LeTV LeEco Le 2 (X620) shows up running a Helio X25 chipset

money does talk

First the results for Meizu MX6 Pro are fake, secondly Mali T880 @ 850 MHz should be as fast as Adreno 510 but it will consume 170% energy of Adreno, third their is a very small difference between consumption of TSMC HPL 28nm & regular 20 nm PL. Other than that S65x series come with better multimedia capabilities (mostly still not utilized) & better cellular radio.

It is fake as the score is abnormally higher compared to both Kirin 950 (@ 50 MHz more & 35% bigger RAM bandwidth) & other X25 scores.

How about your “reliable surce” & comparation of Mate 8 to the Samsung Galaxy A9, even A9 have a larger display they have similar size of battery & don’t forget Kirin 950 is more powerful effects than X25.

https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=27447856&os1=Android&api1=gl&hwtype1=GPU&hwname1=Qualcomm+Adreno+%28TM%29+510&D2=Huawei+Mate+8+%28NXT-xxx%29

As there are a run time data for TRex you can clearly see that power estimations I told are there.

Honey there is much more than just a nm for used process to take into consideration; there is a rooting lib lv, inter blocks & cetera. When I say there is a little difference between bulk not future optimized 20 nm planar to the high optimized hi performance 28 nm of same TSMC I ment that as a fact. You can find estimations on that site for 20 nm Samsung planar & you will see how achieved results are on pair with 28nm HPL to the 28nm PL.

So only thing you need to do is a little reading along with putting some thinking in the use.

By the way mid range SoC’s are the sweet spot for achieving all of your needs, high end tends to put in more power GPU’s that are not actually suitable by their power consumption. At the end A510 is most power efficient mobile GPU up to date, perfectly suitable for FHD gaming, MALI’s won’t catch up anytime soon, me by new announced Power VR series expected next year will narrow the gap a little.

Quite: “All in all, the argument that this 20nm chip should be more power efficient than the competitors’ 28nm is not completely factual and doesn’t seem to hold up in practice. The process still seems young and unoptimized compared to what TSMC offers on 28nm”

Second paragraph at the end of the same page of the same article. I advise you to start thinking about waring eye glasses.

Now let’s assume how TSMC 20nm bulk (as they also never optimized it, instead they jumped to 16nm FinFET) is still a little better than Samsung’s that still doesn’t make it much better than 28nm HPM, as I actually claimed.

Their is a reason why I actually compared Mate 8 to the Galaxy A9 as they are closest match both with 4000 mAh battery even Mate have 6″ display vs 5.5″ on this A9 model, both are Amoled & are only ones with enough data to do a real comparation especially in battery test, by the way results displayed are based on mid scores of all runs so your argument about consistency & Lux on display don’t stand. People usually run tests with auto brightness so there shouldn’t be much difference by the way usual practice for pro testing is putting brightness to 200 nits. The T-Rex battery test difference of 47% more in favor of S652 or A510 if you like clearly speaks about much better power efficiency of A5xx generation. A530 is advertised as most powerful & most power efficient mobile GPU currently available on market which is true but as you won’t need all that power & it will consume much more battery then a scaled down litle brother A510 in most & every game as clocking logic is far from good. Simply A510 is a sweet spot for 1080p gaming & it does this task most power efficient, A530 is capable of pushing 2K even 4K but all of this eats to much of battery (will also consume much more power wile playing in FHD) & on sub 7″ is useless as you won’t see extra details, VR is other pair of slippers but we are not there yet so I won’t touch this topic. Basicall & well known fact why the Adrenos are leader in performance per W are in the way they are designed as single cluster as single cluster scales 100% in performance & every additional adds around 68~75% depending how good design is.

Actuality T880 MP4 implementation on X25 got a greater performance hit than I anticipated in the first place & it’s every way worse & slower than A510 which is not by all means shining in the GFX bench.

https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=31731956&os1=Android&api1=gl&hwtype1=GPU&hwname1=Qualcomm+Adreno+%28TM%29+510&D2=LeTV+LeEco+Le+2+%28X620%29

If you compare 3D Mark scores difference will become more significant.

It’s very hard to provide colorful evidence to blind people.

Now with all data & facts provided from me and none from you expect those that actually did go with my initial hypothesis I expect you at least to buy those eye glasses as it seams you need them, on the other hand I don’t expect to see you becoming any more intelligent as so far you failed badly to use your brains. Well at least you will see better.

Ok you are right about Mate display, still their is no better comparation currently available. Adreno 530 is advertised as more power efficient from previous generation that whose still more power efficient than T7, T8 Mali’s. Auto brightness is what it is. I told that professional benchmarkers set the brightness value to 200 nits, you again have troubles reading…

“Where do you get idea that unoptimized mean not better power consumption comparison to larger nodes?”

From officiall projections when they announced 20nm process compared to same ones of the 28nm HPM. In best case difference is about 4%.

“Both X20/X25 and SD650/652 manufacturing at TSMC. Hence that sentence about Samsung 20nm vs TSMC 28nm it can’t be applied to 20nm TSMC vs 28nm TSMC comparison.”

Actuality I did state how TSMC 20nm process is better & more advanced than Samsung’s I also stated how you won’t find a direct comparation of TSMC 20 nm to 28nm HPM earlier & I will again state how IT’S NOT SIGNIFICANT.

