So if we look at the, very comprehensive, performance analysis done by Anandtech on the US variants of the Galaxy S 3 (AT&T and Tmobile versions) thus far. Two things become evident.
View attachment 28406
1. The EVO LTE is faster in 9/11 tests. And while we're talking about academically faster (most of the tests were close enough to justify saying "they all preform about the same"), there were some tests where there was quite a gulf in performance.
2. Given that we're talking about basically the same hardware in both devices (SOC - S4 with an Adreno 225) There is some differentiatior that, I'm guessing, has to be in software to cause the difference.
How did I come to this conclusion?
Because there's no Sprint variant to test, I took the results from the AT&T and Tmobile versions and averaged them. Yes, this is not perfect, yes, I understand that you shouldn't average averages - and I'd much rather have a sample of real Galaxy S3 Sprint data...but as I'm very sure that Anand's team has meticulously gone through their performance tests, and their good about ensuring that their data is accurate -
Some highlights:
Evo Performance wins:
Sunspider Javascritps - by a whopping 11% margin. Maybe something to do with the lack of customization from code ported over from quad-core Exynos version - but that's pretty dramatic.
Vellamo -by 8.9%. This one confuses the hell out of me. vellamo is Quallcoms own benchmark - so both krait chips should preform equally.
Linpack single thread - 9.8% and multi-threaded at 10.8%
Gl Bench (720p offscreen for both Egypt and Pro) were pretty comporable - HTC pulled away by 3% in both cases. Makes sense, same GPU in both chips
Iperf Wifi Performance - by 42% this one's a killer, and probalby has more to do with the on SOC wifi solution in the HTC vs. the add-in solution employed by the GS3...but man.
GS3 Performance wins:
Browser mark - by 10%. How on earth it managed a win in this spot when it lost equally in Javascript performance is beyond me - but a win none the less
Basemark ES 2.0 - I wouldn't necessarily classify this as a win - both phones performed about the same with both trading off 2% wins in opposite tests. The GS3 didn't loose here though - so I'm counting it as a win
It'll be interesting to see how performance changes with software updates. Both have critical flaws that need to be addressed (EVO - multi-tasking, and GS3 sub-par wifi performance) - but I'm more interested to compare the Sprint variant to the EVO LTE with regard to LTE speed - hopefully, the Sprint GS3 won't suffer similar LTE performance (to wifi performance) - but I'll keep my eye out.
Jack
View attachment 28406
1. The EVO LTE is faster in 9/11 tests. And while we're talking about academically faster (most of the tests were close enough to justify saying "they all preform about the same"), there were some tests where there was quite a gulf in performance.
2. Given that we're talking about basically the same hardware in both devices (SOC - S4 with an Adreno 225) There is some differentiatior that, I'm guessing, has to be in software to cause the difference.
How did I come to this conclusion?
Because there's no Sprint variant to test, I took the results from the AT&T and Tmobile versions and averaged them. Yes, this is not perfect, yes, I understand that you shouldn't average averages - and I'd much rather have a sample of real Galaxy S3 Sprint data...but as I'm very sure that Anand's team has meticulously gone through their performance tests, and their good about ensuring that their data is accurate -
Some highlights:
Evo Performance wins:
Sunspider Javascritps - by a whopping 11% margin. Maybe something to do with the lack of customization from code ported over from quad-core Exynos version - but that's pretty dramatic.
Vellamo -by 8.9%. This one confuses the hell out of me. vellamo is Quallcoms own benchmark - so both krait chips should preform equally.
Linpack single thread - 9.8% and multi-threaded at 10.8%
Gl Bench (720p offscreen for both Egypt and Pro) were pretty comporable - HTC pulled away by 3% in both cases. Makes sense, same GPU in both chips
Iperf Wifi Performance - by 42% this one's a killer, and probalby has more to do with the on SOC wifi solution in the HTC vs. the add-in solution employed by the GS3...but man.
GS3 Performance wins:
Browser mark - by 10%. How on earth it managed a win in this spot when it lost equally in Javascript performance is beyond me - but a win none the less
Basemark ES 2.0 - I wouldn't necessarily classify this as a win - both phones performed about the same with both trading off 2% wins in opposite tests. The GS3 didn't loose here though - so I'm counting it as a win
It'll be interesting to see how performance changes with software updates. Both have critical flaws that need to be addressed (EVO - multi-tasking, and GS3 sub-par wifi performance) - but I'm more interested to compare the Sprint variant to the EVO LTE with regard to LTE speed - hopefully, the Sprint GS3 won't suffer similar LTE performance (to wifi performance) - but I'll keep my eye out.
Jack