I'm sure we will see a 64GB version eventually depending on it's popularity.
I still feel like it's a typo or something. I guess that's still like 6HD movies on 32GB. It's still pretty surprising to me that they'd spec it like that. Maybe they've got info that most of the target consumers keep their stuff in the cloud...
I guess if anyone would know how people are using their devices...
There are two Note 4 reviews on Anandtech, one based on the Exynos model the other Qualcomm's 805, though neither overall as fast as the A8 the 805 did much better, especially in the area of graphics performance, where the Exynos failed miserably.
Here are some of the Note 4 GPU benchmarks based on the 805
There is no doubt the A8 is a faster CPU but not by much and defiantly isn't a POS like you are so adamantly trying to convey. The Nexus 6 will also be faster than the Note 4 as it is not only running Android 5 but won't have the huge overhead that Touchwiz is so famously known for. Now that is a POS. It's fine that you want everyone to know just how fast the iPhone is in comparison to the competition but you don't need to call their chips garbage in the process, simply to make a point. Qualcomm is one of the best ARM chip manufactures in the world and their products are outstanding. Not only that but their an American company with plants in the US, I would think you would want to support them just a little bit more then you are. Also if anything the 805 shows just how fast Qualcomms upcoming 810 64Bit processor will be and if it's faster than A8 and it's looking that most likely will be, what are you going to say then. No, it's better to just enjoy using the products you have without feeling the need to criticize others simply because of some arbitrary benchmarks that you most likely don't even understand anyway.
No, the 805 is a complete POS. Twice the cores and 90% faster clock and can't even beat the A8.
Wow, too much misinformation going on, so instead of quoting everyone I'll just sum up here:
- Android Java apps won't need to be re-compiled to work on Android L when it's on a 64bit device. However, they will NOT take full advantage of the 64bit processor BECAUSE they're written in Java. Java apps running on a 64bit processor are like 32bit iOS apps running on an A7/8 - they work without re-compiling, they're faster than they were before, but they don't utilize the processor to its fullest.
- There are three ways to run an existing app on a 64bit processor. 1) Leave it alone and it will run fine as is. 2) Re-compile the app as 64bit and let the compiler make decisions as to optimizations giving you a 64bit native app. 3) Re-write the app and modify your code as necessary to SPECIFICALLY take advantage of the 64bit processor (for example, modifying algorithms to take advantage of the registers). This is the best way to create a 64bit app.
- iOS 7 wasn't just re-compiled. It was re-written. This is why iOS 7 had little in the way of features. The engineers at Apple would have spent considerable resources just doing the 64bit conversion. It's also doubtful that Android L was simply re-compiled. There's too much going on in an OS to trust it to a simple re-compile. However, higher level apps could be re-written or just re-compiled, there's really no way to know for sure. I also doubt Google apps (or Apple apps) were re-compiled.
- Qualcomm has already stated the 810 will be based on ARM A57 cores. We already know that even the A7 from Apple outperforms the A57 from ARM. So there's no possible way for the 810 to be some kind of powerhouse SoC on the CPU side. It was already outdated even before it was announced. The 810 might have a great GPU, but on Android that's less useful than a good CPU (which is responsible for most of the responsiveness of your software).
Apple did that to themselves, they have been the ones who have consistently wanted a thin and flat device, and finally realised you can't always do it
er, did what? it's a complete non-issue. when I used phone naked I rarely if ever operate flat on a table. but when I tried it, it didn't rock - unless tapping the top-left corner, but that area was unneeded due to left-edge swiping for Back. lastly, I ended up putting a case on it anyway.
3) Re-write the app and modify your code as necessary to SPECIFICALLY take advantage of the 64bit processor (for example, modifying algorithms to take advantage of the registers). This is the best way to create a 64bit app.
The "best way" is the easiest way that satisfies the software requirements. There is no need to re-architect Waze so it fully utilizes a 64-bit platform until it's falls into their normal product cycle, because the app performs fine as it is, with communications providing the only significant lag. Something like GTA: Vice City, yeah, rewrite the core engine.
