Oh another one of you 4 Gig people. This theory has already been discussed ad nauseum, there are huge benefits of a 64 bit processor the least of which is addressing more that 4 Gig of memory. Anand La Shimpi says so as do others more knowledgeable than you. You should stop reading the android blogs because this is there favorite line. Heck even Qualcomm retracted there Senior VP idiots statement as being false for claiming the same thing and demoted him. There a chip maker they know more about it than you or I. And for your info all 1st party apps on iOS are full 64 bit.
Sigh....
The real performance increase in 64 bit is for applications that handle large data arrays; so things like CAD/CAE-like programs, data managers, programs to work with digital media and applied scientific applications, so yes iphone 5s with it current specs would see quite an increase in performance when executing such software.
So yes I agree with you there are other huge benefits of a 64 bit processor besides addressing 4 gigs of memory, but those benefits really only apply to large data array software. Show me that the average user of an iphone 5s or an ipad primarily or at the very least plan on using such software for a good portion of their time when using those devices.
Quote:
Originally Posted by Tallest Skil
If you had, you wouldn’t be repeating this worthless nonsense.
Completely and utterly incorrect. Educate yourself. I’ll repeat it until you do.
Huh? Address bus is used solely for addressing memory and only memory. So the only benefit of a 64 bit address bus is when you have more than 4 GB of memory. If you are not running software that uses more than 2 GB in a single thread or the entire system is not using more than 4 GB of memory there is no performance benefit to using a 64 bit address bus. What else is the address bus used for Tallest Skil?
The only time a 64 bit cpu provides any meaningful performance increase when you don't have more than 4 GB or at the very least 2 GB to run a single thread is when you are executing or running large data array software that takes advantage of the additional registers. Show me that the average user or even significant portion of iphone or ipad users is using such software or plans to use it.
Address bus is used solely for addressing memory and only memory. So the only benefit of a 64 bit address bus is when you have more than 4 GB of memory. If you are not running software that uses more than 2 GB in a single thread or the entire system is not using more than 4 GB of memory there is no performance benefit to using a 64 bit address bus. The only time a 64 bit cpu provides any meaningful performance increase when you don’t have more than 4 GB…
So why does the iPhone 5S show up to a 25% speed improvement on itself between otherwise identical 32-bit and 64-bit compiled apps?
So why does the iPhone 5S show up to a 25% speed improvement on itself between otherwise identical 32-bit and 64-bit compiled apps?
Key word is up to, what is the Mode? Or most common speed improvement? Large data array software such as encryption, encoding, decoding, video editing software will give you those types of speed improvements if not higher speed improvements. Also is it comparing or are you comparing it to the 32 bit version of A7 processor? Lets take for example the most commonly used software, the Internet browser.
Internet Explorer 10 and 11 64 bit is only around 1-2% faster on benchmarks such as sunspider, kraken, v8 benchmarks compared to its 32 bit version. That is hardly a meaningful improvement.
So which software or apps is giving you those 25% speed improvements?
Keep this in mind: higher core clock speeds, faster LPDDR3 memory and quad-core PowerVR graphics chip — all independent improvements that the A7 makes, how much of those 25% speed improvements are attributed to that and not the 64 bit.
What is the most commonly used software that an iphone or ipad user uses? I have a feeling that it is not large data array software.
If you look at ST88 post you will see that according to Futuremark Benchmark the 64 bit version on average is only 7% faster, not enough to move it around on the rankings for fastest mobile phone.
Here is what Futuremark said:
64-bit is not the answer
We started our investigation by compiling a 64-bit version of 3DMark, since the version available in the App Store is 32-bit in order to be compatible with older Apple devices.
Once we had a 64-bit build we tested an iPhone 5s under controlled conditions in our Test Lab. As expected, we found that the results from the 64-bit version were similar to the 32-bit version, with only a 7 percent improvement.
So as I have said before large data array software such as encoding, encryption will give you those types of large speed improvements but if you are talking about the day to day software that the average owner of an iphone or ipad user uses than chances are no it doe not make any real meaningful performance increase.
Quote:
Originally Posted by PhilBoogie
Perhaps his IQ only goes up to 32? More likely to have a single-digit IQ though.
Brilliant response! Did you know that Tallest Skil literally thought address bus did more than address memory.
Quote:
Originally Posted by SolipsismX
Even though it's been mentioned he has completely ignored the new AArch64 ISA.
You know, a 7% improvement is really about what I'd expect to see on average for a 64 bit transition when memory wasn't a constraint. In private speculation in the runup to the 5s device introduction, I didn't think 64 bit was even on the table, assuming there wouldn't be any benefit until more RAM was built into the package.
