10-03-2012 07:09 AM
737 ... 2324252627 ...
tools
  1. Kevin OQuinn's Avatar
    Anandtech didn't run tests on any flagships with jellybean. Geekbench tests overall performance, so I'd value a geekbench test more. And they show that the S3 with Jellybean beats the iPhone 5. I'm going to download geekbench and see what my One X with JB scores.

    Sent from my Nexus 7 using Android Central Forums
    Hmmm...fresh restart on my One X (AT&T) running ParanoidAndroid (Jelly Bean) just got 1566. Which knowing what I know about hardware just doesn't line up. The PowerVR543MP3 is MUCH more powerful than the Adreno 225. Actually, almost any other GPU is (the nvidia and mali included). The CPU cores in the iPhone5 are at least as fast. Maybe later I'll flash an ICS ROM and compare.

    This does nothing to explain HOW software can make such a huge difference on a pure CPU benchmark. Anandtech chose the benchmarks they did because they are more consistent from one platform to the next. Even looking at the browser results the software wouldn't make that much of a difference. It appears as though people are cherry picking the one or two benchmarks that will allow them to say that they have a faster phone, when common sense should tell you that the iPhone 5 has some of the fastest hardware inside of it or any other phone released to date.

    It's ridiculous to think that anything based on A9 can compete directly with a brand new architecture, knowing for a fact that the GPU is also outclassed. I mean that's a Snapdragon S4 Pro QUAD-CORE that is barely keeping pace with the iPhone 5.

    Also, you (not specifically) can't claim that the CPU or GPU is slower, and then use a benchmarks overall result as evidence of that. If you want to argue about a CPU or GPU, use a CPU or GPU specific benchmark. Then you are isolating those particular pieces of hardware for an easier and more direct comparison. Which is why I'll take the Anandtech results. I'm sure they will also test RAM speed and storage speed, too.

    Even if the hardware is slower and software makes up the difference, that should AT LEAST prove how well Apple optimizes the hardware for the software and vice versa.
    09-23-2012 08:48 PM
  2. crackberrytraitor's Avatar
    Keep in mind the A6 is underclocked to 1ghz.
    Yes, 1.02ghz to increase battery life. Same with the S4. It can smoothly run at 1.89 GHz, but is clocked at 1.5 GHz for stable battery life. Either way, geekbench makes it pretty clear that the S4 is at the very least equal to and possibly significantly more powerful than the A6. I think it's pretty clear as well that the S4 pro will perform even better.

    Sent from my One X using Android Central Forums
    09-23-2012 08:50 PM
  3. Kevin OQuinn's Avatar
    My One X with Jellybean scores a 1594 as opposed to the iPhone 5's 1600. When I clock it to 1.7 GHz it scores 1726. That means when not clocked on high, the One X scores 0.3% lower than the iPhone 5, when overclocked it's 7.8 percent faster. I don't have a Tegra 3 phone, but I'll run geekbench on my N7 which has pretty much identical specs to most Tegra phones.

    Also, I'd point out that geekbench shows the iPhone 5 having 1gb ram, but the One X having only 673. This lowers the score and suggests that processor vs processor, the S4 handily wins.

    Sent from my One X using Android Central Forums
    Now slow it down to 1.2ghz and see what the result is. Make it fair. Heck, go to 1.3, because for some reason you think RAM is making a huge difference here (it's not, because some of the RAM that isn't being reported is already being used by the GPU).
    09-23-2012 08:50 PM
  4. crackberrytraitor's Avatar
    My S3 U.S. edition scores 1698 clocked at 1.5 GHz with AOKP. This is probably because of the extra RAM

    Sent from my One X using Android Central Forums
    09-23-2012 08:54 PM
  5. Kevin OQuinn's Avatar
    What's your memory score?
    09-23-2012 08:56 PM
  6. crackberrytraitor's Avatar
    Now slow it down to 1.2ghz and see what the result is. Make it fair. Heck, go to 1.3, because for some reason you think RAM is making a huge difference here (it's not, because some of the RAM that isn't being reported is already being used by the GPU).
    That's not making it fair, we're comparing which is a more powerful overall CPU, not which is more powerful at the same clock speed.

