benchmark new smartphones:

[mclane]

Please test 2 minutes so that we can see heat effects , throttling, on the cpu.

[flither]

Ok, another try, every 30 seconds so easier to compare. Phone in the room temperature

Xiaomi Mi 6, SD 835, Android 7.1.1

At start - ~4280 kNps
30'' - 4222 kNps
60'' - 4072 kNps
90'' - 4016 kNps
120'' - 3677 kNps

[Anton]

Please test 2 minutes so that we can see heat effects , throttling, on the cpu.

After some furter tweaking of the phone, getting 300,000 more nps on the Huawei mate 10 pro, with little to no throttling. ran for 10 minutes.



The issue of throttling is interesting, If your phone starts out at 5,300knps, but after just 2 minutes is throttling back to 4,700knps, then the processor is too powerful for the phone body into which it is build, doesn't matter if the back is glass or metal, the design of the cooling system inside the phone is not adequate. and is not fit for purpose (chess). The kirin 970 does not appear to overheat in the glass body after a long run, it gets warmish, but not excessively.

The poor heat dissipation of glass is overstated in my opinion, my pans are made of glass, and they conduct heat adequately to be used as pans without issue, and the glass body of the phone is very cold to the touch in the mornings, surprisingly so. The cpu was running 7c cooler than a neigbouring plastic phone when brought out of standby this morning, the plastic back phone never feels cold to the touch, so plastic is perhaps the least suitable material for heat dissipation in phone construction.

Image buying a Honda CBR600 with a stated top speed of 240km/h, if you ride it at top speed then discover it throttles back to 180km/h after 2 minutes, you would not be a happy camper, it would be returned to the bike shop pronto, no?

It would be interesting to see the numbers on your Motorola and Essential (or any other phone) after running the test for 10 minutes, does it throttle back any more, or does it maintain the same throttling seen after 2 minutes?? please test...

[flither]

So I went after this idea and retested my Xiaomi Mi 6 (glass back). This time I took off my plastic back cover and tested it after cooling 1 hour in a room temperature.
The first 2 minutes gave much better results than previous test, although later massive throttling took place...

Start to ~1 minute: 4650 kNps
2' - 4035 kNps
3' - 3793 kNps
4' - 3545 kNps
5' - 3453 kNps
6' - 3118 kNps
7' - 3061 kNps
8' - 2962 kNps
11' - 2793 kNps
15' - 2628 kNps

-40% lost due to throttling, but only 6% battery lost

I'm thinking about getting rid of thermal config files, but is it worth "just" for chess?

[mclane]

The reason why we test at 2 minutes is exactly that we want to measure a realistic value and not a PEAK value.
Therefore the software (droid fish) should be in default stage (do not change the app) and the phone should be without background apps in normal Condition, just delete the tasks in the task list.
Some phones need that you go into energy settings and put speed to full speed and not smart mode or energy saving mode that could reduce the speed.
In iOS mobile phone Stockfish app can be used, on windows systems any latest stockfish app.
On windows systems make sure the laptop uses fast mode and not any energy reducing mode that weakens the speed.

[kinderchocolate]

I'm thinking about getting rid of thermal config files, but is it worth "just" for chess?

No. On the mobile platform, battery consistency is by far much more important than raw calculation speed.

[mclane]

I have bought the “pocophone f1” in the 6 GB / 128 GB version via Gearbest. It costs around 300 Euro and has a big display , 6.18“ LCD,
And a fast 845 Snapdragon cpu.
It is a flagship killer.
What I liked most is that it has a damned plastic backside instead of 2 glass sides.
In the moment glass front and back is popular and it is not easy to get metal or plastic phones that are fast and state of the art.

But the pocophone has it.

Plastic will not break when you let it fall on the ground.
And plastic costs not 599 dollar to repair the broken backside then, like the new iPhones cost if you repair the backside.

Pocophone started fast with over 5259 peak and throttled to 4559 after 120 seconds thinking with its 8 cores.

Antutu was 264074.

The 845 is only slightly faster then a 835.
My Moto z2 force has 208087 in antutu,

The next thing u do is putting a QI pad into the usb C slot and bring it between the silicon case and the plastic backside.
And then I can use the phone on any QI loader, and the plug in the usb C also protects it from dust,

There comes a silicon case with the phone,
But this has. I protection for the front, only protecting the camera hub on the backside,
I want to protect both, so I will order another one.

IMO the pocophone could be a flagship killer for the one plus6 or other flagships.

The price is very low, isn’t it ?

Camera is not that good like a Samsung or iPhone but it is decent,
The face protection and the fingerprint reader on the backside works very good,
It has also double tap to wake.

[adnoh]

iPhone XS - 5119 kN/s

Smaller.jpg

[stavros]

why 2 mins? if i want to analyse positions for >10 mins the thermal throttling will be massive.
you will be surpised when you see the "flag ships" phones real speed at least 10 mins anal.
show me a "fast" phone with lower than 20%-40% throttling after >10 mins and i will paid x10 for it.
for chess speed the "fast" and expensive flagships are so bogus.watch what you buy!

[mclane]

It’s 2 minutes because this usually gets the machines in heat trouble.