New IPad Pro 10.5 chess benchmark results?

[h1a8]

I mean pc

[MikeB]

Well apple says that it is 30% faster in computing.
If that is true then the benchmark should be around 3224kn/s

(30% up from 2480kn/s iPad pro 9.7)

I just hope the komodo team can create komodo for ios. I would love to use it in stockfish app.

And with asmFish it should be ~ 4200 kn/s

Seriously? Why is asmFish faster than stockfish (3230 vs 2480)?

Because Mohammed did a good job in rewriting Asmfish from C++ into assembly.

CFish is also faster that Stockfish (rewrite from C++ to C). But Ronald got tired of keeping CFish current and it did not get the testing effort like Asmfish.

So asmfish is stronger than Stockfish for pic?

Correct - every 1% gain might give you about 1.4 extra ELO if you believe doubling speed is worth 100 ELO (1.01^70 = ~2; 100/70 = ~1.4 ELO).

haha - just noticed that it approximately follows the "Rule of 72", the bankers will know what that means..;>)

[MikeB]

Hello all,

Is there already some early stockfish benchmark results of the newly released IPad Pro 10.5 ?

rgds

I will be able to tell you in a week or so... the estimate is 30% faster than the last iPad Pro released ...

https://browser.geekbench.com/v4/cpu/3036382

https://browser.geekbench.com/v4/cpu/3048924

The key here is to look at the overall integer score which is the best proxy for how well a processor will do with chess engines.

30% integer gain is certainly true with the single core geekbench scores, but the multi core interger score ( new iPad has 6 cores) is almost DOUBLE?! - so this could be a MONSTER 5,000,000 node per second 16 oz machine in multi core mode - we're have to wait and see... to put in in perspective, if true, about 100x speed of the $3500 Revelation or about ~550 ELO stronger for 1/5 of the cost ( wifi only model)

[MikeB]

Because Mohammed did a good job in rewriting Asmfish from C++ into assembly.

tha qualifies for one of the understatements of the year, "amazing" is probably more like it.. . ;>)

[Hai]

Well apple says that it is 30% faster in computing.
If that is true then the benchmark should be around 3224kn/s

(30% up from 2480kn/s iPad pro 9.7)

I just hope the komodo team can create komodo for ios. I would love to use it in stockfish app.

And with asmFish it should be ~ 4200 kn/s

Seriously? Why is asmFish faster than stockfish (3230 vs 2480)?

Because Mohammed did a good job in rewriting Asmfish from C++ into assembly.

CFish is also faster that Stockfish (rewrite from C++ to C). But Ronald got tired of keeping CFish current and it did not get the testing effort like Asmfish.

So asmfish is stronger than Stockfish for pic?

Correct - every 1% gain might give you about 1.4 extra ELO if you believe doubling speed is worth 100 ELO (1.01^70 = ~2; 100/70 = ~1.4 ELO).

haha - just noticed that it approximately follows the "Rule of 72", the bankers will know what that means..;>)

It is very important to have asmFish, from 3000 to ->4000 kn/s, on ios.

[Hai]

Well apple says that it is 30% faster in computing.
If that is true then the benchmark should be around 3224kn/s

(30% up from 2480kn/s iPad pro 9.7)

I just hope the komodo team can create komodo for ios. I would love to use it in stockfish app.

And with asmFish it should be ~ 4200 kn/s

Seriously? Why is asmFish faster than stockfish (3230 vs 2480)?

Because Mohammed did a good job in rewriting Asmfish from C++ into assembly.

CFish is also faster that Stockfish (rewrite from C++ to C). But Ronald got tired of keeping CFish current and it did not get the testing effort like Asmfish.

So asmfish is stronger than Stockfish for pic?

Correct - every 1% gain might give you about 1.4 extra ELO if you believe doubling speed is worth 100 ELO (1.01^70 = ~2; 100/70 = ~1.4 ELO).

haha - just noticed that it approximately follows the "Rule of 72", the bankers will know what that means..;>)

It is very important to have asmFish, from 3000 to ->4000 kn/s, on ios.

Great news, improvements after first ipad pro http://talkchess.com/forum/viewtopic.ph ... 0&start=60 : (integer single core +1/3) integer multi-core ~doubled! = It should have now ~5000 kn/s.

Someone must create asmfish for ios, (to be used in "SmallFish / SmallChess for ios" for example). AsmFish 30% faster = ~6500 kn/s on ios devices .

[Frank Quisinsky]

Hi Michael,

I am very happy with my OnePlus 3T and waiting of OnePlus 5 (successor). My OnePlus 3T is very very fast with 6Gb RAM and Snapdragon 821. And after around an half hour or an hour CPU is running with 94% ... only 6% lost with all cores. Perfect for chess ... I am using Samsung Tab A 2016 with OctoCore 1.6 processor too. Also very fast for chess.

But how I can make a bench test with more cores. Made only Antutu and OnePlus 3T have 161.500 here.

I am using Droidfish with Stockfish 8, Texel 1.06, Hakkapeliitta 3.0, Senpai 1.0 und ordered Chiron 4.

Sure, that the new OnePlus 5 (comming soon) will have a higher perferformance as Apple hardware.

Best
Frank

PS: Today in the garden I made a draw in C53 after 38 moves tournament conditions (40 in 120 moves) on OnePlus 3T vs. Hakkapellitta 3.0 1 core. Very happy about it. A good day ... tomorrow the second game with black pieces.

[tpoppins]

Someone would have to write an Armfish, which would be a huge effort.

https://github.com/lantonov/asmFish/tree/master/armFish
https://github.com/lantonov/asmFish/tre ... tables/ARM

[h1a8]

I just got iPad Pro 10.5 and Stockfish gets an incredible 4220kn/s after 1 min.
The problem is that this isn't Stockfish 8 but an earlier version.
I'm going to use iPad Pro primarily for chess analysis on the go.
Currently, what's the strongest engine for iOS for analysis?
Does SmallFish use the 64bit version of Stockfish or the 32bit version? And does it use all the cores?

I'm waiting patiently to buy Komodo for iOS when it's released. I prefer Komodo to Stockfish.

[Lion]

Hello,

Why asmfish ? Is it better/stronger than Stockfish ?

rgds