If all this is true the big losser is tegra 3. A5 beats it to a pul and the new iPad 3 will do it even worse plus potentially having LTE support which tegra 3 can't. Also the dual core cortex 15 chios that don't support LTE...
Also Intel is suppose to take mobile by 2014-15 with there GPU performance and finally cracking the battery live issue. By that time IMHO arm will have the market and their wont be any space for intel...
BeltsBear you would likely lost that bet and here is why.
When taking in account battery life it's really just simple math. The A5 processor is fabricated on a 45nm process. The new processors coming out this year are fabricated on 28 or 32nm which is a considerable shrink.
This means less power and size and higher yields (hopefully) per wafer. What each vendor then does is make a decision. They have a certain range of power consumption their products can handle. Thus they must stay within this Thermal Design Power (TDP) range.
So the options are to increase the clock frequency of a dual core (as the smaller process allows for this) or bolt on another two cores running about the same frequency (which is what most vendors are doing)
But that doesn't answer your question really. Would Apple stick with Dual Core or move to Quad Core. Well that comes down to the the ARM processor itself. The Cortex A9 has 8 pipleline stage which is low in contrast to an Intel Core I5/I7 14 stage pipeline.
The more pipelines the higher you can clock a processor.
So at this point with only 8 pipelines the ARM core isn't likely going to clock high enough to make a significant different in speed as opposed to the benefits of adding another two cores (more execution units, cache, everything)
So in this bet the odds would be that you would lose unless Apple pulls the proverbial rabbit from the hat.
You seem to know a lot about processors but are ignoring the fact that more cores do not always feel faster. Getting software to use them 100% efficiently can be a challenge even for Apple.
Apple will increase the speed enough to make it feel just right, and will do more with the graphics processor then they do with the main processor. That is why it is being rumored as the A5X. If it had 4 cores it would probably be called the A6.
While I think they are right about the sales and will certainly buy one myself at the earliest opportunity, I think it's a bit of a stretch to call this a "significant update."
It will look almost identical to the second generation version. It will be slightly faster and have a sharper screen.
The screen alone is a monumental update.
The processor gain probably won't be realised because of the increased graphics demands but it the update will be nicely rounded out by Siri and better camera optics.
I'd be more likely to consider cosmetic and form-factor changes as insignificant updates. The iPad 2 was a notable exception in this regard because the rounded back of the iPad 1 was completely fugly, unnecessary and impeded function.
I have an uneasy feeling there is some massive shorting of Apple stock going on at the moment with all this insane boosting on very little evidence. It will only take one thing missing from the fanboy wish list (dual core vs quad, 3G rather than LTE) and AAPL will nosedive.
You seem to know a lot about processors but are ignoring the fact that more cores do not always feel faster. Getting software to use them 100% efficiently can be a challenge even for Apple.
Apple will increase the speed enough to make it feel just right, and will do more with the graphics processor then they do with the main processor. That is why it is being rumored as the A5X. If it had 4 cores it would probably be called the A6.
It will be two cores.
I admit to being intrigued friend. It could go either way but one bloke made an interesting comment. He said something like he hadn't seen mention of any processors that were fabbed at 28nm that were dual core.
I have an uneasy feeling there is some massive shorting of Apple stock going on at the moment with all this insane boosting on very little evidence. It will only take one thing missing from the fanboy wish list (dual core vs quad, 3G rather than LTE) and AAPL will nosedive.
I am not disappointed, I'm just saying it's an overstatement to refer to it as "significant."
You can't be serious. The entire face of the iPad is a touch screen, and that is getting a very major improvement. Beyond "significant." Going from the iPad/iPad 2 resolution to a retina display is huge. That alone is reason to be excited about the iPad 3, but we're also most likely getting a significantly faster processor and a significantly better camera. Plus Siri? And who knows what else... but that's not significant enough for you?
LTE is the piece I think has the least chance of making it intO the next iPad, until a minor spec bump in the fall that adds LTE to the iPhone at the same time. Quad res, better cams, Siri, and a better processor (whatever it ends up being) possibly 32GB base but I may jut be hopeful on the last one.
LTE is the piece I think has the least chance of making it intO the next iPad, until a minor spec bump in the fall that adds LTE to the iPhone at the same time.
Apple doesn't do "spec bumps" on iOS devices. It's all or nothing.
You seem to know a lot about processors but are ignoring the fact that more cores do not always feel faster. Getting software to use them 100% efficiently can be a challenge even for Apple.
Apple will increase the speed enough to make it feel just right, and will do more with the graphics processor then they do with the main processor. That is why it is being rumored as the A5X. If it had 4 cores it would probably be called the A6.
Microsoft will never be a long-term success because they aren't a variety of fruit. Check out this new baby: Raspberry Pi, starts at $25, runs Linux.
The computing world will soon be divided into cheap running free software (these things) and great designs (Apple). Where does Microsoft fit in?
Never underestimate the importance of a case...
