I was interested in having a performance discussion around the Galaxy S4 (S600 Variant) in direct contrast to the HTC One. This is, of course, a polarizing topic ? and thus, I?ll have to set some rules:
1) If you?re interested in talking about performance differences, you?re welcome to ? this needs to be backed up with fact, not fanciful arguments about ?feel,? or intangible performance metrics
2) If you?re only interested in talking about intangibles, you?re not welcome to discuss them here
3) I don?t want to talk about the build quality, display quality, removable battery or SD card options as ?value? to offset the device. This is NOT a topic about the value of a device, rather, a discussion about the pure performance of said devices
4) I don?t particularly care which device you like better ? performance is about fact, preference is about opinion ? only one is valid for this type of discussion
So the big question is ?Why?? right? Why bother having a discussion about the performance of a device (or devices) in this case ? that should perform similarly?good question ? the answer is?they don?t perform similarly ? in fact, on average, there?s a pretty big gulf in performance if you?re talking in technical terms. ?What?s big? is the next question. If I look at all of the numbers summed and average (and understand that benchmarks are benchmarks ? and there?s always a margin of error) the Galaxy S4 performs about 15% faster overall. That?s a pretty big difference between ?like? devices ? and certainly not one that can be explained purely by looking at the 200 MHz difference in clock speed between the devices.
So what do I mean by that? Let?s look at some of the test results to understand. I?ve got my table broken out to examine not only the sheer performance differences between the flagships ? but also the generational leap between the respective devices and their predecessors. How much of a gulf is there ? a pretty big one.
The Galaxy S4 is, again on average, 40.6% faster than the Galaxy S3 (results are based on the Qualcomm equivalents of both ? I?m not touching the Exynos chips today).
The HTC One is 43% faster than the HTC One X (again, looking purely at the Qualcomm versions ? not the Tegra or XL version of the One X)
I think, to level set here, that both companies need to be commended for such a performance leap from generation to generation. Things only get interesting when you start looking at the devices against one another. From a top down perspective:
The Galaxy S4 takes the performance crown in 8 of 9 tests, with an average performance advantage of ~15%. But when you start to look at the tests, there are some that are much more divergent from this average. GL Benchmark 2.5HD (native) for example ? demonstrates a 19% performance advantage for the S4: which is interesting ? because if you look at the ?1080p? which, theoretically, should line up like for like (since 1900*1080 is the default rendering resolution) ? there?s only a 15% advantage for the GS4. Geekbench is even more interesting, coming in at an insane 22% difference between the two (with the Galaxy S4 as the clear winner).
The HTC One does net a win, in HTML5 rendering performance in Vellamo ? which is again, fascinating given that it?s a lower clocked variant of the same chip found in the GS4 ? and still outperforms it by nearly 11.5%.
So that brings me to an interesting line in the sand, ?the lag.? Man oh man, if I never hear this term again ? it?ll be too soon. Not because I don?t acknowledge the little hiccups and differences in animations from time to time on the GS4 (especially when compared to the HTC One) ? but I?m struggling to understand how the masses perceive that somehow Touchwiz is at fault, given that raw performance is so one sided in the GS4?s favor. I understand, conceptually, that the animations are a little nuts, but given the horsepower difference ? you can no longer empirically state that touchwiz is a factor ? it?s obvious that Samsung is doing something better in Touchwiz than HTC is doing with sense (in terms of raw performance).
It?ll be interesting to see how the ?Nexus experience? GS4 performs against the touchwiz version. I?m going to go out on a limb and guess?roughly the same, or slightly worse. Before you go beating me up ? remember, there?s obviously some heady optimizations being done here ? a simple clock speed difference cannot account for the above performance differences ? that Touchwiz will be absent, indicates that some of those optimizations might not make it over to the (what I?m understand as) largely AOSP device. I?m happy to eat my words though ? prove me wrong Samsung. I think if anything, this largely states a case for NOT bringing a similar HTC version out to market. A Nexus HTC One would do little to improve the view point that the HTC One is superior to the GS4 (which is what HTC so desperately needs to prove) from a pure performance standpoint anyway. Sans the fancy camera software (which would undoubtedly NOT be transitioned to a Nexus AOSP experience) ? that shooter isn?t going to look nearly as good against a raw 13mp camera (again, handicapped by AOSP camera software as well) ? and the performance of the stock HTC vs. GS4 isn?t going to change greatly with taking away any optimizations that HTC has done with their s600 either.
All that said, the above is in no way an indication that I think one device is better or worse. There are certainly tradeoffs to be made (and not discussed here) ? but from a raw performance perspective ? this is an interesting hardware generation indeed.