You can't be serious with that question. Talking about the A7 processor is addressing the ISA.
Yes, I'm serious. Where did you address AArch64? Where did you discredit the new ISA by saying it offers no benefits to the OS or SW?
PS: In your previous post you're compressing a bunch of different types of 64-bit items into one poorly contrived argument. You can't use IE 11 as being 1% faster as proof that the new ISA in AArch64 offers no benefit.
Yes, I'm serious. Where did you address AArch64? Where did you discredit the new ISA by saying it offers no benefits to the OS or SW?
PS: In your previous post you're compressing a bunch of different types of 64-bit items into one poorly contrived argument. You can't use IE 11 as being 1% faster as proof that the new ISA in AArch64 offers no benefit.
Already answered that question go back and read and the answer is not that simple.
No, no you didn't. You discredited everything that comes with AArch64 and made sweeping ignorant generalizations and even made foolish comments about RAM capacity and CPU bitness.
PS: I checked your posts. You didn't use ISA, AArch64, or the colloquial ARM64 once in your comments.
Even though it's been mentioned he has completely ignored the new AArch64 ISA.
yes he has. He overlooks that AArch64 doubled the number of registers over 32bit ARM. 32-bit ARM provides 16 integer registers, of which one is a dedicated program counter, two more are give over to a stack pointer and link register, and the other 13 are available for general use. With AArch 64 there are 32 integer registers, with a dedicated zero register, dedicated link register and dedicated frame pointer register. One further register is dedicated to the platform leaving 28 general purpose registers.
AArch 64 also increases the number of floating point registers as well. AArch 32 has 32 32 bit floating point registers. Which can overlap into various sized of floating points as needed. AArch 64 simplifies this and has 32 128 bit floating point registers that can be used for smaller amounts of data, but are dedicated and do not overlap which is a big performance win. With AArch 64 Physical RAM address size is decouple from CPU fitness.Data Bus size the amount of data fetched from RAM is likewise decoupled. A CPU instruction may request a certain size of data but the amount can be independentlycontrolled by either fetching in smaller amounts or fetching more than is necessary.Also a revised and streamlined the instruction set for a big performance gain. Also apple took advantage of AArch64 in a big way with an inline retain count, which eliminates the need to perform costly hash table lookup for retain and release operations in the common case. Those operations are very common in most Objective-C code and is a big performance win. Per-object resource cleanup flags make object deallocation a lot faster. The cost of creating and destroying an object is now roughly cut in half by AArch64. Tagged pointers also make for a nice performance win as well as reducing memory usage.
These are just some of the performance increases that have made huge differences in the A7. But hey think what you want. The proof is in the testing and the A7 is clearly the fastest mobile processor on the market.
Oh another one of you 4 Gig people. This theory has already been discussed ad nauseum, there are huge benefits of a 64 bit processor the least of which is addressing more that 4 Gig of memory. Anand La Shimpi says so as do others more knowledgeable than you. You should stop reading the android blogs because this is there favorite line. Heck even Qualcomm retracted there Senior VP idiots statement as being false for claiming the same thing and demoted him. There a chip maker they know more about it than you or I. And for your info all 1st party apps on iOS are full 64 bit.
To requote what has been quoted for months now since someone in this thread can't be bothered to do any research before posting...
"Apple isn’t very focused on delivering a larger memory address space today however. As A64 is a brand new ISA, there are other benefits that come along with the move. Similar to the x86-64 transition, the move to A64 comes with an increase in the number of general purpose registers. ARMv7 had 15 general purpose registers (and 1 register for the program counter), while ARMv8/A64 now has 31 that are each 64-bits wide. All 31 registers are accessible at all times. Increasing the number of architectural registers decreases register pressure and can directly impact performance. The doubling of the register space with x86-64 was responsible for up to a 10% increase in performance.
The original ARM architecture made all instructions conditional, which had a huge impact on the instruction space. The number of conditional instructions is far more limited in ARMv8/A64.
The move to ARMv8 also doubles the number of FP/NEON registers (from 16 to 32) as well as widens all of them registers to 128-bits (up from 64-bits). Support for 128-bit registers can go a long way in improving SIMD performance. Whereas simply doubling register count can provide moderate increases in performance, doubling the size of each register can be far more significant given the right workload. There are also new advanced SIMD instructions that are a part of ARMv8. Double precision SIMD FP math is now supported among other things.