    Sent from my Nexus 7 using Android Central Forums
    09-23-2012 09:13 PM
  7. crackberrytraitor's Avatar
    What's your memory score?
    The S3 or the One X?

    Sent from my Nexus 7 using Android Central Forums
    09-23-2012 09:13 PM
  8. Kevin OQuinn's Avatar
    That's not making it fair, we're comparing which is a more powerful overall CPU, not which is more powerful at the same clock speed.

    Sent from my Nexus 7 using Android Central Forums
    Ok. So clocked 300mhz faster it's .3% slower? Got it. Remember that we know the instruction set is the same, but the architecture is not. This means that Apple could have optimized things in a different way than Qualcomm.

    The S3 or the One X?

    Sent from my Nexus 7 using Android Central Forums
    Both if you have them handy, but I was asking for it from the S3.
    09-23-2012 09:17 PM
  9. crackberrytraitor's Avatar
    Yes, but the A6 probably can't even go up to 1500 MHz. Clock speed isn't relevant. That's like saying a single core cpu is more powerful than a quad core cpu if the single core is 15% more efficient per mhz, which is incorrect, just try running some advanced computations through both CPUs, the single core that's 15% more efficient is going to perform poorer overall. We're comparing which device performs better overall, with the CPU at native speed.

    The One X got a 1729 memory score with 673 mb ram recognised. The S3 got a 1986 with 1744 mb recognised. The iPhone 5 got an 1811 with 1015 mb recognised.

    The fact is, with jellybean the S3, both U.S. and international editions, beat the iPhone 5 in overall performance. It's very, very clear.
    09-23-2012 09:30 PM
  10. JHBThree's Avatar
    Yes, but the A6 probably can't even go up to 1500 MHz. Clock speed isn't relevant. We're comparing which device performs better overall, with the CPU at native speed.

    The One X got a 1729 memory score with 673 mb ram recognised. The S3 got a 1986 with 1744 mb recognised. The iPhone 5 got an 1811 with 1015 mb recognised.

    The fact is, with jellybean the S3, both U.S. and international editions, beat the iPhone 5 in overall performance. It's very, very clear.
    You cannot fairly compare two chips when one is clocked at 1ghz and one is 1.5.

    Also, given the A6s pedigree, I would not be shocked if it could be clocked above 1.5 without breaking a sweat.
    09-23-2012 09:32 PM
  11. crackberrytraitor's Avatar
    You cannot fairly compare two chips when one is clocked at 1ghz and one is 1.5.

    Also, given the A6s pedigree, I would not be shocked if it could be clocked above 1.5 without breaking a sweat.
    Read the edit to my above post to see why what you're saying is not accurate.

    Sent from my One X using Android Central Forums
    09-23-2012 09:36 PM
  12. crackberrytraitor's Avatar
    You cannot fairly compare two chips when one is clocked at 1ghz and one is 1.5.

    Also, given the A6s pedigree, I would not be shocked if it could be clocked above 1.5 without breaking a sweat.
    Yeah, no. They wouldn't spend the extra cash manufacturing chips that can be stably clocked 50 percent faster than their native speed. That's beyond unlikely.

    Sent from my One X using Android Central Forums
    09-23-2012 09:41 PM
  13. JHBThree's Avatar
    Yeah, no. They wouldn't spend the extra cash manufacturing chips that can be stably clocked 50 percent faster than their native speed. That's beyond unlikely.

    Sent from my One X using Android Central Forums
    In your judgement, which doesn't mean anything. Given that Apple designed this chip, that means it will be found in some capacity in Apple products for at least a year, but probably longer. Apple never designs something without some flexibility in its design, and they never skimp on spending on components.

    Sent from my SCH-I535 using Tapatalk 2
    pauldroidr2d2 likes this.
    09-23-2012 09:45 PM
  14. JHBThree's Avatar
    Read the edit to my above post to see why what you're saying is not accurate.

    Sent from my One X using Android Central Forums
    It is entirely accurate when you're comparing the S4 and A6.

