Well, when I first saw 'steroid boost', 'faking performance' in this article, I thought that Note 3 would overclock it's CPU somehow for the the benchmark apps.
But it turns out that Note 3 does not overclock it's CPU for the benchmarks. Sure it clocks, for the benchmarks, at 2.3gHz which is exactly SD800 was advertised for. Where is 'steroid boost' and 'faking performance'?
It does overclock the CPU over what would normally happen without the boosting. With this trickery it keeps every core active running at full speed which doesn't happen under normal circumstances as it will turn certain cores off and dynamically change clock speed. That's why when Ars Technica tricked the Note 3 by renaming the benchmark apps that the numbers went down significantly. Seems you didn't even bother to read the article. From Ars:
Quote:
Above is a picture of Geekbench and of Stealthbench, which is identical to Geekbench in every way except for a different package name. With Geekbench, System Monitor shows that the CPU is locked into 2.3GHz mode and all cores are active, but in Stealthbench, the CPU is allowed to idle, shut off cores, and switch power modes, the same way it does in any other app. We have successfully disabled the special benchmark mode.
Geekbench is a popular benchmarking app, so the Note is programmed to give it special treatment. It has never heard of "Stealthbench," though, so despite being the exact same app, it does not get the special benchmark boost. The Note will run thisbenchmark like 99.99999 percent of the other apps on the device. The next step, then, is to run the two benchmarks and compare the CPU's benchmark mode with its non-benchmark mode.
The difference is remarkable. In Geekbench's multicore test, the Note 3's benchmark mode gives the device a 20 percent boost over its "natural" score. With the benchmark boosting logic stripped away, the Note 3 drops down to LG G2 levels, which is where we initially expected the score to be, given the identical SoCs. This big of a boost means that the Note 3 is not just messing with the CPU idle levels; significantly more oomph is unlocked when the device runs a benchmark.
Base on which metric? The dual core A7 @ 1.3Ghz beats the Samsung Exynos 8-cores @ 1.8Ghz on both efficiency and performance hands down.
Not necessarily.
We know that the A7 at 1.3 GHz RUNNING iOS beats the Exynos @1.8 GHz RUNNING ANDROID. The fact that they're running different operating systems makes it more complex. Maybe the performance is due to the CPU and maybe it's the OS (or maybe both).
If you want to say that one process is faster than the other, you would need to be running the same OS on both.
It does overclock the CPU over what would normally happen without the boosting. With this trickery it keeps every core active running at full speed which doesn't happen under normal circumstances as it will turn certain cores off and dynamically change clock speed. That's why when Ars Technica tricked the Note 3 by renaming the benchmark apps that the numbers went down significantly. Seems you didn't even bother to read the article. From Ars:
No it does not overclock it's 2.3gHz CPU. So, there is no steroid booster nor faking performance here, no matter how you translate it.
We know that the A7 at 1.3 GHz RUNNING iOS beats the Exynos @1.8 GHz RUNNING ANDROID. The fact that they're running different operating systems makes it more complex. Maybe the performance is due to the CPU and maybe it's the OS (or maybe both).
If you want to say that one process is faster than the other, you would need to be running the same OS on both.
That is not necessarily true either, I agree in some test like browser or UI response depends a lot of the OS, but other like Linpack base on compute arithmetic doesn't much.
We know that the A7 at 1.3 GHz RUNNING iOS beats the Exynos @1.8 GHz RUNNING ANDROID. The fact that they're running different operating systems makes it more complex. Maybe the performance is due to the CPU and maybe it's the OS (or maybe both).
If you want to say that one process is faster than the other, you would need to be running the same OS on both.
Benchmarks like Geekbench are good for comparing performance of processors since they don't rely on API's to get their numbers. So it's quite accurate even across OS's.
Now benchmarks like Sunspider or Octane are much different since they are running through API's on the respective OS. This is where the optimization within an OS has a significant effect on the results. And when you look at these benchmarks the A7 on iOS pulls far ahead of the competition.