1) If you?re interested in talking about performance differences, you?re welcome to ? this needs to be backed up with fact, not fanciful arguments about ?feel,? or intangible performance metrics
2) If you?re only interested in talking about intangibles, you?re not welcome to discuss them here
3) I don?t want to talk about the build quality, display quality, removable battery or SD card options as ?value? to offset the device. This is NOT a topic about the value of a device, rather, a discussion about the pure performance of said devices
4) I don?t particularly care which device you like better ? performance is about fact, preference is about opinion ? only one is valid for this type of discussion
So the big question is ?Why?? right? Why bother having a discussion about the performance of a device (or devices) in this case ? that should perform similarly?good question ? the answer is?they don?t perform similarly ? in fact, on average, there?s a pretty big gulf in performance if you?re talking in technical terms. ?What?s big? is the next question. If I look at all of the numbers summed and average (and understand that benchmarks are benchmarks ? and there?s always a margin of error) the Galaxy S4 performs about 15% faster overall. That?s a pretty big difference between ?like? devices ? and certainly not one that can be explained purely by looking at the 200 MHz difference in clock speed between the devices.
So what do I mean by that? Let?s look at some of the test results to understand. I?ve got my table broken out to examine not only the sheer performance differences between the flagships ? but also the generational leap between the respective devices and their predecessors. How much of a gulf is there ? a pretty big one.
The Galaxy S4 is, again on average, 40.6% faster than the Galaxy S3 (results are based on the Qualcomm equivalents of both ? I?m not touching the Exynos chips today).
The HTC One is 43% faster than the HTC One X (again, looking purely at the Qualcomm versions ? not the Tegra or XL version of the One X)
I think, to level set here, that both companies need to be commended for such a performance leap from generation to generation. Things only get interesting when you start looking at the devices against one another. From a top down perspective:
The Galaxy S4 takes the performance crown in 8 of 9 tests, with an average performance advantage of ~15%. But when you start to look at the tests, there are some that are much more divergent from this average. GL Benchmark 2.5HD (native) for example ? demonstrates a 19% performance advantage for the S4: which is interesting ? because if you look at the ?1080p? which, theoretically, should line up like for like (since 1900*1080 is the default rendering resolution) ? there?s only a 15% advantage for the GS4. Geekbench is even more interesting, coming in at an insane 22% difference between the two (with the Galaxy S4 as the clear winner).
The HTC One does net a win, in HTML5 rendering performance in Vellamo ? which is again, fascinating given that it?s a lower clocked variant of the same chip found in the GS4 ? and still outperforms it by nearly 11.5%.
So that brings me to an interesting line in the sand, ?the lag.? Man oh man, if I never hear this term again ? it?ll be too soon. Not because I don?t acknowledge the little hiccups and differences in animations from time to time on the GS4 (especially when compared to the HTC One) ? but I?m struggling to understand how the masses perceive that somehow Touchwiz is at fault, given that raw performance is so one sided in the GS4?s favor. I understand, conceptually, that the animations are a little nuts, but given the horsepower difference ? you can no longer empirically state that touchwiz is a factor ? it?s obvious that Samsung is doing something better in Touchwiz than HTC is doing with sense (in terms of raw performance).
It?ll be interesting to see how the ?Nexus experience? GS4 performs against the touchwiz version. I?m going to go out on a limb and guess?roughly the same, or slightly worse. Before you go beating me up ? remember, there?s obviously some heady optimizations being done here ? a simple clock speed difference cannot account for the above performance differences ? that Touchwiz will be absent, indicates that some of those optimizations might not make it over to the (what I?m understand as) largely AOSP device. I?m happy to eat my words though ? prove me wrong Samsung. I think if anything, this largely states a case for NOT bringing a similar HTC version out to market. A Nexus HTC One would do little to improve the view point that the HTC One is superior to the GS4 (which is what HTC so desperately needs to prove) from a pure performance standpoint anyway. Sans the fancy camera software (which would undoubtedly NOT be transitioned to a Nexus AOSP experience) ? that shooter isn?t going to look nearly as good against a raw 13mp camera (again, handicapped by AOSP camera software as well) ? and the performance of the stock HTC vs. GS4 isn?t going to change greatly with taking away any optimizations that HTC has done with their s600 either.
All that said, the above is in no way an indication that I think one device is better or worse. There are certainly tradeoffs to be made (and not discussed here) ? but from a raw performance perspective ? this is an interesting hardware generation indeed.