    Sent from my SCH-I535 using Tapatalk 2
    09-23-2012 09:46 PM
  15. crackberrytraitor's Avatar
    It is entirely accurate when you're comparing the S4 and A6.

    Sent from my SCH-I535 using Tapatalk 2
    No, it really isn't. CPU efficiency isn't the same thing as CPU power. If you want to say the A6 is more efficient, go ahead. If you want to say it's more powerful, well, quite simply, you're wrong.

    Sent from my One X using Android Central Forums
    09-23-2012 09:55 PM
  16. crackberrytraitor's Avatar
    In your judgement, which doesn't mean anything. Given that Apple designed this chip, that means it will be found in some capacity in Apple products for at least a year, but probably longer. Apple never designs something without some flexibility in its design, and they never skimp on spending on components.

    Sent from my SCH-I535 using Tapatalk 2
    The iPhone 4s could run at a max 1ghz but was clocked stock at 800mhz. If you think the A6 can run at 1.5 GHz when they clock it a 1.02, I'm sorry, but you don't know what you're talking about. 1.2, maybe 1.3 if we're highballing.

    Sent from my One X using Android Central Forums
    09-23-2012 10:00 PM
  17. JHBThree's Avatar
    The iPhone could run at a max 1ghz but was clocked stock at 800mhz. If you think the A6 can run at 1.5 GHz when they clock it a 1.02, I'm sorry, but you don't know what you're talking about. 1.2, maybe 1.3 if we're highballing.

    Sent from my One X using Android Central Forums
    You clearly don't know anything about the architecture apple used.

    Sent from my SCH-I535 using Tapatalk 2
    09-23-2012 10:02 PM
  18. JHBThree's Avatar
    No, it really isn't. CPU efficiency isn't the same thing as CPU power. If you want to say the A6 is more efficient, go ahead. If you want to say it's more powerful, well, quite simply, you're wrong.

    Sent from my One X using Android Central Forums
    They are comparable. The A6 architecture is equivalent to the S4.

    Sent from my SCH-I535 using Tapatalk 2
    09-23-2012 10:04 PM
  19. Kevin OQuinn's Avatar
    Yeah, no. They wouldn't spend the extra cash manufacturing chips that can be stably clocked 50 percent faster than their native speed. That's beyond unlikely.

    Sent from my One X using Android Central Forums
    Not really. We know nothing of how they're binning these chips. Look at all the different models of Tegra 3 due to binning. It's entirely realistic to think there is some OC potential with the A6.

    No, it really isn't. CPU efficiency isn't the same thing as CPU power. If you want to say the A6 is more efficient, go ahead. If you want to say it's more powerful, well, quite simply, you're wrong.

    Sent from my One X using Android Central Forums
    Proof? Anandtech PROVES that the CPU and GPU are more powerful. Geekbench, which you seem to quote and go by so often, don't specify enough.

    The iPhone could run at a max 1ghz but was clocked stock at 800mhz. If you think the A6 can run at 1.5 GHz when they clock it a 1.02, I'm sorry, but you don't know what you're talking about. 1.2, maybe 1.3 if we're highballing.

    Sent from my One X using Android Central Forums
    Geekbench shows that it'll go up to 1.2ghz stock.

    You clearly don't know anything about the architecture apple used.

    Sent from my SCH-I535 using Tapatalk 2
    Nobody does really. We know about the instruction set, but not the architecture. Even with the pic of the die all we have are educated guesses.



    My memory score was a 1671. See how that particular score doesn't scale with quantity? That's my problem with saying extra RAM alone is why the score is higher. It has more to do with the efficiency of the pipeline than anything else in that particular case. Anyway, I'm not going to trust Geekbench to tell me that a particular device is more powerful than any other device, when I can clearly see evidence that one device is MUCH more powerful.

    Any idea how Geekbench goes about getting those numbers? It seems to run awful quickly as opposed to something like Vellamo, which takes much longer to run, and breaks down in great detail what it tested to arrive at those results. I'm not saying we should use Vellamo, was just citing as an example.