I think it's safe to say the bare processor A7 is just as fast as the SD800. And that iOS 7 on the A7 is much faster than SD800 on Android.
1. 64-bit CPUs do not have advantage over 32-bit CPUs when running 32-bit tests/benchmarks as the software does not take advantage of the additional registers and their greater size. ARM Cortex A15 already has a 64-bit data path between RAM and CPU cache.
2. Snapdragon is running at a 77% higher frequency, meaning more than twice the power needed (non-linear increase).
3. Snapdragon has twice the cores, meaning twice the consumption when all cores work (as is the case with benchmarks). That's why ARM created the big.LITTLE architecture. Samsung's implementation, however, is awful.
So, for close to 5 times the power consumption you get mind-blowing 3% better performance scores.
4. Benchmarks usually scale extremely well (close to perfect) on multi-core platforms. That is never the case with real-world applications, except for some cases that are usually run on supercomputers, or on GPUs.
So, it seems to me that the iPhone 5s has the most powerful mobile CPU. Apple chose not to over-do it with 4 cores, nor boost the frequency. Because A7's purpose is to power aphone, not a data center.
Stop it, your facts are confusing the issue!
Quote:
Originally Posted by MikeJones
Yeah, and to do so they need the double the cores, double the memory and nearly 40% higher clock speed per core (the non boosted speed rating). Which basically means, they didn't really win anything.
Exactly.
Ars calls it Samsung's "special benchmark mode". That says it all, really. The apologists are working overtime over there in the comments section. Apple isn't perfect and isn't for everyone - I don't have a problem with people choosing a different platform (even if it has a dubious early history), one of my friends loves his HTC One, couldn't be happier. Even my other friend the Windows zealot, loves his Windows phone to death (and if given half a chance will tell you over-and-over all about why it's better!), good on him. But why people would defend a scumbag outfit like Samsung, I'll never know.
Scamscum has to be one of the most morally bankrupt tech companies in the world. What a POS excuse of a company. Does it speak to the morality of South Korean business ethics in general? I sure hope not.
"Samsung Recruited to Join ‘MobileBench’ Consortium Tasked With Creating New Mobile Benchmarks"
And in baseball news, Alex Rodriguez and Ryan Braun are collaborating on new tests for performance-enhancing drugs.
Or like WADA asking the same of Lance Armstrong.
Quote:
it aims to offer more useful tools for mobile platform designers and "more reliable indices" for assessing user experience. MobileBench plans to establish impartial guidelines and a more sophisticated evaluation methodology for both its first benchmark tool, MobileBench and MobileBench-UX, for testing system-level applications.
It gets better. From Anandtech re: Benchmark cheating:
Quote:
Anand Lal Shimpi - Tuesday, October 01, 2013
We've been struggling with how to deal with this one for a little while now. Unfortunately this optimization is far more widespread among Android OEMs and not limited to Samsung or the Galaxy Note 3. We hinted at it in our original international SGS4 investigation and tried to get other OEMs to stop back then as well but with little success.
But why people would defend a scumbag outfit like Samsung, I'll never know.
Let's be fair to every manufacturer. I think there is nothing Samsung needs to defend here. Note 3 does not clock its CPU over 2.3gHz which SD800 is advertised for. There is no such things like 'benchmark booster', 'faking performance', 'benchmark shenanigans'. I am not defending Samsung, but simply putting my opinion here. (I could not see anyone defending Samsung here)
I also think that Samsung phones are ugly and I don't like TW, but my wife loves her Note 1 and she is buying Note 3.
Let's be fair to every manufacturers. I think there is nothing Samsung needs to defend here. Note 3 does not clock its CPU over 2.3gHz which SD800 is advertised for. There is no such things like 'benchmark booster', 'faking performance', 'benchmark shenanigans'. I am not defending Samsung, but simply putting my opinion here. (I could not see anyone defending Samsung here)
Ars Technica and Anandtech disagree with you. I'll take their word over a random forum poster any day.
