It is currently Thu Mar 28, 2024 2:58 am

All times are UTC - 8 hours [ DST ]




Post new topic Reply to topic  [ 523 posts ]  Go to page Previous  1 ... 19, 20, 21, 22, 23, 24, 25 ... 27  Next
Author Message
Offline
 Post subject: Re: New android app "BadukAI"
Post #421 Posted: Thu Mar 17, 2022 6:19 pm 
Dies in gote

Posts: 51
Liked others: 0
Was liked: 0
Rank: KGS 1 dan
Thanks a lot! :bow:

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #422 Posted: Mon Mar 21, 2022 2:14 am 
Dies in gote

Posts: 51
Liked others: 0
Was liked: 0
Rank: KGS 1 dan
Haha, Katago 1.11.0 is released!
https://github.com/lightvector/KataGo/r ... ag/v1.11.0

Can you continue updating? :salute:

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #423 Posted: Mon Mar 21, 2022 2:57 pm 
Lives with ko

Posts: 174
Liked others: 13
Was liked: 151
Fortunately, this is not necessary. There are only a few commits between release 1.11 and the revision that I built for BadukAI 1.14. And these commits are mainly concerned with the code for training and with documentation, which don't matter for BadukAI. So the KataGo code in BadukAI 1.14 is practically equivalent to release 1.11.

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #424 Posted: Tue Mar 22, 2022 9:01 am 
Gosei
User avatar

Posts: 1348
Liked others: 202
Was liked: 203
in the worst case, depending on the hardware, what is the speed gain of an optimized network? and at the best?

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #425 Posted: Tue Mar 22, 2022 1:28 pm 
Lives with ko

Posts: 174
Liked others: 13
Was liked: 151
If you compare the speeds of optimized and original network, you get:

a) If you choose the optimal number of search threads (in terms of speed) for each network:
In the worst (HW dependent) case the speed gain is factor 2 (e.g. 20 playouts/second instead of 10).
In the best (HW dependent) case the speed gain is factor 24 (e.g. 240 playouts/second instead of 10).

b) If you choose one search thread for both networks:
In the worst (HW dependent) case the speed gain is factor 2.
In the best (HW dependent) case the speed gain is factor 40.

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #426 Posted: Tue Mar 22, 2022 6:27 pm 
Dies in gote

Posts: 51
Liked others: 0
Was liked: 0
Rank: KGS 1 dan
I find a compatibility issue!
I have two phones(Dimensity 720 and Snapdragon 662)
When I use 40B weight, the output of winning rate is wrong.


Attachments:
E95E81FC057E42712F6E195DB4361FF8.jpg
E95E81FC057E42712F6E195DB4361FF8.jpg [ 840.07 KiB | Viewed 36064 times ]
Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #427 Posted: Wed Mar 23, 2022 12:27 am 
Dies in gote

Posts: 51
Liked others: 0
Was liked: 0
Rank: KGS 1 dan
In my vivo y52s (Dimensity 720)
I have test all weights
Kata 10b -> ok
kata 20b -> bad
lz 15b -> ok
lz 40b ->ok


Attachments:
a.png
a.png [ 1.05 MiB | Viewed 36045 times ]
Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #428 Posted: Wed Mar 23, 2022 2:29 pm 
Lives with ko

Posts: 174
Liked others: 13
Was liked: 151
Since version 1.14 BadukAI runs the optimized KG networks (20b and 40b) on NPU instead of CPU on Mediatek processors. This works well (and fast) on Dimensity 1100 (according to user reports). But your tests show that on the (slightly older) Dimensity 720 the NPU can't handle the KG network correctly and produces garbage output. What a pity.

Since I don't know on which Mediatek processors it works and on which it doesn't, I will add a switch to KataGo's config settings in the next release, so that every user can turn this on or off depending on whether it works on his/her device.

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #429 Posted: Wed Mar 23, 2022 5:14 pm 
Dies in gote

Posts: 51
Liked others: 0
Was liked: 0
Rank: KGS 1 dan
OK,thanks :bow:

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #430 Posted: Thu Mar 24, 2022 9:46 am 
Gosei
User avatar

Posts: 1348
Liked others: 202
Was liked: 203
akigo wrote:
If you compare the speeds of optimized and original network, you get:

a) If you choose the optimal number of search threads (in terms of speed) for each network:
In the worst (HW dependent) case the speed gain is factor 2 (e.g. 20 playouts/second instead of 10).
In the best (HW dependent) case the speed gain is factor 24 (e.g. 240 playouts/second instead of 10).

b) If you choose one search thread for both networks:
In the worst (HW dependent) case the speed gain is factor 2.
In the best (HW dependent) case the speed gain is factor 40.

stunningly!!!

