"The 2-3% difference Apple is saying it sees between the battery life of the two processors is well within its manufacturing tolerances for any device, even two iPhones with the same exact processor. In other words, your iPhone and someone else’s iPhone with the same guts likely vary as much as 3%, regardless of who made them."
This appears to be a little bit of a win for TSMC. Even though it's just a little bit, better is still better.
I assumed the Samsung chip would have been slightly better but that does not appear to be the case. Good work TSMC.
Typically, as semiconductor processes shrink, the active power of a gate decreases, while the leakage of the gate increases. The leakage power is constant no matter what the gate is doing, while the active power depends if the gate is clocked or not. The overall power is a combination of the active and the leakage. The leakage power consumption is several factors smaller than the dynamic power consumption. So if the gate is constantly used, the active power will dominate the power consumption. If the gate is not used, then the leakage power will dominate the power consumption.
Thus, a shrink from one process node to a smaller process node does not lead to a linear drop in power consumption. This is then complicated by what those gates are doing, the voltage of the gates, and the distance between them. So considering that going from 16nm to 14nm is 87.5% shrink at the same voltage and running the same code, the best possible case would be 12.5% decrease assuming a linear drop in power. But since power savings is not typically linear in a shrink at those geometries, expect at best half of this savings, say 6%. Furthermore, since not all gates are being used all the time, many will be just leaking increasing the power consumption of the unused gates. This typically drops it again to half, so 3%. That is inline with Apple's 2-3% difference.
The apps that measure battery life by running the system full on, will not take into account the leakage power. So this is not a good measure of a phone that is in standby most of the day. Thus, those apps will tend to show more battery savings than one that did some activity, fell back into standby and then wake up a later time and did some more activities.
What they're really saying is that we intensively use the phone only part of the time. If you measured those intense use periods the battery life might increase 10% (eg: 5hrs goes to 5h30 of intense use). But once you factor in all the time it just sits in your pocket etc, it's a much lower amount (eg: 15hrs goes to 15h30 of average use, 3.3% difference).
edit: actually that's wrong. If it was 10% intense use, then average use would also be 10%.
15h would go to 16.5 hours average use. My logic was flawed (and Apple says it's 2-3%).
I was under the impression that the opposite was true. So smaller die size results in worse energy efficiency?
The said in the story that they were surprised that the samsung version which is built on the smaller 14nm consumed more energy in there tests than 16nm version from TSMC. Normally smaller is less power. Trust samsung to screw that up too.
I don't know, but I can only assume 2-3% is well within established acceptable variances for any product with a battery. On a ten hour battery, a 3% reduction is 18 minutes. Real world usage variances could easily obscure any differences--ones that would only be seen with two side-by-side phones with different chips. In other words, even that much of a difference is meaningless in the real world.
I don't know, but I can only assume 2-3% is well within established acceptable variances for any product with a battery. On a ten hour battery, a 3% reduction is 18 minutes. Real world usage variances could easily obscure any differences--ones that would only be seen with two side-by-side phones with different chips. In other words, even that much of a difference is meaningless in the real world.
Especially given day to day my usage probably varies more than that in terms of power intensive versus light power usage functions. So the slop of a few percent would get lost in the noise of everyday use.
http://arstechnica.com/tech-policy/2015/08/former-tsmc-employee-leaked-secrets-to-samsung-taiwan-supreme-court-says/ This in spite of the fact that the copycat clowns at Samsung stole trade secrets from TSMC for the 28nm process. I really hope Apple switches to TSMC completely, for the iPhone 7. Since 10nm will be ready late 2016 (by TSMC's own admission), Apple cannot go smaller for the iPhone 7, but this will give TSMC enough time to ramp up production capacity.
At least from a political perspective, I'm happy to have a TSMC iPhone 6s. :-) (Yes, I know there are other Simpson components in the phone).
This in spite of the fact that the copycat clowns at Samsung stole trade secrets from TSMC for the 28nm process. I really hope Apple switches to TSMC completely, for the iPhone 7. Since 10nm will be ready late 2016 (by TSMC's own admission), Apple cannot go smaller for the iPhone 7, but this will give TSMC enough time to ramp up production capacity.
At least from a political perspective, I'm happy to have a TSMC iPhone 6s. :-) (Yes, I know there are other Simpson components in the phone).