Let's be fair to every manufacturers. I think there is nothing Samsung needs to defend here. Note 3 does not clock its CPU over 2.3gHz which SD800 is advertised for. There is no such things like 'benchmark booster', 'faking performance', 'benchmark shenanigans'. I am not defending Samsung, but simply putting my opinion here. (I could not see anyone defending Samsung here)
I also think that Samsung phones are ugly and I don't like TW, but my wife loves her Note 1 and she is buying Note 3.
This benchmark tweak was definitely sleazy on Samsung's part. Even though it may meet their advertised spec its not a spec they can meet sustainably. I'm not really surprised by the behavior as I kind of expect it from them. It's like car manufacturers building their bumpers more robustly in only one spot to score better in crash tests since they know the exact test. Blech.
That aside, I do get a lot of hate just for pointing out what the data shows. I don't think there is 'no benefit' to 64 bit architecture and nowhere do I state that. Clearly there must be because when running 64-bit native apps the 5s is almost as fast as the 32 bit Note.
That is all I'm saying, but people get offended by it. You can change the topic to be about mileage or battery life or whatever else you want, that's fine- change the criteria of the benchmark to battery life or some other metric and then Apple can win that test if it makes you feel better. In a side by side test of the two phones, as delivered, the Note beat the 5s in *speed* by a slight amount. At least that is what the data shows- but people get done reading that article and looking at the same data that come to the conclusion "Apple is way faster" simply because they like the brand baffles me.
But why people would defend a scumbag outfit like Samsung, I'll never know.
Let's be fair to every manufacturers. I think there is nothing Samsung needs to defend here. Note 3 does not clock its CPU over 2.3gHz which SD800 is advertised for. There is no such things like 'benchmark booster', 'faking performance', 'benchmark shenanigans'. I am not defending Samsung, but simply putting my opinion here. (I could not see anyone defending Samsung here)
I also think that Samsung phones are ugly and I don't like TW, but my wife loves her Note 1 and she is buying Note 3.
The point is not that it clocks the CPU above its advertised capability, it's that for normal use it clearly has to constrain the CPU well below that capability. That much is unambiguously apparent from the disparity in performance when the device is fooled into not recognizing the app as a benchmark just by changing the app name. So enabling the full capability just for benchmark apps, while not overclocking, is deliberately misleading and certainly justified to be characterized as cheating.
The mental gymnastics on display here in attempting to defend Samsung are rather disappointing.
Ars Technica and Anandtech disagree with you. I'll take their word over a random forum poster any day.
Now I remember why I added him to my block list. Although when someone quotes him I'm stuck reading the stupidity.
It's so funny how the forums are exploding everywhere with Samsung apologists. You can tell it really bothers them that Apple has made such a great SoC in the A7.
For close to 30 years people have been discussing processor efficiency. It's long been agreed that a processor that can do the same work at a lower clock has the superior architecture. Likewise with a processor that can do the same work with fewer cores. Or a processor that does it while consuming less power.
But now we have a serious problem in the universe. Apple, a useless company that doesn't innovate, has created a processor that does all three tremendously well. This has caused scores of people (like Frood) to simply throw out 30 years of accepted facts regarding what constitutes good processor design simply because it's made by a company they dislike.
This goes far beyond the usual level of "pathetic".
Now I remember why I added him to my block list. Although when someone quotes him I'm stuck reading the stupidity.
It's so funny how the forums are exploding everywhere with Samsung apologists. You can tell it really bothers them that Apple has made such a great SoC in the A7.
For close to 30 years people have been discussing processor efficiency. It's long been agreed that a processor that can do the same work at a lower clock has the superior architecture. Likewise with a processor that can do the same work with fewer cores. Or a processor that does it while consuming less power.
But now we have a serious problem in the universe. Apple, a useless company that doesn't innovate, has created a processor that does all three tremendously well. This has caused scores of people (like Frood) to simply throw out 30 years of accepted facts regarding what constitutes good processor design simply because it's made by a company they dislike.
This goes far beyond the usual level of "pathetic".
First of all, I am not a Samsung apologist. And that's fine that you blocked me for whatever the reason.
This limiting CPU speed for some of its applications in Android are not new as Ars notified. So this is a hardly news.