“YOU-CAN’T-ESTIMATE-EVER-POWER-CONSUMPTION-OF-GPU-FROM-THAT-PARTICULAR-BENCHMARK-HENCE-THERE-IS-A-LOT-MORE-AFFECTS-POWER-DRAIN”

THEIR IS NOTHING EFFECTING A POWER DRAIN MORE THAN GPU UTILIZATION!!! Besides Gxf is pretty lite on CPU block, display is second largest consumer in this case. Certainly Samsungs UI & bunch of apps and frameworks in the background will case bigger power drain than none of those on Huawei along with EmotionUI that lacks even app drawer. I can’t do accurate projection but let’s say 35~40% is in self margins compared to so high clocked T880 MP4.

Make this in personal by finding either better comparation or some facts to any of your claims. So far I did present some not so good comparations but it’s still much better to none that you presented so far. Pardon my ignorance but you didn’t present anything that I could ignore. & stop with that faggat shit.

“Just learn to stop when you proven wrong.” talking to your self aren’t you? I am older than you. I didn’t called you a honey, sweetheart, sauger so if that someone after that tells you to stop with faggat stuff & that insult you then you probably have problems…

I never claimed how it’s 70% more efficient I claimed 35~40%.

You can stick your link where sun don’t shine.

Here you go a link with a little more detailed power/performance/power savings technics, proces optimizations inter blocks & cetera along with their additional dependences when combined together. It’s still not in depth. If you want I have deeper wite paper for 28nm FD-SOI?

https://www.synopsys.com/Company/Publications/SynopsysInsight/Pages/Art4-low-power-IssQ2-12.aspx?cmp=Insight-I2-2012-Art4

By the way 28nm HPM is 40% more power efficient than 28nm HP, 20nm is HP & it’s compared to HP as TSMC always do. Now this would actually suggest how 28nm HPM is more efficient than 20 nm HP which is not entirely true if you take into consideration & SRAM (which eats a large die size) along with density, they probably cost approximately the same now. So at the end it seams how actually 28nm HPM culd actually be around 4~5% better in terms of power consumption. Qualcomm learned a hard way how 20nm HP is not an improvement over 28nm HPM on last year S810.

Edit: we can consider both claims about comparation’s of 28nm HPM & 20nm HP to the 28nm HP even their are from a same official source with a grain of salt actual results on full blown complex SoC are smaller & around 20% 28nm HPM & 15% for 20nm HP.

Actuality I did proven that: A510 is more efficient than T880 MP4 @ 850 MHz, shown how it’s faster & showed how 28nm HPM is actually even better than 20 nm HP from TSMC. Those are the facts, we can argue about actual projections but ewen if we had actual & identical hardware like for instance this LeTV in both versions we still couldn’t make fully acute numbers out of them.

Now I actually have right to be pissed off as I whose right all of the time wile you ware annoying litle crab. In the future don’t make a bold claims & hard headedly assume things without any real hard stand points. I know also to behave quite nice when I am approached nicely or at least normally with a normal question.

You really are dumb ass acting as a little kid.

If I stayed that T880 will consume 170% of the power of A510 is using that translates in to 40% efficiency, if I claimed 70% (& I didn’t) more power efficient that would mean it uses 30% power to achieve same performance as T880 or how T880 uses 320% power.

The Meizu MX6 Pro result is fake as not only that it scores more than any other X25 it scores more than HiSilicon Kirin 950 & that part is simply impossible.

TSMC 28nm process is very scalable, HPM is based on HP that is adapted to use less power & provide good performance. In the figure 2 of linked article their is a simple example how it’s possible to achieve additional power consumption cut (30% less leaking) on it. The part you simply aren’t able to get in your head is how TSMC’s 20nm process is simply not scalable nor adjustable (no one bothered to translate libs for abandoned process). At the base TSMC 28nm HPM & TSMC 20nm use same base voltage so they will consume relatively same power. How ever their is much more in to it from using specific inter blocks & how blocks are rutted & it’s possible to do this better on 28nm HPM than on 20nm as most of the tools are not available for the 20nm process. Qualcomm actually achieved same to better power consumption on HPM than LP back in the S800, S600 day’s. They are by far most experienced in using it. Me by the best example that is good documented how important rooting, placement & inter blocks is one tied to Samsung Exunos SoC’s with Cortex A15 as they achieved in 3 cycles to get it’s power consumption down almost by 50% wile switching only from 28nm to 20nm. You’re own claim is how Samsung’s 20 nm is 30% more power efficient, the rewiev showed us differently & this power consumption of A15 is actually based on real measurements & not projections.

Take a look if you have time:

http://cyberparse.co.uk/2015/02/10/arm-a53a57t760-investigated-samsung-galaxy-note-4-exynos-review/

By the way this is much deeper analys then one you are referring at, actually one you are referring at is based on this & simplified.

Exunos 5433 managed to drove 7.11 Watt from a big cluster (±100 mW margin).

At the end TSMC is advertised it’s 28nm HPC+ process that consumes less power & costs less than 20nm wile delivering comparative performance of 20nm & 28nm HPM wile being compatible with 28nm HPM tools. So you see (I hope) it’s not all in nm, their is much more in implementation & it’s possible to achieve better implementation on 28nm. Even Samsung that actually did optimized it’s 20nm process along with tools and libs is abandoning it in favor of 14nm FinFET & 28nm FD-SOI.