If true, no biggie for Apple. But this can be a death nail for Samsung who just reported 'profit' with the chips division. Big win for TSMC, death for Samsung and no big Deal for Apple
If true, no biggie for Apple. But this can be a death nail for Samsung who just reported 'profit' with the chips division. Big win for TSMC, death for Samsung and no big Deal for Apple
Just keep cutting back on the Samsung, and order more of the TSMC.
This in spite of the fact that the copycat clowns at Samsung stole trade secrets from TSMC for the 28nm process. I really hope Apple switches to TSMC completely, for the iPhone 7. Since 10nm will be ready late 2016 (by TSMC's own admission), Apple cannot go smaller for the iPhone 7, but this will give TSMC enough time to ramp up production capacity.
At least from a political perspective, I'm happy to have a TSMC iPhone 6s. :-) (Yes, I know there are other Simpson components in the phone).
@Shervin : Not sure if TSMC would get ahead of Samsung in 10nm. But, from political / economic perspective, sure, we all love Taiwan made SOCs (TSMC) over US made SOCs (Samsung) -- whatever works for Apple.
Part of me thinks Apple should not have responded to this. Apple's response is going to give it even more attention, attention it probably wouldn't otherwise have received. This wasn't really a "gate" yet and was mostly just floating around tech/rumor text sites. Now Apple has legitimized it as a story and I won't be surprised if some idiots who ran an app to find out what processor they have will run to an Apple store demanding a phone with a TSMC chip. And I won't be surprised if some lawyer is out there just waiting to file a class action lawsuit claiming Apple lied about battery life. :rolleyes:
They had to respond. Articles were getting posted on more main stream media asking if you got the good iPhone Or the bad one. Apple doesn't need people worrying they have the "bad" one with an hour plus difference in battery. They say 2-3% and they should know. that is small enough that I want the TSMC one, but I won't return a Samsung based one to try and get the other.
If true, no biggie for Apple. But this can be a death nail for Samsung who just reported 'profit' with the chips division. Big win for TSMC, death for Samsung and no big Deal for Apple
Maybe it's another auto correct error like @SpamSandwich found in this thread but it's death knell.
Comments
According to the linked TechCrunch article:
"The 2-3% difference Apple is saying it sees between the battery life of the two processors is well within its manufacturing tolerances for any device, even two iPhones with the same exact processor. In other words, your iPhone and someone else’s iPhone with the same guts likely vary as much as 3%, regardless of who made them."
This appears to be a little bit of a win for TSMC. Even though it's just a little bit, better is still better.
I assumed the Samsung chip would have been slightly better but that does not appear to be the case. Good work TSMC.
Typically, as semiconductor processes shrink, the active power of a gate decreases, while the leakage of the gate increases. The leakage power is constant no matter what the gate is doing, while the active power depends if the gate is clocked or not. The overall power is a combination of the active and the leakage. The leakage power consumption is several factors smaller than the dynamic power consumption. So if the gate is constantly used, the active power will dominate the power consumption. If the gate is not used, then the leakage power will dominate the power consumption.
Thus, a shrink from one process node to a smaller process node does not lead to a linear drop in power consumption. This is then complicated by what those gates are doing, the voltage of the gates, and the distance between them. So considering that going from 16nm to 14nm is 87.5% shrink at the same voltage and running the same code, the best possible case would be 12.5% decrease assuming a linear drop in power. But since power savings is not typically linear in a shrink at those geometries, expect at best half of this savings, say 6%. Furthermore, since not all gates are being used all the time, many will be just leaking increasing the power consumption of the unused gates. This typically drops it again to half, so 3%. That is inline with Apple's 2-3% difference.
The apps that measure battery life by running the system full on, will not take into account the leakage power. So this is not a good measure of a phone that is in standby most of the day. Thus, those apps will tend to show more battery savings than one that did some activity, fell back into standby and then wake up a later time and did some more activities.
3% of 24 hours is 45 minutes. That matters during a day.
Anyone now if it's possible to specifically order/ask for an iPhone with the TMSC chip?
One could of course use the return policy until one gets lucky...
A 2-3% battery life increase is still a fair bit.