ARMv8 also adds some new cryptographic instructions for hardware acceleration of AES and SHA1/SHA256 algorithms. These hardware AES/SHA instructions have the potential for huge increases in performance, just like we saw with the introduction of AES-NI on Intel CPUs a few years back. Both the new advanced SIMD instructions and AES/SHA instructions are really designed to enable a new wave of iOS apps."
"It's also important to point out the things that "64-bit" does not refer to, as there's a lot of confusion in this area as well. In particular, "64-bit" does not include:
Physical RAM address size. The number of bits used to actually talk to RAM (and therefore the amount of RAM the hardware can support) is decoupled from the question of CPU bitness. ARM CPUs have ranged from 26 bits to 40 bits, and this can be changed independently from the rest.
Data bus size. The amount of data fetched from RAM or cache is likewise decoupled. Individual CPU instructions may request a certain amount of data, but the amount of data actually fetched can be independent, either by splitting the fetch into smaller parts, or fetching more than is necessary. The iPhone 5 already fetches data from memory in 64-bit chunks, and chunk sizes of up to 192 bits exist in the PC world.
Anything related to floating-point. FPU register size and internal design is independent, and ARM CPUs have had 64-bit FPU registers since well before ARM64.
[…]
Adding it all together, it's a pretty big win. My casual benchmarking indicates that basic object creation and destruction takes about 380ns on a 5S running in 32-bit mode, while it's only about 200ns when running in 64-bit mode. If any instance of the class has ever had a weak reference and an associated object set, the 32-bit time rises to about 480ns, while the 64-bit time remains around 200ns for any instances that were not themselves the target.
In short, the improvements to Apple's runtime make it so that object allocation in 64-bit mode costs only 40-50% of what it does in 32-bit mode. If your app creates and destroys a lot of objects, that's a big deal.
[…]
Apple took advantage of the transition to make some changes of their own. The biggest change is an inline retain count, which eliminates the need to perform a costly hash table lookup for retain and release operations in the common case. Since those operations are so common in most Objective-C code, this is a big win. Per-object resource cleanup flags make object deallocation quite a bit faster in certain cases. All in all, the cost of creating and destroying an object is roughly cut in half. Tagged pointers also make for a nice performance win as well as reduced memory use.
ARM64 is a welcome addition to Apple's hardware. We all knew it would happen eventually, but few expected it this soon. It's here now, and it's great."
Note that AnandTech shows an improvement in AES by 825% (must access link to see results) simply by running as 64-bit. How exactly not a benefit? As I've stated before I think it's worth considering that Touch ID wouldn't be possible without A64's improved crypto and phenomenal AES and SHA speed performance.
The original ARM architecture made all instructions conditional, which had a huge impact on the instruction space. The number of conditional instructions is far more limited in ARMv8/A64.
Comments
Oh another one of you 4 Gig people. This theory has already been discussed ad nauseum, there are huge benefits of a 64 bit processor the least of which is addressing more that 4 Gig of memory. Anand La Shimpi says so as do others more knowledgeable than you. You should stop reading the android blogs because this is there favorite line. Heck even Qualcomm retracted there Senior VP idiots statement as being false for claiming the same thing and demoted him. There a chip maker they know more about it than you or I. And for your info all 1st party apps on iOS are full 64 bit.
Sigh....
The real performance increase in 64 bit is for applications that handle large data arrays; so things like CAD/CAE-like programs, data managers, programs to work with digital media and applied scientific applications, so yes iphone 5s with it current specs would see quite an increase in performance when executing such software.
So yes I agree with you there are other huge benefits of a 64 bit processor besides addressing 4 gigs of memory, but those benefits really only apply to large data array software. Show me that the average user of an iphone 5s or an ipad primarily or at the very least plan on using such software for a good portion of their time when using those devices.
If you had, you wouldn’t be repeating this worthless nonsense.
Completely and utterly incorrect. Educate yourself. I’ll repeat it until you do.
Huh? Address bus is used solely for addressing memory and only memory. So the only benefit of a 64 bit address bus is when you have more than 4 GB of memory. If you are not running software that uses more than 2 GB in a single thread or the entire system is not using more than 4 GB of memory there is no performance benefit to using a 64 bit address bus. What else is the address bus used for Tallest Skil?
The only time a 64 bit cpu provides any meaningful performance increase when you don't have more than 4 GB or at the very least 2 GB to run a single thread is when you are executing or running large data array software that takes advantage of the additional registers. Show me that the average user or even significant portion of iphone or ipad users is using such software or plans to use it.
