Anony, 07 Feb 2017So SD625 is much faster than SD652/SD650??Not that 'much' irl performance but SD625's real benefit is ultra-power efficiency, deploying 8x Cortex A-53 at 14nm
Well, first things first great job GSMarena Team! And i would like to suggest a smartphone screen guide and its improvement from the older generation
Great article. Thank you very much!
This so hard to simply compare performance of SoCs though several generations.
Looking forward for Apple chips. And it wil be awesome if you can add used power for cores in mW. These data are very rare, because this is very difficult (or nearly impossible) to calculate it during device test.
Good direction guys, I hope you will build on it.
Couple of points where I think you can go even beyond what you did.
1. Next time try geekbench 4. I mean I see sd810 up there in terms of performance with the sd820 in geekbench 3 and we all know there's a bit of a gap between those two in real world performance.
2. Include a per watt performance metric, it's meaningful and it impacts directly battery life.
3. Try to get also the apple A - chipsets into play, at the end of the day they are still custom ARM so they belong here
4. I would suggest separate charts for single core and multi core. You do realize there are chipsets that score very good in multicore like the samsung A5 exynos 7580/7880 but they can't compare to the likes of exynos 8890 in single core performance. Real world performance we all see what the 8890 in the s7 does and the A5 does not even come close.
Anonymous, 08 Feb 2017Geekbench 3 is a bad test. They should have gone with Geekbench 4.
In there it's very palpa... moreGeekBench 3 is far from a perfect choice, but since we relied on our own test database in order to guarantee consistency, it was the pure CPU test that offered the largest amount of samples, going back a couple of years.
Also, there is a single-core chart provided on page 3. It just takes a little bit of time to load.
Amazing job. Well done!!!!
Yes absolutely, please I would you like to see more projects like this in the future!!!!
Great Job! Although would be great to know whether all chipmakers provide support for major OS releases like MM, Nougat etc. to make it easier for folks to decide. Common notion is Qualcomm>Kirin=Exynos>Mediatek in terms of performance, efficiency and future updates but let's see if it can be proven!
First of all a big thanks for the work, the performance comparison of (just) multicore score tends to give the readers a completely different picture clearly far from truth. Yes you guys have mentioned it, but you need to be focused on what kind-of an idea you are planting into a less informed reader. If you are familiar to the site called 'cpubench', they have some benchmarking for Android as well, though not widely used to compare in the smartphone world. They do the benchmarking just like antutu with the performance analysed from different components like storage, CPU, GPU etc and they give it a "score". And based on that they rank a phone model can be boiled down to (SOC). My suggestion to you guys is that with all the numbers you have,, like the single core score (which is according to me at least) more important than multicore score, and the GPU scores etc, how about come up with a rating system where you identify the importance of each division and according to the score a phone achieve in each division (the rating system has to be made in such a way that it shows what's the best according to "~real world" scenarios).
Your works are becoming more important as people search for an information regarding smartphone and your site is most certainly accessed by that user, as you can influence a world wide market so easily, make sure you it's precisely studied and clearly explained. Keep up the good work and always try to give the readers the most accurate information that there is. Good luck.
Hats off! One of the most comprehensive articles I have ever read on chipsets. This will surely be bookmarked. Eagerly waiting for SD835 and see how it fares on these charts. But seriously, great work guyz!
Like a G6 :v, 08 Feb 2017They explained the what is a CPU, its performance, and I guess its architecture. If you take a... moreThat's why I asked about it. We got the process and architecture explained, then we're given the benchmarks of the SoCs, but including only CPU score.The title says "mobile chipset guide", but they're saying only about "half" of the chipset...
Take a bow, GSMArena!!! An article I will bookmark, and hope for regular updates! :)
Nice post. I read it from top to bottom. Handy for referencing later.
Pabliell, 08 Feb 2017Cool, but where are the GPU numbers? SoC is both CPU and GPU. so why only compare CPUs? With t... moreThey explained the what is a CPU, its performance, and I guess its architecture. If you take a look at this article, they never mentioned GPU.
Cool, but where are the GPU numbers? SoC is both CPU and GPU. so why only compare CPUs? With the GPU scores combined the Kirin processors would not be on top, because of low GPU cores count.
Samsung would be 1st, Qualcomm 2nd, then Huawei with Kirin.
Most people around the world don't look for the latest Snapdragon SoC, they look for the newest phone and that's it, why do you think Apple and Samsung latest phones are more popular than the previous ones? It's not their performance they're looking for, people just want to show up the phones and brag that they got the most expensive iPhone 7 Plus or Samsung Galaxy S7 Edge. // That's the reason why these phones are not designed for power/heavy users anymore, because the Galaxy S-series and iPhones have worse battery life than other phones out there.
Thanks so much to the team at GSM Arena for such a great article. I am a big fan of you for not being biased, very skillful, expert and respectable in the tech field. Your site is the first one i think of when i want to see a real review of smartphones. Good luck to you and keep up the great job.
All the best,
From Angkor Wat temple, Siem Reap, Cambodia.
I hope this article and others like it will finally succeed in drilling into the ignorant's brain that the different chipsets aren't significantly better than the other in real world performance.
Anony, 07 Feb 2017So SD625 is much faster than SD652/SD650??Geekbench 3 is a bad test. They should have gone with Geekbench 4.
In there it's very palpable how bad Qualcomm *actually* is (barely breaking the 4k point all the while the rest was battling for 6k last year).
I hope they redo the test but using Geekbench 4 this time around. Also please add single core charts as well as more custom workloads (like SpecInt or phoronix's very own bench suitel