- Qualcomm has already stated the 810 will be based on ARM A57 cores. We already know that even the A7 from Apple outperforms the A57 from ARM. So there's no possible way for the 810 to be some kind of powerhouse SoC on the CPU side. It was already outdated even before it was announced. The 810 might have a great GPU, but on Android that's less useful than a good CPU (which is responsible for most of the responsiveness of your software).
Sorry, when compared to the A7, what are you talking about, the only ARM processor currently available that even utilizes the A57 cores is the AMD Opteron A1100 and that's a server CPU. Freescale has one in the works as well, also for servers but it won't be available till next year. Outdated before released, yea, I think you need to do some research on the A57, not only is it ARM's latest and greatest but there are no current benchmarks pitting an A7 against a A57, at least non that I've seen and I've looked extensively as I am very interested in the Opteron. So which A57 CPU are you referring too, there's only one that could have possibly been used and mark my words the Opteron is faster, much faster but that's a server CPU and not a mobile? Also the Qualcomm will have almost double the amount of memory bandwidth as the A8 as their using a 2x 64Bit memory controller vs. 2x 32Bit found in the A8 and DDR4. I think you've mistaken the 810 for something else and please stop referring the Android programming language as Java.
Sorry, when compared to the A7, what are you talking about, the only ARM processor currently available that even utilizes the A57 cores is the AMD Opteron A1100 and that's a server CPU. Freescale has one in the works as well, also for servers but it won't be available till next year. Outdated before released, yea, I think you need to do some research on the A57, not only is it ARM's latest and greatest but there are no current benchmarks pitting an A7 against a A57, at least non that I've seen and I've looked extensively as I am very interested in the Opteron. So which A57 CPU are you referring too, there's only one that could have possibly been used and mark my words the Opteron is faster, much faster but that's a server CPU and not a mobile? Also the Qualcomm will have almost double the memory bandwidth as the A8 as their using a 2x 64Bit memory controller vs. 2x 32Bit found in the A8 and DDR4. I think you've mistaken the 810 for something else.
Now I know you're just trolling when you post complete and utter BS like this. On my way out of the office, but I'll explain in great detail why the A57 is garbage and why you are so far off base it's pathetic.
Now I know you're just trolling when you post cmplete and utter BS like this. On my way out of the office, but I'll explain in great detail why the A57 is garbage and why you are so far off base it's pathetic.
Yes, yes, the Note 4 and I'm sure your going home to type up, excuse me copy and paste what reviewers have said about it, though technically the Exynos 5433 is a based on the ARM A57 design it is not a true implementation of it. Not only was the memory controller hobbled, 64Bit disabled but it doesn't use DDR4 memory, along with a a slue of architectural differences that make it not a true A57. It was a half assed rushed job from Samsung and cannot be compared to Qualcomms 810 or AMD's Opteron in any way. We are talking about 64bit chips here, not 32bit. So I'm sorry if you think I was purposely leaving out the crap that is the Exynos 5433 but that is not and I repeat not what Qualcomm is going to do with their chip. It will be a true 64bit chip, it will have DDR4 and it will have a true 64bit memory controller. Will it be faster than Apples A8, I think it has a good chance but I don't care.
Related: there was an GooglePlay banner that was seen today with the Galaxy S5 in a stream next to Moto G and HTC One M8. It was a mockup with the home icon (which hopefully would go away). I think the availability of a non-touchwiz S5 would calm people down regarding the size and/or price of the Nexus 6.
If it wasn't for Samsung's software, and its price, I'd probably like that phone. I'm not a big Samsung fan, but they haven't poisoned my dogs yet or anything. I mostly just don't want any skinning on Android.