Address bus is used solely for addressing memory and only memory. So the only benefit of a 64 bit address bus is when you have more than 4 GB of memory. If you are not running software that uses more than 2 GB in a single thread or the entire system is not using more than 4 GB of memory there is no performance benefit to using a 64 bit address bus. The only time a 64 bit cpu provides any meaningful performance increase when you don’t have more than 4 GB…
So why does the iPhone 5S show up to a 25% speed improvement on itself between otherwise identical 32-bit and 64-bit compiled apps?
Perhaps his IQ only goes up to 32? More likely to have a single-digit IQ though.
This link may answer your question...
http://www.patentlyapple.com/patently-apple/2013/09/apples-fingerprint-sensor-patent-claims-more-patent-graphics.html#more
Even though it's been mentioned he has completely ignored the new AArch64 ISA.
So why does the iPhone 5S show up to a 25% speed improvement on itself between otherwise identical 32-bit and 64-bit compiled apps?
Key word is up to, what is the Mode? Or most common speed improvement? Large data array software such as encryption, encoding, decoding, video editing software will give you those types of speed improvements if not higher speed improvements. Also is it comparing or are you comparing it to the 32 bit version of A7 processor? Lets take for example the most commonly used software, the Internet browser.
Internet Explorer 10 and 11 64 bit is only around 1-2% faster on benchmarks such as sunspider, kraken, v8 benchmarks compared to its 32 bit version. That is hardly a meaningful improvement.
So which software or apps is giving you those 25% speed improvements?
Keep this in mind: higher core clock speeds, faster LPDDR3 memory and quad-core PowerVR graphics chip — all independent improvements that the A7 makes, how much of those 25% speed improvements are attributed to that and not the 64 bit.
What is the most commonly used software that an iphone or ipad user uses? I have a feeling that it is not large data array software.
If you look at ST88 post you will see that according to Futuremark Benchmark the 64 bit version on average is only 7% faster, not enough to move it around on the rankings for fastest mobile phone.
Here is what Futuremark said:
64-bit is not the answer
We started our investigation by compiling a 64-bit version of 3DMark, since the version available in the App Store is 32-bit in order to be compatible with older Apple devices.
Once we had a 64-bit build we tested an iPhone 5s under controlled conditions in our Test Lab. As expected, we found that the results from the 64-bit version were similar to the 32-bit version, with only a 7 percent improvement.
http://www.futuremark.com/pressreleases/understanding-3dmark-results-from-the-apple-iphone-5s-and-ipad-air
So as I have said before large data array software such as encoding, encryption will give you those types of large speed improvements but if you are talking about the day to day software that the average owner of an iphone or ipad user uses than chances are no it doe not make any real meaningful performance increase.
Perhaps his IQ only goes up to 32? More likely to have a single-digit IQ though.
Brilliant response! Did you know that Tallest Skil literally thought address bus did more than address memory.
Even though it's been mentioned he has completely ignored the new AArch64 ISA.
No I haven't.
Where did you address the ISA?
Where did you address the ISA?
You can't be serious with that question. Talking about the A7 processor is addressing the ISA.
Yes, I'm serious. Where did you address AArch64? Where did you discredit the new ISA by saying it offers no benefits to the OS or SW?
PS: In your previous post you're compressing a bunch of different types of 64-bit items into one poorly contrived argument. You can't use IE 11 as being 1% faster as proof that the new ISA in AArch64 offers no benefit.
Yes, I'm serious. Where did you address AArch64? Where did you discredit the new ISA by saying it offers no benefits to the OS or SW?
PS: In your previous post you're compressing a bunch of different types of 64-bit items into one poorly contrived argument. You can't use IE 11 as being 1% faster as proof that the new ISA in AArch64 offers no benefit.
No you can't be serious.
So you're still not going to answer why you think the new ISA is just marketing hype invented by Apple?
So you're still not going to answer why you think the new ISA is just marketing hype invented by Apple?
Already answered that question go back and read and the answer is not that simple.
No, I made no comment as to that. Try again without putting words in others’ mouths.
No, no you didn't. You discredited everything that comes with AArch64 and made sweeping ignorant generalizations and even made foolish comments about RAM capacity and CPU bitness.
PS: I checked your posts. You didn't use ISA, AArch64, or the colloquial ARM64 once in your comments.
Did you know you're incorrectly equating the number of bits with the address bus with number of bits used in other parts of the system.
Even though it's been mentioned he has completely ignored the new AArch64 ISA.
yes he has. He overlooks that AArch64 doubled the number of registers over 32bit ARM. 32-bit ARM provides 16 integer registers, of which one is a dedicated program counter, two more are give over to a stack pointer and link register, and the other 13 are available for general use. With AArch 64 there are 32 integer registers, with a dedicated zero register, dedicated link register and dedicated frame pointer register. One further register is dedicated to the platform leaving 28 general purpose registers.