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #431 Posted: Fri Mar 25, 2022 7:50 am 
Dies in gote

Posts: 51
Liked others: 0
Was liked: 0
Rank: KGS 1 dan
My friend bought a new phone (Redmi K30 Pro with Dimensity 9000 chip), but the speed of kata 20b is very slow, it seems that there is no acceleration, only 20n/s.
The performance of Dimensity 9000 is stronger than that of Dimensity 1100, it should exceed 140n/s.

Is it possible to provide a new version that allows me to test different acceleration schemes. :salute:

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #432 Posted: Fri Mar 25, 2022 1:49 pm 
Lives with ko

Posts: 174
Liked others: 13
Was liked: 151
Can you share the android system log for the time period when KataGo is starting ? Then I could check for error messages which give a hint what's going wrong. Otherwise I wouldn't know what to change ...

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #433 Posted: Sat Mar 26, 2022 3:09 pm 
Lives with ko

Posts: 174
Liked others: 13
Was liked: 151
I made a new version (1.15) available on play store https://play.google.com/store/apps/details?id=net.kir.baduk_ai and github https://github.com/aki65/aki65.github.io/releases/tag/v1.15.0 with the change I announced in post #428.

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #434 Posted: Sun Mar 27, 2022 9:42 am 
Lives with ko

Posts: 174
Liked others: 13
Was liked: 151
lovego wrote:
My friend bought a new phone (Redmi K30 Pro with Dimensity 9000 chip), but the speed of kata 20b is very slow, it seems that there is no acceleration, only 20n/s.
The performance of Dimensity 9000 is stronger than that of Dimensity 1100, it should exceed 140n/s.

Is it possible to provide a new version that allows me to test different acceleration schemes. :salute:


You can do this with the version from https://github.com/aki65/aki65.github.io/releases/tag/v1.15.1 (see instructions there).

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #435 Posted: Sun Mar 27, 2022 5:35 pm 
Dies in gote

Posts: 51
Liked others: 0
Was liked: 0
Rank: KGS 1 dan
Haha, my Dimensity 720 works well when useSpecificNpuAccelerator is 3/4/5 :clap:

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #436 Posted: Mon Mar 28, 2022 2:37 am 
Dies in gote

Posts: 51
Liked others: 0
Was liked: 0
Rank: KGS 1 dan
Dimensity 9000 is about 300n/s (KG 20b) when useSpecificNpuAccelerator is 3
The fastest chip at present!
:tmbup:

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #437 Posted: Mon Mar 28, 2022 3:55 am 
Dies in gote

Posts: 51
Liked others: 0
Was liked: 0
Rank: KGS 1 dan
Oh, I get 400+n/s when numSearchThread is 3 and paralletEvalution is on

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #438 Posted: Mon Mar 28, 2022 1:36 pm 
Lives with ko

Posts: 174
Liked others: 13
Was liked: 151
It's great to see that. Thank you very much for your testing. I would like to add an entry for Dimensity 9000 in the performance table on BadukAI's homepage. Since all the visit counts in this table were measured in this position (black to play)



it would be nice if you could check the speed of Dimensity 9000 in this position.

It's important to compare speeds only in the same position, since the type of position affects the visit count: In a highly symmetrical position (for example an empty board), the search tree will include a lot of positions which are mirrors of already visited positions. KataGo will handle a visit of such a position by cache lookup instead of computation, so the visit count will be higher.

By the way, correct move is E3 (not too difficult, I am told, but already beyond my level), so also a good way to check if computations are correct.

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #439 Posted: Mon Mar 28, 2022 5:42 pm 
Dies in gote

Posts: 51
Liked others: 0
Was liked: 0
Rank: KGS 1 dan
OK, I'll let my friend try.
Another thing, in version 1.16, the LZ engine crashes when the analysis is paused, and the speed is very slow on Dimensity 9000, can LZ choose the NPU accelerator to use?

Top
 Profile  
 
Offline
 Post subject: Re: New android app "BadukAI"
Post #440 Posted: Mon Mar 28, 2022 7:02 pm 
Dies in gote

Posts: 51
Liked others: 0
Was liked: 0
Rank: KGS 1 dan
Regarding board recognition, I found a good solution!
Use yolov3 to identify black stone, white stone, corners (no stone), sidelines (no stone) and empty position.

https://blog.csdn.net/wyzegg/article/details/108029789
https://github.com/zhuoyiyao97/YOLO-GO


Attachments:
20200815234608841.jpeg
20200815234608841.jpeg [ 548.11 KiB | Viewed 35527 times ]
Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 523 posts ]  Go to page Previous  1 ... 19, 20, 21, 22, 23, 24, 25 ... 27  Next

All times are UTC - 8 hours [ DST ]


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group