Yes, yes, the Note 4 and I'm sure your going home to type up, excuse me copy and paste what reviewers have said about it, though technically the Exynos 5433 is a based on the ARM A57 design it is not a true implementation of it. Not only was the memory controller hobbled, 64Bit disabled but it doesn't use DDR4 memory, along with a a slue of architectural differences that make it not a true A57. It was a half assed rushed job from Samsung and cannot be compared to Qualcomms 810 or AMD's Opteron in any way. We are talking about 64bit chips here, not 32bit. So I'm sorry if you think I was purposely leaving out the crap that is the Exynos 5433 but that is not and I repeat not what Qualcomm is going to do with their chip. It will be a true 64bit chip, it will have DDR4 and it will have a true 64bit memory controller. Will it be faster than Apples A8, I think it has a good chance but I don't care.
Sorry, but I don't need to use the Exynos 5433 (even though that's also a garbage processor). I'm going to use material from ARM themselves which discusses the performance of the A57 in 32 and 64 bit modes.
This is ARM's reference page for the A57. Note how their simple bar chart shows the A57 about 20-40% faster. But this isn't a good source of data. I only posted it up as it's an earlier prediction ARM made (and also because they have some other useful information about the A57 here).
This is Geekbench 3 data straight from ARM comparing the performance of the A15, A53 and A57. These charts are important for a couple of reasons. For starters, this is ARM talking, not Samsung or Qualcomm or some random internet poster trying to predict performance of a processor based on something as stupid as memory bus width (**cough** you know who **cough**). Secondly, ARM normalizes their charts to 1.0, which is standard practice when comparing processors that run the same architecture. BTW, Geekbench also does this. In this case, I'm going to normalize all the numbers to 1.0GHz to make comparing processors easier.
Looking at the bottom chart, ARM claims the A57 in 32bit mode is 1.20x as fast as an A15 at the same clock. In 64bit mode the A57 is 1.45x as fast. The difference in performance between 32bit and 64bit is almost exactly 20%. This makes a lot of sense as 64bit code is usually only about 20% faster than 32bit code for normal functions (excluding things like encoding or encryption where it's much faster).
So what about the A15? Thankfully, there are lots of A15 processors out there for us to benchmark.
Samsung Exynos 5420 uses A15 cores running at 1.9GHz. The K1 Tegra 32bit processor also uses the A15 cores. So how do they perform?
Exynos scores 888 at 1.90GHz. This gives us a figure of 467 when normalized to 1.0GHz.
Tegra K1 scores 1088 at 2.128GHz. This gives is a figure of 511 when normalized to 1.0GHz.
For sake of argument, I'm going to use the fastest ever tested A15 from the Tegra K1.
Now based on ARM's own numbers, an A57 in 32bit is 1.2x and an A57 in 64bit is 1.45x as fast as an A15.
That means the A57 scores 613 in 32bit mode and 741 in 64bit mode. According to ARM themselves.
What about the Exynos 5433? Well, it scores 1128 at 1.9GHz. Converted to 1.0GHz and it scores...wait for it.....594. Compared to the 613 according to ARM it appears the Exynos isn't really the POS I stated, in that it's very close to ARM's number (in 32bit mode).
So there goes your (and mine) first argument, that the Exynos is garbage and we should not use its benchmark results. It appears the Exynos A57 performs about as well as ARM says it should.
Now what about the A7/A8? The A7 normalized to 1.0GHz scores 1076 in 64bit mode and the A8 scores 1142.
I'm not sure what you think is "impressive", but I'd say a processor that scores 1076/1142 is a hell of a lot faster than one that scores 741 (according to ARM).
Now I'd LOVE for you to explain why you think the Qualcomm 810 is going to somehow magically do better than what ARM says the A57 is capable of when the Exynos 5433 performs as ARM predicted and Qualcomm isn't making a custom version of the A57 (like they did with Krait).
In this case, I'm going to normalize all the numbers to 1.0GHz to make comparing processors easier.
...
Exynos scores 888 at 1.90GHz. This gives us a figure of 467 when normalized to 1.0GHz.
Tegra K1 scores 1088 at 2.128GHz. This gives is a figure of 511 when normalized to 1.0GHz.
For sake of argument, I'm going to use the fastest ever tested A15 from the Tegra K1.
Now based on ARM's own numbers, an A57 in 32bit is 1.2x and an A57 in 64bit is 1.45x as fast as an A15.
