Can someone tell me if they have some kind of difference? The Android compiling from Windows is still under experimentation, on my device it works. If everything is ok I'll put a tutorial on how to do it.
Can someone tell me if they have some kind of difference? The Android compiling from Windows is still under experimentation, on my device it works. If everything is ok I'll put a tutorial on how to do it.
Hello Mr. Goi, from what I saw, the link goes to PC engines platform and not the Android engines , yes it was Corchess but nor for Android,
R u sure ur in the right section of topic's ?
Greetings
Can someone tell me if they have some kind of difference? The Android compiling from Windows is still under experimentation, on my device it works. If everything is ok I'll put a tutorial on how to do it.
There is no difference between them same version of clang and equal speed.
Windows
► Show Spoiler
Linux
► Show Spoiler
Skynet's build are over 10% faster than both PC NDK Version.
Compiled using Termux+GCC-11
Can someone tell me if they have some kind of difference? The Android compiling from Windows is still under experimentation, on my device it works. If everything is ok I'll put a tutorial on how to do it.
Hello Mr. Goi, from what I saw, the link goes to PC engines platform and not the Android engines , yes it was Corchess but nor for Android,
R u sure ur in the right section of topic's ?
Greetings
Last edited by kramnik on 20/08/2021, 6:03, edited 1 time in total.
Can someone tell me if they have some kind of difference? The Android compiling from Windows is still under experimentation, on my device it works. If everything is ok I'll put a tutorial on how to do it.
There is no difference between them same version of clang and equal speed.
Windows
► Show Spoiler
Linux
► Show Spoiler
Skynet's build are over 10% faster than both PC NDK Version.
Compiled using Termux+GCC-11
kramnik wrote: ↑20/08/2021, 6:01
Thank you very much for the hint. So the Termux compilation is the most effective...
The only downside of doing this on termux (with -march=native flag) is, it only runs on the same type of cpu, which could be result in compatibility issues when distributing it for other devices. Most effective, when compiling on the same machine where the chess engine is used at the end.
So they succeeded to really have a native compiler for Android, not a sandbox. That's great!
The only problem with Termux is that GCC is deprecated there. No, there is another one: the official Termux repository seems to be down:
One .ore thing: if the -march=native flag is avoided we'll have a binary working with other Android devices?
kramnik wrote: ↑20/08/2021, 6:01
Thank you very much for the hint. So the Termux compilation is the most effective...
The only downside of doing this on termux (with -march=native flag) is, it only runs on the same type of cpu, which could be result in compatibility issues when distributing it for other devices. Most effective, when compiling on the same machine where the chess engine is used at the end.
kramnik wrote: ↑20/08/2021, 23:13
So they succeeded to really have a native compiler for Android, not a sandbox. That's great!
The only problem with Termux is that GCC is deprecated there. No, there is another one: the official Termux repository seems to be down:
One .ore thing: if the -march=native flag is avoided we'll have a binary working with other Android devices?
kramnik wrote: ↑20/08/2021, 6:01
Thank you very much for the hint. So the Termux compilation is the most effective...
The only downside of doing this on termux (with -march=native flag) is, it only runs on the same type of cpu, which could be result in compatibility issues when distributing it for other devices. Most effective, when compiling on the same machine where the chess engine is used at the end.
Playstore Version is outdated we are using Fdroid Version.
kramnik wrote: ↑20/08/2021, 23:13
One .ore thing: if the -march=native flag is avoided we'll have a binary working with other Android devices?
I'm not the expert for the command line parameters for gcc, but something like -march=armv8-a came into my mind. Beside that, there also exists other parameters like -mtune and -mcpu, which can be combined. Studying the help file of gcc will help.