    ---------- Post Merged at 10:24 PM ---------- Previous Post was at 10:22 PM ----------

    Also, clockspeed is ALWAYS relevant when talking about benchmarks, since the easiest way to improve benchmark scores is to increase clockspeed.
    09-23-2012 10:24 PM
  20. crackberrytraitor's Avatar
    Not really. We know nothing of how they're binning these chips. Look at all the different models of Tegra 3 due to binning. It's entirely realistic to think there is some OC potential with the A6.



    Proof? Anandtech PROVES that the CPU and GPU are more powerful. Geekbench, which you seem to quote and go by so often, don't specify enough.



    Geekbench shows that it'll go up to 1.2ghz stock.



    Nobody does really. We know about the instruction set, but not the architecture. Even with the pic of the die all we have are educated guesses.



    My memory score was a 1671. See how that particular score doesn't scale with quantity? That's my problem with saying extra RAM alone is why the score is higher. It has more to do with the efficiency of the pipeline than anything else in that particular case. Anyway, I'm not going to trust Geekbench to tell me that a particular device is more powerful than any other device, when I can clearly see evidence that one device is MUCH more powerful.

    Any idea how Geekbench goes about getting those numbers? It seems to run awful quickly as opposed to something like Vellamo, which takes much longer to run, and breaks down in great detail what it tested to arrive at those results. I'm not saying we should use Vellamo, was just citing as an example.

    ---------- Post Merged at 10:24 PM ---------- Previous Post was at 10:22 PM ----------

    Also, clockspeed is ALWAYS relevant when talking about benchmarks, since the easiest way to improve benchmark scores is to increase clockspeed.
    You can't overclock the tegra 3 by 50% of it's overall clock speed. Come on now, seriously guys. If we want to look at Apple's history, the iPhone 4 could be overclocked 200mhz.

    efficiency per mhz IS NOT what we're comparing here. We're comparing which is the most powerful overall. Underpowering a CPU isn't fair for an overall comparison. That's like saying if you're having a big guy fight a little guy that the big guy should wear weights to make it fair. But if you're trying to see who is the better overall fighter making it "fair" is no what's important.

    The fact is, in a test of overall system performance with no CPU hackery, the S3 with the latest android handily beats the i5 with the latest iOS, and the One X ties with it (okay, it technically loses with a 0.3 percent lower score).

    Sent from my One X using Android Central Forums
    09-23-2012 10:54 PM
  21. JHBThree's Avatar
    Except its actually 30% on the A6 since it can go up to 1.2 with no overclocking.

    Sent from my SCH-I535 using Tapatalk 2
    09-23-2012 10:58 PM
  22. crackberrytraitor's Avatar
    Except its actually 30% on the A6 since it can go up to 1.2 with no overclocking.

    Sent from my SCH-I535 using Tapatalk 2
    Didn't know that. If it goes to 1.2 natively I could see it going to 1.5 maybe 1.6 tops OC. However my point still stands. Geekbench pushes the CPU to it's max clocked speed. The S3 beat it handily. That means that with no CPU hackery involved the GS3 with the latest Android triumphs over the i5 with the latest iOS in overall performance. Not to mention the GS3 can over clock to around 2.1 GHz (the One X around 1.9 with current software, but capable of 2.1 as well). If you increase the clock speed of the A6 from 1.2 to 1.5 and the S4 from 1.5 to 2.1 the S4 is still going to stomp it.

    Sent from my One X using Android Central Forums
    09-23-2012 11:17 PM
  23. Kevin OQuinn's Avatar
    No, it really isn't. CPU efficiency isn't the same thing as CPU power. If you want to say the A6 is more efficient, go ahead. If you want to say it's more powerful, well, quite simply, you're wrong.

    Sent from my One X using Android Central Forums
    Didn't know that. If it goes to 1.2 natively I could see it going to 1.5 maybe 1.6 tops OC. However my point still stands. Geekbench pushes the CPU to it's max clocked speed. The S3 beat it handily. That means that with no CPU hackery involved the GS3 with the latest Android triumphs over the i5 with the latest iOS in overall performance. Not to mention the GS3 can over clock to around 2.1 GHz (the One X around 1.9 with current software, but capable of 2.1 as well). If you increase the clock speed of the A6 from 1.2 to 1.5 and the S4 from 1.5 to 2.1 the S4 is still going to stomp it.