That means the A57 scores 613 in 32bit mode and 741 in 64bit mode. According to ARM themselves.
What about the Exynos 5433? Well, it scores 1128 at 1.9GHz. Converted to 1.0GHz and it scores...wait for it.....594. Compared to the 613 according to ARM it appears the Exynos isn't really the POS I stated, in that it's very close to ARM's number (in 32bit mode).
...
Now what about the A7/A8? The A7 normalized to 1.0GHz scores 1076 in 64bit mode and the A8 scores 1142.
I'm not sure what you think is "impressive", but I'd say a processor that scores 1076/1142 is a hell of a lot faster than one that scores 741 (according to ARM).
How do you justify normalizing to the same clock speed when you are comparing different architectures? Some architectures (like Qualcomm) may be simply designed to run at higher clock speeds to accomplish the same work that other architectures (like the A7/A8) would perform at lower clock speeds. It's more meaningful to instead normalize by overall power consumption, since it could be the case (which needs to be checked of course) that the higher clock speeds of the Qualcomm are partly counterbalanced by a simpler design, or equivalently that the lower clock speeds of the A7 require more transistors to achieve.
Comments
I'm sure we will see a 64GB version eventually depending on it's popularity.
I still feel like it's a typo or something. I guess that's still like 6HD movies on 32GB. It's still pretty surprising to me that they'd spec it like that. Maybe they've got info that most of the target consumers keep their stuff in the cloud...
I guess if anyone would know how people are using their devices...
AD Hominens. Because strawman was just not good enough of a logical argument.
No one cares 2 hoots for your apologies. I don't earn money for getting apologies from trolls.
But you were wrong. If you were a better grown up you'd just accept the inaccuracy in your comment and move on.
There are two Note 4 reviews on Anandtech, one based on the Exynos model the other Qualcomm's 805, though neither overall as fast as the A8 the 805 did much better, especially in the area of graphics performance, where the Exynos failed miserably.
Here are some of the Note 4 GPU benchmarks based on the 805
There is no doubt the A8 is a faster CPU but not by much and defiantly isn't a POS like you are so adamantly trying to convey. The Nexus 6 will also be faster than the Note 4 as it is not only running Android 5 but won't have the huge overhead that Touchwiz is so famously known for. Now that is a POS. It's fine that you want everyone to know just how fast the iPhone is in comparison to the competition but you don't need to call their chips garbage in the process, simply to make a point. Qualcomm is one of the best ARM chip manufactures in the world and their products are outstanding. Not only that but their an American company with plants in the US, I would think you would want to support them just a little bit more then you are. Also if anything the 805 shows just how fast Qualcomms upcoming 810 64Bit processor will be and if it's faster than A8 and it's looking that most likely will be, what are you going to say then. No, it's better to just enjoy using the products you have without feeling the need to criticize others simply because of some arbitrary benchmarks that you most likely don't even understand anyway.
No, the 805 is a complete POS. Twice the cores and 90% faster clock and can't even beat the A8.
Wow, too much misinformation going on, so instead of quoting everyone I'll just sum up here:
- Android Java apps won't need to be re-compiled to work on Android L when it's on a 64bit device. However, they will NOT take full advantage of the 64bit processor BECAUSE they're written in Java. Java apps running on a 64bit processor are like 32bit iOS apps running on an A7/8 - they work without re-compiling, they're faster than they were before, but they don't utilize the processor to its fullest.
- There are three ways to run an existing app on a 64bit processor. 1) Leave it alone and it will run fine as is. 2) Re-compile the app as 64bit and let the compiler make decisions as to optimizations giving you a 64bit native app. 3) Re-write the app and modify your code as necessary to SPECIFICALLY take advantage of the 64bit processor (for example, modifying algorithms to take advantage of the registers). This is the best way to create a 64bit app.
- iOS 7 wasn't just re-compiled. It was re-written. This is why iOS 7 had little in the way of features. The engineers at Apple would have spent considerable resources just doing the 64bit conversion. It's also doubtful that Android L was simply re-compiled. There's too much going on in an OS to trust it to a simple re-compile. However, higher level apps could be re-written or just re-compiled, there's really no way to know for sure. I also doubt Google apps (or Apple apps) were re-compiled.