AArch 64 also increases the number of floating point registers as well. AArch 32 has 32 32 bit floating point registers. Which can overlap into various sized of floating points as needed. AArch 64 simplifies this and has 32 128 bit floating point registers that can be used for smaller amounts of data, but are dedicated and do not overlap which is a big performance win. With AArch 64 Physical RAM address size is decouple from CPU fitness.Data Bus size the amount of data fetched from RAM is likewise decoupled. A CPU instruction may request a certain size of data but the amount can be independentlycontrolled by either fetching in smaller amounts or fetching more than is necessary.Also a revised and streamlined the instruction set for a big performance gain. Also apple took advantage of AArch64 in a big way with an inline retain count, which eliminates the need to perform costly hash table lookup for retain and release operations in the common case. Those operations are very common in most Objective-C code and is a big performance win. Per-object resource cleanup flags make object deallocation a lot faster. The cost of creating and destroying an object is now roughly cut in half by AArch64. Tagged pointers also make for a nice performance win as well as reducing memory usage.
These are just some of the performance increases that have made huge differences in the A7. But hey think what you want. The proof is in the testing and the A7 is clearly the fastest mobile processor on the market.
To requote what has been quoted for months now since someone in this thread can't be bothered to do any research before posting...
The original ARM architecture made all instructions conditional, which had a huge impact on the instruction space. The number of conditional instructions is far more limited in ARMv8/A64.
The move to ARMv8 also doubles the number of FP/NEON registers (from 16 to 32) as well as widens all of them registers to 128-bits (up from 64-bits). Support for 128-bit registers can go a long way in improving SIMD performance. Whereas simply doubling register count can provide moderate increases in performance, doubling the size of each register can be far more significant given the right workload. There are also new advanced SIMD instructions that are a part of ARMv8. Double precision SIMD FP math is now supported among other things.
ARMv8 also adds some new cryptographic instructions for hardware acceleration of AES and SHA1/SHA256 algorithms. These hardware AES/SHA instructions have the potential for huge increases in performance, just like we saw with the introduction of AES-NI on Intel CPUs a few years back. Both the new advanced SIMD instructions and AES/SHA instructions are really designed to enable a new wave of iOS apps."
and…
Physical RAM address size. The number of bits used to actually talk to RAM (and therefore the amount of RAM the hardware can support) is decoupled from the question of CPU bitness. ARM CPUs have ranged from 26 bits to 40 bits, and this can be changed independently from the rest.
Data bus size. The amount of data fetched from RAM or cache is likewise decoupled. Individual CPU instructions may request a certain amount of data, but the amount of data actually fetched can be independent, either by splitting the fetch into smaller parts, or fetching more than is necessary. The iPhone 5 already fetches data from memory in 64-bit chunks, and chunk sizes of up to 192 bits exist in the PC world.
Anything related to floating-point. FPU register size and internal design is independent, and ARM CPUs have had 64-bit FPU registers since well before ARM64.
[…]
Adding it all together, it's a pretty big win. My casual benchmarking indicates that basic object creation and destruction takes about 380ns on a 5S running in 32-bit mode, while it's only about 200ns when running in 64-bit mode. If any instance of the class has ever had a weak reference and an associated object set, the 32-bit time rises to about 480ns, while the 64-bit time remains around 200ns for any instances that were not themselves the target.
In short, the improvements to Apple's runtime make it so that object allocation in 64-bit mode costs only 40-50% of what it does in 32-bit mode. If your app creates and destroys a lot of objects, that's a big deal.
[…]
Apple took advantage of the transition to make some changes of their own. The biggest change is an inline retain count, which eliminates the need to perform a costly hash table lookup for retain and release operations in the common case. Since those operations are so common in most Objective-C code, this is a big win. Per-object resource cleanup flags make object deallocation quite a bit faster in certain cases. All in all, the cost of creating and destroying an object is roughly cut in half. Tagged pointers also make for a nice performance win as well as reduced memory use.
ARM64 is a welcome addition to Apple's hardware. We all knew it would happen eventually, but few expected it this soon. It's here now, and it's great."
Note that AnandTech shows an improvement in AES by 825% (must access link to see results) simply by running as 64-bit. How exactly not a benefit? As I've stated before I think it's worth considering that Touch ID wouldn't be possible without A64's improved crypto and phenomenal AES and SHA speed performance.
Wouldn't that be ARMv7 here?
Apple's A7 uses ARMv8.