But calling 'benchmark booster', 'faking performance', 'benchmark shenanigans' to one of Android manuracturers and accusing them as 'cheater' because they tested benchmarks at the CPU speed that was advertised for is pure evil journalism to me.
Comments
Well, when I first saw 'steroid boost', 'faking performance' in this article, I thought that Note 3 would overclock it's CPU somehow for the the benchmark apps.
But it turns out that Note 3 does not overclock it's CPU for the benchmarks. Sure it clocks, for the benchmarks, at 2.3gHz which is exactly SD800 was advertised for. Where is 'steroid boost' and 'faking performance'?
It does overclock the CPU over what would normally happen without the boosting. With this trickery it keeps every core active running at full speed which doesn't happen under normal circumstances as it will turn certain cores off and dynamically change clock speed. That's why when Ars Technica tricked the Note 3 by renaming the benchmark apps that the numbers went down significantly. Seems you didn't even bother to read the article. From Ars:
Above is a picture of Geekbench and of Stealthbench, which is identical to Geekbench in every way except for a different package name. With Geekbench, System Monitor shows that the CPU is locked into 2.3GHz mode and all cores are active, but in Stealthbench, the CPU is allowed to idle, shut off cores, and switch power modes, the same way it does in any other app. We have successfully disabled the special benchmark mode.
Geekbench is a popular benchmarking app, so the Note is programmed to give it special treatment. It has never heard of "Stealthbench," though, so despite being the exact same app, it does not get the special benchmark boost. The Note will run this benchmark like 99.99999 percent of the other apps on the device. The next step, then, is to run the two benchmarks and compare the CPU's benchmark mode with its non-benchmark mode.
The difference is remarkable. In Geekbench's multicore test, the Note 3's benchmark mode gives the device a 20 percent boost over its "natural" score. With the benchmark boosting logic stripped away, the Note 3 drops down to LG G2 levels, which is where we initially expected the score to be, given the identical SoCs. This big of a boost means that the Note 3 is not just messing with the CPU idle levels; significantly more oomph is unlocked when the device runs a benchmark.
Not necessarily.
We know that the A7 at 1.3 GHz RUNNING iOS beats the Exynos @1.8 GHz RUNNING ANDROID. The fact that they're running different operating systems makes it more complex. Maybe the performance is due to the CPU and maybe it's the OS (or maybe both).
If you want to say that one process is faster than the other, you would need to be running the same OS on both.
I don't think Samsung was surprised by Apple's 64 bit architecture (lol, they built it).
That is the most wrong statment I have ever seen. They did not build the A7. They manufactured the Apple Designed Build of the A7.
They had absolutely no hand in building the architecture of the A7.
If you meant manufactured it then I can agree with you.
It does overclock the CPU over what would normally happen without the boosting. With this trickery it keeps every core active running at full speed which doesn't happen under normal circumstances as it will turn certain cores off and dynamically change clock speed. That's why when Ars Technica tricked the Note 3 by renaming the benchmark apps that the numbers went down significantly. Seems you didn't even bother to read the article. From Ars:
No it does not overclock it's 2.3gHz CPU. So, there is no steroid booster nor faking performance here, no matter how you translate it.
Not necessarily.
We know that the A7 at 1.3 GHz RUNNING iOS beats the Exynos @1.8 GHz RUNNING ANDROID. The fact that they're running different operating systems makes it more complex. Maybe the performance is due to the CPU and maybe it's the OS (or maybe both).
If you want to say that one process is faster than the other, you would need to be running the same OS on both.
That is not necessarily true either, I agree in some test like browser or UI response depends a lot of the OS, but other like Linpack base on compute arithmetic doesn't much.
Very informative article, thanks. This summarizes things nicely.
If you want an extremely technical article on 64 bit ARMv8 used in the A7 read here.
There are a ton more reasons that the standard stupid "addresses more that 4 Gig is the only advantage of 64 bit" useless statement is a non issue.
That is the very least of reasons to go to ARMv8 and 64 bit.