- Qualcomm has already stated the 810 will be based on ARM A57 cores. We already know that even the A7 from Apple outperforms the A57 from ARM. So there's no possible way for the 810 to be some kind of powerhouse SoC on the CPU side. It was already outdated even before it was announced. The 810 might have a great GPU, but on Android that's less useful than a good CPU (which is responsible for most of the responsiveness of your software).
Over ten percent of this topic's content.
You can almost smell the desperation, can't you?
er, did what? it's a complete non-issue. when I used phone naked I rarely if ever operate flat on a table. but when I tried it, it didn't rock - unless tapping the top-left corner, but that area was unneeded due to left-edge swiping for Back. lastly, I ended up putting a case on it anyway.
non-issue in either use case.
what a f'ing stupid codename for an OS btw
Yeah, "Lemon Bon Bon" was already taken...
(I wonder what happened to Lemon, anyway?)
3) Re-write the app and modify your code as necessary to SPECIFICALLY take advantage of the 64bit processor (for example, modifying algorithms to take advantage of the registers). This is the best way to create a 64bit app.
The "best way" is the easiest way that satisfies the software requirements. There is no need to re-architect Waze so it fully utilizes a 64-bit platform until it's falls into their normal product cycle, because the app performs fine as it is, with communications providing the only significant lag. Something like GTA: Vice City, yeah, rewrite the core engine.
- Qualcomm has already stated the 810 will be based on ARM A57 cores. We already know that even the A7 from Apple outperforms the A57 from ARM. So there's no possible way for the 810 to be some kind of powerhouse SoC on the CPU side. It was already outdated even before it was announced. The 810 might have a great GPU, but on Android that's less useful than a good CPU (which is responsible for most of the responsiveness of your software).
Sorry, when compared to the A7, what are you talking about, the only ARM processor currently available that even utilizes the A57 cores is the AMD Opteron A1100 and that's a server CPU. Freescale has one in the works as well, also for servers but it won't be available till next year. Outdated before released, yea, I think you need to do some research on the A57, not only is it ARM's latest and greatest but there are no current benchmarks pitting an A7 against a A57, at least non that I've seen and I've looked extensively as I am very interested in the Opteron. So which A57 CPU are you referring too, there's only one that could have possibly been used and mark my words the Opteron is faster, much faster but that's a server CPU and not a mobile? Also the Qualcomm will have almost double the amount of memory bandwidth as the A8 as their using a 2x 64Bit memory controller vs. 2x 32Bit found in the A8 and DDR4. I think you've mistaken the 810 for something else and please stop referring the Android programming language as Java.
Yeah, "Lemon Bon Bon" was already taken...
(I wonder what happened to Lemon, anyway?)
He soured on posting here due to bitterness.
No, the 805 is a complete POS. Twice the cores and 90% faster clock and can't even beat the A8.
Speed is not the only factor. Keep in mind that even at $650, the Nexus 6's price is still $260* less than the iPhone 6 Plus.
*approx. splitting the difference between $859 and $969 since there's no 32GB iPhone 6 Plus
No, the 805 is a complete POS. Twice the cores and 90% faster clock and can't even beat the A8.
Ugh, do your homework before posting this kind of gibberish and a faster clock doesn't always relate to more speed, again read up.
Ad hominems are things like calling people trolls, which you've done twice. What I did was insinuate immaturity, and you've now borne that out twice.
I did discuss my error in my reply to you. It's the part you edited out of your quote.
Get some rest. Tomorrow will be a big day at school. Hopefully your teacher will cover irony.