That will be in the major consumer release. Here's the progression of the PC industry:
hobbyist toys without cases, some assembly required
nice cases (Apple II)
business cred (IBM)
mass market cred (cheap clones)
commodities (ultra-cheap clones)
no profits because high markups shift to cool portable devices
Raspberry Pi looks like doing the first 5 steps by September. If anyone can get a Linux-capable PC for $25-$35 + cheap power supply, mouse, keyboard and screen, what's Microsoft's business case for selling software for hundreds of dollars into this market? These things are about as fast as a Pentium II 300MHz which can run pretty good apps, and that will put a lot of pressure on the free software world to slim down.
The more pipelines the higher you can clock a processor.
So at this point with only 8 pipelines the ARM core isn't likely going to clock high enough to make a significant different in speed as opposed to the benefits of adding another two cores (more execution units, cache, everything)
True up to a point (you mean pipeline stages, not pipelines). There are diminishing returns from more pipe stages, since you have more logic overhead as you add more stages. Also, the deeper the pipeline, the higher the cost of branch mis-predictions, since you have more instructions at various stages to discard. You can also gain speed by making the caches bigger without more cores or a more aggressive pipeline, at lower cost in energy. What you care about most is not clock speed but throughput and a higher clock speed does not guarantee higher throughput if cache misses aren't reduced and you get more pipeline stalls.
LTE is the piece I think has the least chance of making it intO the next iPad, until a minor spec bump in the fall that adds LTE to the iPhone at the same time. Quad res, better cams, Siri, and a better processor (whatever it ends up being) possibly 32GB base but I may jut be hopeful on the last one.
Not really. Quad Core isn't a guarantee. LTE is not as expendable as people think here.
True up to a point (you mean pipeline stages, not pipelines). There are diminishing returns from more pipe stages, since you have more logic overhead as you add more stages. Also, the deeper the pipeline, the higher the cost of branch mis-predictions, since you have more instructions at various stages to discard. You can also gain speed by making the caches bigger without more cores or a more aggressive pipeline, at lower cost in energy. What you care about most is not clock speed but throughput and a higher clock speed does not guarantee higher throughput if cache misses aren't reduced and you get more pipeline stalls.
Indeed the SoC cache doubled to 1MB with the A5 which certainly helped. Remember the "heady" days of Intel Netburst architecture with often more than 20 pipeline stages (Prescott had 31!!!) don't want to return to those days.
I'm pretty excited about the ARM Cortex A15. 15 stage integer pipeline, up to 25 stage floating point pipeline and up to 4MB L2 cache.
Comments
Just remember what people originally thought of the "significance" of the 4S' upgraded specs.
I really need a built in, slide out and flip up thin keyboard. and more responsiveness.
Your dream come true:
http://www.nationmultimedia.com/top40/detail/7176
BeltsBear you would likely lost that bet and here is why.
When taking in account battery life it's really just simple math. The A5 processor is fabricated on a 45nm process. The new processors coming out this year are fabricated on 28 or 32nm which is a considerable shrink.
This means less power and size and higher yields (hopefully) per wafer. What each vendor then does is make a decision. They have a certain range of power consumption their products can handle. Thus they must stay within this Thermal Design Power (TDP) range.
So the options are to increase the clock frequency of a dual core (as the smaller process allows for this) or bolt on another two cores running about the same frequency (which is what most vendors are doing)
But that doesn't answer your question really. Would Apple stick with Dual Core or move to Quad Core. Well that comes down to the the ARM processor itself. The Cortex A9 has 8 pipleline stage which is low in contrast to an Intel Core I5/I7 14 stage pipeline.
The more pipelines the higher you can clock a processor.
So at this point with only 8 pipelines the ARM core isn't likely going to clock high enough to make a significant different in speed as opposed to the benefits of adding another two cores (more execution units, cache, everything)
So in this bet the odds would be that you would lose unless Apple pulls the proverbial rabbit from the hat.
You seem to know a lot about processors but are ignoring the fact that more cores do not always feel faster. Getting software to use them 100% efficiently can be a challenge even for Apple.
Apple will increase the speed enough to make it feel just right, and will do more with the graphics processor then they do with the main processor. That is why it is being rumored as the A5X. If it had 4 cores it would probably be called the A6.
It will be two cores.
While I think they are right about the sales and will certainly buy one myself at the earliest opportunity, I think it's a bit of a stretch to call this a "significant update."
It will look almost identical to the second generation version. It will be slightly faster and have a sharper screen.
The screen alone is a monumental update.
The processor gain probably won't be realised because of the increased graphics demands but it the update will be nicely rounded out by Siri and better camera optics.
I'd be more likely to consider cosmetic and form-factor changes as insignificant updates. The iPad 2 was a notable exception in this regard because the rounded back of the iPad 1 was completely fugly, unnecessary and impeded function.
You seem to know a lot about processors but are ignoring the fact that more cores do not always feel faster. Getting software to use them 100% efficiently can be a challenge even for Apple.
Apple will increase the speed enough to make it feel just right, and will do more with the graphics processor then they do with the main processor. That is why it is being rumored as the A5X. If it had 4 cores it would probably be called the A6.
It will be two cores.