Now benchmarks like Sunspider or Octane are much different since they are running through API's on the respective OS. This is where the optimization within an OS has a significant effect on the results. And when you look at these benchmarks the A7 on iOS pulls far ahead of the competition.
I think it's safe to say the bare processor A7 is just as fast as the SD800. And that iOS 7 on the A7 is much faster than SD800 on Android.
1. 64-bit CPUs do not have advantage over 32-bit CPUs when running 32-bit tests/benchmarks as the software does not take advantage of the additional registers and their greater size. ARM Cortex A15 already has a 64-bit data path between RAM and CPU cache.
2. Snapdragon is running at a 77% higher frequency, meaning more than twice the power needed (non-linear increase).
3. Snapdragon has twice the cores, meaning twice the consumption when all cores work (as is the case with benchmarks). That's why ARM created the big.LITTLE architecture. Samsung's implementation, however, is awful.
So, for close to 5 times the power consumption you get mind-blowing 3% better performance scores.
4. Benchmarks usually scale extremely well (close to perfect) on multi-core platforms. That is never the case with real-world applications, except for some cases that are usually run on supercomputers, or on GPUs.
So, it seems to me that the iPhone 5s has the most powerful mobile CPU. Apple chose not to over-do it with 4 cores, nor boost the frequency. Because A7's purpose is to power a phone, not a data center.
Stop it, your facts are confusing the issue!
Yeah, and to do so they need the double the cores, double the memory and nearly 40% higher clock speed per core (the non boosted speed rating). Which basically means, they didn't really win anything.
Exactly.
Ars calls it Samsung's "special benchmark mode". That says it all, really. The apologists are working overtime over there in the comments section. Apple isn't perfect and isn't for everyone - I don't have a problem with people choosing a different platform (even if it has a dubious early history), one of my friends loves his HTC One, couldn't be happier. Even my other friend the Windows zealot, loves his Windows phone to death (and if given half a chance will tell you over-and-over all about why it's better!), good on him. But why people would defend a scumbag outfit like Samsung, I'll never know.
DF:
And in baseball news, Alex Rodriguez and Ryan Braun are collaborating on new tests for performance-enhancing drugs.
Or like WADA asking the same of Lance Armstrong.
Laugh out loud.
Since when is cheating the test not faking it! Yes, we all know what a class act Android is!
It gets better. From Anandtech re: Benchmark cheating:
We've been struggling with how to deal with this one for a little while now. Unfortunately this optimization is far more widespread among Android OEMs and not limited to Samsung or the Galaxy Note 3. We hinted at it in our original international SGS4 investigation and tried to get other OEMs to stop back then as well but with little success.
I love this response in the comments:
The cut and paste age.
But why people would defend a scumbag outfit like Samsung, I'll never know.
Let's be fair to every manufacturer. I think there is nothing Samsung needs to defend here. Note 3 does not clock its CPU over 2.3gHz which SD800 is advertised for. There is no such things like 'benchmark booster', 'faking performance', 'benchmark shenanigans'. I am not defending Samsung, but simply putting my opinion here. (I could not see anyone defending Samsung here)
I also think that Samsung phones are ugly and I don't like TW, but my wife loves her Note 1 and she is buying Note 3.
Let's be fair to every manufacturers. I think there is nothing Samsung needs to defend here. Note 3 does not clock its CPU over 2.3gHz which SD800 is advertised for. There is no such things like 'benchmark booster', 'faking performance', 'benchmark shenanigans'. I am not defending Samsung, but simply putting my opinion here. (I could not see anyone defending Samsung here)
Ars Technica and Anandtech disagree with you. I'll take their word over a random forum poster any day.
Let's be fair to every manufacturers. I think there is nothing Samsung needs to defend here. Note 3 does not clock its CPU over 2.3gHz which SD800 is advertised for. There is no such things like 'benchmark booster', 'faking performance', 'benchmark shenanigans'. I am not defending Samsung, but simply putting my opinion here. (I could not see anyone defending Samsung here)
I also think that Samsung phones are ugly and I don't like TW, but my wife loves her Note 1 and she is buying Note 3.