Sorry, when compared to the A7, what are you talking about, the only ARM processor currently available that even utilizes the A57 cores is the AMD Opteron A1100 and that's a server CPU. Freescale has one in the works as well, also for servers but it won't be available till next year. Outdated before released, yea, I think you need to do some research on the A57, not only is it ARM's latest and greatest but there are no current benchmarks pitting an A7 against a A57, at least non that I've seen and I've looked extensively as I am very interested in the Opteron. So which A57 CPU are you referring too, there's only one that could have possibly been used and mark my words the Opteron is faster, much faster but that's a server CPU and not a mobile? Also the Qualcomm will have almost double the memory bandwidth as the A8 as their using a 2x 64Bit memory controller vs. 2x 32Bit found in the A8 and DDR4. I think you've mistaken the 810 for something else.
Now I know you're just trolling when you post complete and utter BS like this. On my way out of the office, but I'll explain in great detail why the A57 is garbage and why you are so far off base it's pathetic.
Now, this Nexus 6 will hardly fit in a pocket.
Now I know you're just trolling when you post cmplete and utter BS like this. On my way out of the office, but I'll explain in great detail why the A57 is garbage and why you are so far off base it's pathetic.
Yes, yes, the Note 4 and I'm sure your going home to type up, excuse me copy and paste what reviewers have said about it, though technically the Exynos 5433 is a based on the ARM A57 design it is not a true implementation of it. Not only was the memory controller hobbled, 64Bit disabled but it doesn't use DDR4 memory, along with a a slue of architectural differences that make it not a true A57. It was a half assed rushed job from Samsung and cannot be compared to Qualcomms 810 or AMD's Opteron in any way. We are talking about 64bit chips here, not 32bit. So I'm sorry if you think I was purposely leaving out the crap that is the Exynos 5433 but that is not and I repeat not what Qualcomm is going to do with their chip. It will be a true 64bit chip, it will have DDR4 and it will have a true 64bit memory controller. Will it be faster than Apples A8, I think it has a good chance but I don't care.
It is more like the Note 4 is hobbled to samsung.
Related: there was an GooglePlay banner that was seen today with the Galaxy S5 in a stream next to Moto G and HTC One M8. It was a mockup with the home icon (which hopefully would go away). I think the availability of a non-touchwiz S5 would calm people down regarding the size and/or price of the Nexus 6.
If it wasn't for Samsung's software, and its price, I'd probably like that phone. I'm not a big Samsung fan, but they haven't poisoned my dogs yet or anything. I mostly just don't want any skinning on Android.
Get some rest. Tomorrow will be a big day at school. Hopefully your teacher will cover irony.
lmao
Yes, yes, the Note 4 and I'm sure your going home to type up, excuse me copy and paste what reviewers have said about it, though technically the Exynos 5433 is a based on the ARM A57 design it is not a true implementation of it. Not only was the memory controller hobbled, 64Bit disabled but it doesn't use DDR4 memory, along with a a slue of architectural differences that make it not a true A57. It was a half assed rushed job from Samsung and cannot be compared to Qualcomms 810 or AMD's Opteron in any way. We are talking about 64bit chips here, not 32bit. So I'm sorry if you think I was purposely leaving out the crap that is the Exynos 5433 but that is not and I repeat not what Qualcomm is going to do with their chip. It will be a true 64bit chip, it will have DDR4 and it will have a true 64bit memory controller. Will it be faster than Apples A8, I think it has a good chance but I don't care.
Sorry, but I don't need to use the Exynos 5433 (even though that's also a garbage processor). I'm going to use material from ARM themselves which discusses the performance of the A57 in 32 and 64 bit modes.
http://www.arm.com/products/processors/cortex-a/cortex-a57-processor.php
This is ARM's reference page for the A57. Note how their simple bar chart shows the A57 about 20-40% faster. But this isn't a good source of data. I only posted it up as it's an earlier prediction ARM made (and also because they have some other useful information about the A57 here).
http://www.anandtech.com/show/7995/arm-shares-updated-cortex-a53a57-performance-expectations
This is Geekbench 3 data straight from ARM comparing the performance of the A15, A53 and A57. These charts are important for a couple of reasons. For starters, this is ARM talking, not Samsung or Qualcomm or some random internet poster trying to predict performance of a processor based on something as stupid as memory bus width (**cough** you know who **cough**). Secondly, ARM normalizes their charts to 1.0, which is standard practice when comparing processors that run the same architecture. BTW, Geekbench also does this. In this case, I'm going to normalize all the numbers to 1.0GHz to make comparing processors easier.