What they're really saying is that we intensively use the phone only part of the time. If you measured those intense use periods the battery life might increase 10% (eg: 5hrs goes to 5h30 of intense use). But once you factor in all the time it just sits in your pocket etc, it's a much lower amount (eg: 15hrs goes to 15h30 of average use, 3.3% difference).
edit: actually that's wrong. If it was 10% intense use, then average use would also be 10%.
15h would go to 16.5 hours average use. My logic was flawed (and Apple says it's 2-3%).
Ahh well.
I was under the impression that the opposite was true. So smaller die size results in worse energy efficiency?
The said in the story that they were surprised that the samsung version which is built on the smaller 14nm consumed more energy in there tests than 16nm version from TSMC. Normally smaller is less power. Trust samsung to screw that up too.
This in spite of the fact that the copycat clowns at Samsung stole trade secrets from TSMC for the 28nm process. I really hope Apple switches to TSMC completely, for the iPhone 7. Since 10nm will be ready late 2016 (by TSMC's own admission), Apple cannot go smaller for the iPhone 7, but this will give TSMC enough time to ramp up production capacity.
At least from a political perspective, I'm happy to have a TSMC iPhone 6s. :-) (Yes, I know there are other Simpson components in the phone).
http://arstechnica.com/tech-policy/2015/08/former-tsmc-employee-leaked-secrets-to-samsung-taiwan-supreme-court-says/
This in spite of the fact that the copycat clowns at Samsung stole trade secrets from TSMC for the 28nm process. I really hope Apple switches to TSMC completely, for the iPhone 7. Since 10nm will be ready late 2016 (by TSMC's own admission), Apple cannot go smaller for the iPhone 7, but this will give TSMC enough time to ramp up production capacity.
At least from a political perspective, I'm happy to have a TSMC iPhone 6s. :-) (Yes, I know there are other Simpson components in the phone).
D'oh!
If true, no biggie for Apple. But this can be a death nail for Samsung who just reported 'profit' with the chips division. Big win for TSMC, death for Samsung and no big Deal for Apple
Just keep cutting back on the Samsung, and order more of the TSMC.
http://arstechnica.com/tech-policy/2015/08/former-tsmc-employee-leaked-secrets-to-samsung-taiwan-supreme-court-says/
This in spite of the fact that the copycat clowns at Samsung stole trade secrets from TSMC for the 28nm process. I really hope Apple switches to TSMC completely, for the iPhone 7. Since 10nm will be ready late 2016 (by TSMC's own admission), Apple cannot go smaller for the iPhone 7, but this will give TSMC enough time to ramp up production capacity.
At least from a political perspective, I'm happy to have a TSMC iPhone 6s. :-) (Yes, I know there are other Simpson components in the phone).
@Shervin : Not sure if TSMC would get ahead of Samsung in 10nm. But, from political / economic perspective, sure, we all love Taiwan made SOCs (TSMC) over US made SOCs (Samsung) -- whatever works for Apple.
They had to respond. Articles were getting posted on more main stream media asking if you got the good iPhone Or the bad one. Apple doesn't need people worrying they have the "bad" one with an hour plus difference in battery. They say 2-3% and they should know. that is small enough that I want the TSMC one, but I won't return a Samsung based one to try and get the other.
Take a peek under the hood.
If true, no biggie for Apple. But this can be a death nail for Samsung who just reported 'profit' with the chips division. Big win for TSMC, death for Samsung and no big Deal for Apple
Maybe it's another auto correct error like @SpamSandwich found in this thread but it's death knell.
D'oh!
I disagree. I think it is wise to quash these stupid rumors.
FingerGate.
BendGate.
ChipGate.
Quash them all. I wish Apple was just as aggressive in killing off stupid rumors from the far east about weak demand from dumb ass suppliers.
There is only one "-Gate" and that is ?Gate. Anything that Apple releases will have some horrendous problem. Every other "-Gate" is just a subset! /s
This is, in effect, a double negative, so it's easy to see why it created confusion.
As to the facts - operating speed, operating voltage, and gate switching times, are important, too. This suggests it is just one factor.
Samsung deliberately hindering Apple iPhone?
There is only one "-Gate" and that is ?Gate. Anything that Apple releases will have some horrendous problem. Every other "-Gate" is just a subset! /s
Everyone remembers back in 1997 when Microsoft purchased $150 million of Apple shares.
I really think we should start referring to that as "Gates-Gate".