This benchmark tweak was definitely sleazy on Samsung's part. Even though it may meet their advertised spec its not a spec they can meet sustainably. I'm not really surprised by the behavior as I kind of expect it from them. It's like car manufacturers building their bumpers more robustly in only one spot to score better in crash tests since they know the exact test. Blech.
That aside, I do get a lot of hate just for pointing out what the data shows. I don't think there is 'no benefit' to 64 bit architecture and nowhere do I state that. Clearly there must be because when running 64-bit native apps the 5s is almost as fast as the 32 bit Note.
That is all I'm saying, but people get offended by it. You can change the topic to be about mileage or battery life or whatever else you want, that's fine- change the criteria of the benchmark to battery life or some other metric and then Apple can win that test if it makes you feel better. In a side by side test of the two phones, as delivered, the Note beat the 5s in *speed* by a slight amount. At least that is what the data shows- but people get done reading that article and looking at the same data that come to the conclusion "Apple is way faster" simply because they like the brand baffles me.
But why people would defend a scumbag outfit like Samsung, I'll never know.
Let's be fair to every manufacturers. I think there is nothing Samsung needs to defend here. Note 3 does not clock its CPU over 2.3gHz which SD800 is advertised for. There is no such things like 'benchmark booster', 'faking performance', 'benchmark shenanigans'. I am not defending Samsung, but simply putting my opinion here. (I could not see anyone defending Samsung here)
I also think that Samsung phones are ugly and I don't like TW, but my wife loves her Note 1 and she is buying Note 3.
The point is not that it clocks the CPU above its advertised capability, it's that for normal use it clearly has to constrain the CPU well below that capability. That much is unambiguously apparent from the disparity in performance when the device is fooled into not recognizing the app as a benchmark just by changing the app name. So enabling the full capability just for benchmark apps, while not overclocking, is deliberately misleading and certainly justified to be characterized as cheating.
The mental gymnastics on display here in attempting to defend Samsung are rather disappointing.
Ars Technica and Anandtech disagree with you. I'll take their word over a random forum poster any day.
Now I remember why I added him to my block list. Although when someone quotes him I'm stuck reading the stupidity.
It's so funny how the forums are exploding everywhere with Samsung apologists. You can tell it really bothers them that Apple has made such a great SoC in the A7.
For close to 30 years people have been discussing processor efficiency. It's long been agreed that a processor that can do the same work at a lower clock has the superior architecture. Likewise with a processor that can do the same work with fewer cores. Or a processor that does it while consuming less power.
But now we have a serious problem in the universe. Apple, a useless company that doesn't innovate, has created a processor that does all three tremendously well. This has caused scores of people (like Frood) to simply throw out 30 years of accepted facts regarding what constitutes good processor design simply because it's made by a company they dislike.
This goes far beyond the usual level of "pathetic".
Now I remember why I added him to my block list. Although when someone quotes him I'm stuck reading the stupidity.
It's so funny how the forums are exploding everywhere with Samsung apologists. You can tell it really bothers them that Apple has made such a great SoC in the A7.
For close to 30 years people have been discussing processor efficiency. It's long been agreed that a processor that can do the same work at a lower clock has the superior architecture. Likewise with a processor that can do the same work with fewer cores. Or a processor that does it while consuming less power.
But now we have a serious problem in the universe. Apple, a useless company that doesn't innovate, has created a processor that does all three tremendously well. This has caused scores of people (like Frood) to simply throw out 30 years of accepted facts regarding what constitutes good processor design simply because it's made by a company they dislike.
This goes far beyond the usual level of "pathetic".
First of all, I am not a Samsung apologist. And that's fine that you blocked me for whatever the reason.
This limiting CPU speed for some of its applications in Android are not new as Ars notified. So this is a hardly news.
But calling 'benchmark booster', 'faking performance', 'benchmark shenanigans' to one of Android manuracturers and accusing them as 'cheater' because they tested benchmarks at the CPU speed that was advertised for is pure evil journalism to me.