Looking at the bottom chart, ARM claims the A57 in 32bit mode is 1.20x as fast as an A15 at the same clock. In 64bit mode the A57 is 1.45x as fast. The difference in performance between 32bit and 64bit is almost exactly 20%. This makes a lot of sense as 64bit code is usually only about 20% faster than 32bit code for normal functions (excluding things like encoding or encryption where it's much faster).
So what about the A15? Thankfully, there are lots of A15 processors out there for us to benchmark.
Samsung Exynos 5420 uses A15 cores running at 1.9GHz. The K1 Tegra 32bit processor also uses the A15 cores. So how do they perform?
Exynos scores 888 at 1.90GHz. This gives us a figure of 467 when normalized to 1.0GHz.
Tegra K1 scores 1088 at 2.128GHz. This gives is a figure of 511 when normalized to 1.0GHz.
For sake of argument, I'm going to use the fastest ever tested A15 from the Tegra K1.
Now based on ARM's own numbers, an A57 in 32bit is 1.2x and an A57 in 64bit is 1.45x as fast as an A15.
That means the A57 scores 613 in 32bit mode and 741 in 64bit mode. According to ARM themselves.
What about the Exynos 5433? Well, it scores 1128 at 1.9GHz. Converted to 1.0GHz and it scores...wait for it.....594. Compared to the 613 according to ARM it appears the Exynos isn't really the POS I stated, in that it's very close to ARM's number (in 32bit mode).
So there goes your (and mine) first argument, that the Exynos is garbage and we should not use its benchmark results. It appears the Exynos A57 performs about as well as ARM says it should.
Now what about the A7/A8? The A7 normalized to 1.0GHz scores 1076 in 64bit mode and the A8 scores 1142.
I'm not sure what you think is "impressive", but I'd say a processor that scores 1076/1142 is a hell of a lot faster than one that scores 741 (according to ARM).
Now I'd LOVE for you to explain why you think the Qualcomm 810 is going to somehow magically do better than what ARM says the A57 is capable of when the Exynos 5433 performs as ARM predicted and Qualcomm isn't making a custom version of the A57 (like they did with Krait).
In this case, I'm going to normalize all the numbers to 1.0GHz to make comparing processors easier.
...
Exynos scores 888 at 1.90GHz. This gives us a figure of 467 when normalized to 1.0GHz.
Tegra K1 scores 1088 at 2.128GHz. This gives is a figure of 511 when normalized to 1.0GHz.
For sake of argument, I'm going to use the fastest ever tested A15 from the Tegra K1.
Now based on ARM's own numbers, an A57 in 32bit is 1.2x and an A57 in 64bit is 1.45x as fast as an A15.
That means the A57 scores 613 in 32bit mode and 741 in 64bit mode. According to ARM themselves.
What about the Exynos 5433? Well, it scores 1128 at 1.9GHz. Converted to 1.0GHz and it scores...wait for it.....594. Compared to the 613 according to ARM it appears the Exynos isn't really the POS I stated, in that it's very close to ARM's number (in 32bit mode).
...
Now what about the A7/A8? The A7 normalized to 1.0GHz scores 1076 in 64bit mode and the A8 scores 1142.
I'm not sure what you think is "impressive", but I'd say a processor that scores 1076/1142 is a hell of a lot faster than one that scores 741 (according to ARM).
How do you justify normalizing to the same clock speed when you are comparing different architectures? Some architectures (like Qualcomm) may be simply designed to run at higher clock speeds to accomplish the same work that other architectures (like the A7/A8) would perform at lower clock speeds. It's more meaningful to instead normalize by overall power consumption, since it could be the case (which needs to be checked of course) that the higher clock speeds of the Qualcomm are partly counterbalanced by a simpler design, or equivalently that the lower clock speeds of the A7 require more transistors to achieve.