    Sent from my One X using Android Central Forums
    Not entirely accurate. It took a little bit of trickery just to get the iPhone to clock up to 1.2ghz (and they used Geekbench to determine it) because Apple is so aggressive with the gating.


    Most powerful still comes down to which one has the fastest hardware (since that's the power). Testing strictly the hardware, and going based off strictly the hardware, the iPhone 5 has everything else beat. You're talking about what Apple talks about, which is what that does for the end user. Which is fine, because ultimately that's what matters.

    The GPU in the IPhone 5 is fast. Really fast. The CPU is equivalent to the S4 (let's just assume for arguments sake). The GPU is the difference maker. And I'm not sure, but I would bet that Apple offloads a lot of the UI drawing to the GPU to save on power. Android can do this, too (I think CM has a hack that allows it on some devices/kernels), and it's supposed to make the UI smoother, save power, and leave the CPU for doing other types of things. This is where the S4 and Tegra 3 have a significant disadvantage. The GPU's aren't as fast as what Apple uses (the Mali in the Exynos is slower, too).


    It just seems like we're talking about raw power, but you're talking about what the software does with it. I also wonder how Geekbench is coded and running its tests on each device. Different OS's, completely different coding languages. Safe bet they don't behave exactly the same because of that.
    09-24-2012 12:03 AM
  24. crackberrytraitor's Avatar
    Not entirely accurate. It took a little bit of trickery just to get the iPhone to clock up to 1.2ghz (and they used Geekbench to determine it) because Apple is so aggressive with the gating.


    Most powerful still comes down to which one has the fastest hardware (since that's the power). Testing strictly the hardware, and going based off strictly the hardware, the iPhone 5 has everything else beat. You're talking about what Apple talks about, which is what that does for the end user. Which is fine, because ultimately that's what matters.

    The GPU in the IPhone 5 is fast. Really fast. The CPU is equivalent to the S4 (let's just assume for arguments sake). The GPU is the difference maker. And I'm not sure, but I would bet that Apple offloads a lot of the UI drawing to the GPU to save on power. Android can do this, too (I think CM has a hack that allows it on some devices/kernels), and it's supposed to make the UI smoother, save power, and leave the CPU for doing other types of things. This is where the S4 and Tegra 3 have a significant disadvantage. The GPU's aren't as fast as what Apple uses (the Mali in the Exynos is slower, too).


    It just seems like we're talking about raw power, but you're talking about what the software does with it. I also wonder how Geekbench is coded and running its tests on each device. Different OS's, completely different coding languages. Safe bet they don't behave exactly the same because of that.
    I would point out that there are almost GPU tests that have been run on the i5 that are not at least partially influenced by software. The output provided to the benchmarks used is based on the efficiency of the routing software rather than the hardware, GLbenchmark in particular. In the few tests of pure hardware, such as floating point calculations per second, the S4 beats the A6 hands down.

    I'm honestly not just blindly defending Android. I've worked with CPUs for many years, all the way back to my Apple II with it's badass early CMOS chip. I used to defend Apple, back when PC clones were invading the market. I think that the A6 is a great improvement. But I am not at all convinced that the A6 is a superior piece of hardware, rather I think it is managed by particularly efficient, cut-down software.

    And btw, Geekbench performs the same on any O.S. It measures the outputs in designated categories. A more efficient O.S. may give a higher score, but because it puts out a higher output from improved hardware management, not from a software variance in Geekbench from platform to platform.

    Sent from my Nexus 7 using Android Central Forums
    09-24-2012 01:46 AM
  25. pauldroidr2d2's Avatar
    Something to ease the tension a bit.

    Anyone else eager to LAUGH at what apple has to offer Sept 12th?-starwars_zps5da998f1.jpg
    09-24-2012 01:55 AM
737 ... 2324252627 ...
LINK TO POST COPIED TO CLIPBOARD