Apple has been more than happy to disclose any and all sales figures when those sales figures have been good/great and less inclined to do it when they aren't.
The single most influential improvement they could make with the 2nd generation model would be to dramatically improve the battery life. In my conversations with people who would have considered buying the 1st gen model, that was always the main reason they chose not to. A lot of people just can't get over the fact that there were digital watches 40 years ago that could go two years before needing a new battery. Smart watch or not, 2 days on a charge feels like a major step back.
Comments
There are other things to consider but I could see a huge number of very happy customers once they see the results of refactored hardware and software.
In in a nut shell the new SoC manufacturing tech could move watch farther ahead than many might suspect. This isn't so much SoC as a stacked arrangement of dies System on Cake (SoC) if you will. {you heard the term here first!!!}
I already thought they were cashing things in the watch, but seems they werent'.
I don't know what you mean by "A5 derived", but the current model is single core and < 600 MHz. The A4 was 800 MHz.
If a legitimate case can be made that something faster than an A4 is needed, then I'd suggest a single core A6 before going to the A7.
People that tested it said it was equivalent to the A5 in CPU speed.
Could already be a downlocked A6 in that thing.
The slowness complaint was mostly for non native apps, which means the network chain to retreive the app info is a problem.
The way to do it would be to prefect some info/some apps and cache it, off course for this to work the mem/storage subsystem and network should be very fast.
I don't see the point of going to a big A7-class core until we're at, say, 7nm process tech (using TSMC's definition of that term).
Check out this article: http://www.anandtech.com/show/9381/the-apple-watch-review/3
They found the S1 to be a single core, vanilla ARM Cortex-A7 chip running at 520 MHz.
The A4 was also a Cortex-A7 that had been tweaked to run at higher clock speeds -- about 800 MHz (25% faster than the 3GS, which was also an ARM Cortext-A7). See: http://www.anandtech.com/show/3794/the-iphone-4-review/12
The A5 was just two A4 cores. So to say that the S1 is based on the A5 really only makes sense if the S1 was dual core (which it isn't), since being dual core is the only thing that distinguishes the A5 from the A4.
If you hold all else constant, then clock speed is absolutely a meaningful differentiation, and with the 3GS, A4, and S1, it would appear that most else is held constant. The A5 only varies core counts.
Not until the A6 did Apple really start to differentiate its core designs from ARM's in any meaningful way.
Using a 3GS-class CPU in the watch makes plenty of sense to me. The 3GS was a perfectly capable phone for the software of its time, and I doubt the watch will be running software that is any more CPU-intensive than what ran on the 3GS any time soon. I'd use process shrinks to boost the clock speed of the CPU, but not increase its transistor budget. Instead, I'd use the extra transistors to add things like LTE -- that's way more valuable.
Edit: Oops the watch already uses SRAM not DRAM for memory so that would make the gap in speed from SRAM to XPoint even larger than I thought it would be.