I admit to being intrigued friend. It could go either way but one bloke made an interesting comment. He said something like he hadn't seen mention of any processors that were fabbed at 28nm that were dual core.
I'm further intrigued by the mention of Macroscalar as a Trademark and set of patents for Apple as discussed here http://arstechnica.com/apple/news/20...-processor.ars
Makes me wonder if Apple's confronting the necessity of core scaling by working towards keeping the cores full of data.
March 7th can't come soon enough. Cheers!
I have an uneasy feeling there is some massive shorting of Apple stock going on at the moment with all this insane boosting on very little evidence. It will only take one thing missing from the fanboy wish list (dual core vs quad, 3G rather than LTE) and AAPL will nosedive.
Welcome to every keynote ever.
I am not disappointed, I'm just saying it's an overstatement to refer to it as "significant."
You can't be serious. The entire face of the iPad is a touch screen, and that is getting a very major improvement. Beyond "significant." Going from the iPad/iPad 2 resolution to a retina display is huge. That alone is reason to be excited about the iPad 3, but we're also most likely getting a significantly faster processor and a significantly better camera. Plus Siri? And who knows what else... but that's not significant enough for you?
Wow.
LTE is the piece I think has the least chance of making it intO the next iPad, until a minor spec bump in the fall that adds LTE to the iPhone at the same time.
Apple doesn't do "spec bumps" on iOS devices. It's all or nothing.
You seem to know a lot about processors but are ignoring the fact that more cores do not always feel faster. Getting software to use them 100% efficiently can be a challenge even for Apple.
Apple will increase the speed enough to make it feel just right, and will do more with the graphics processor then they do with the main processor. That is why it is being rumored as the A5X. If it had 4 cores it would probably be called the A6.
It will be two cores.
You could both be right.
It could be a dual core CPU and quad core GPU.
J.
Microsoft will never be a long-term success because they aren't a variety of fruit. Check out this new baby: Raspberry Pi, starts at $25, runs Linux.
The computing world will soon be divided into cheap running free software (these things) and great designs (Apple). Where does Microsoft fit in?
Never underestimate the importance of a case...
That will be in the major consumer release. Here's the progression of the PC industry:
- hobbyist toys without cases, some assembly required
- nice cases (Apple II)
- business cred (IBM)
- mass market cred (cheap clones)
- commodities (ultra-cheap clones)
- no profits because high markups shift to cool portable devices
Raspberry Pi looks like doing the first 5 steps by September. If anyone can get a Linux-capable PC for $25-$35 + cheap power supply, mouse, keyboard and screen, what's Microsoft's business case for selling software for hundreds of dollars into this market? These things are about as fast as a Pentium II 300MHz which can run pretty good apps, and that will put a lot of pressure on the free software world to slim down.[...]
The more pipelines the higher you can clock a processor.
So at this point with only 8 pipelines the ARM core isn't likely going to clock high enough to make a significant different in speed as opposed to the benefits of adding another two cores (more execution units, cache, everything)
True up to a point (you mean pipeline stages, not pipelines). There are diminishing returns from more pipe stages, since you have more logic overhead as you add more stages. Also, the deeper the pipeline, the higher the cost of branch mis-predictions, since you have more instructions at various stages to discard. You can also gain speed by making the caches bigger without more cores or a more aggressive pipeline, at lower cost in energy. What you care about most is not clock speed but throughput and a higher clock speed does not guarantee higher throughput if cache misses aren't reduced and you get more pipeline stalls.
Apple doesn't do "spec bumps" on iOS devices. It's all or nothing.
funny, my second iPhone was a 16GB edge model. i think it was a spec bump from my 8gb iphone after it was stolen
funny, my second iPhone was a 16GB edge model. i think it was a spec bump from my 8gb iphone after it was stolen
Oh, right, they did add a 16, didn't they?
They also dropped the 4, so that makes sense.
As an aside, I love owning the Apple product with the shortest lifespan ever.
LTE is the piece I think has the least chance of making it intO the next iPad, until a minor spec bump in the fall that adds LTE to the iPhone at the same time. Quad res, better cams, Siri, and a better processor (whatever it ends up being) possibly 32GB base but I may jut be hopeful on the last one.
Not really. Quad Core isn't a guarantee. LTE is not as expendable as people think here.
True up to a point (you mean pipeline stages, not pipelines). There are diminishing returns from more pipe stages, since you have more logic overhead as you add more stages. Also, the deeper the pipeline, the higher the cost of branch mis-predictions, since you have more instructions at various stages to discard. You can also gain speed by making the caches bigger without more cores or a more aggressive pipeline, at lower cost in energy. What you care about most is not clock speed but throughput and a higher clock speed does not guarantee higher throughput if cache misses aren't reduced and you get more pipeline stalls.
Indeed the SoC cache doubled to 1MB with the A5 which certainly helped. Remember the "heady" days of Intel Netburst architecture with often more than 20 pipeline stages (Prescott had 31!!!) don't want to return to those days.
I'm pretty excited about the ARM Cortex A15. 15 stage integer pipeline, up to 25 stage floating point pipeline and up to 4MB L2 cache.