I never intended to use other networks. I just wanted a compiling experiment for fun.voffka wrote: ↑Wed Apr 28, 2021 10:58 am Hello Gabor,
It is possible to adjust those things in makefie and embed other networks in Igel as long as architecture matches. But keep in mind that the end result will not be Igel engine - it will be something else as other networks will not qualify for Igel Generation Network rules I mentioned some time ago (https://github.com/vshcherbyna/igel#ige ... tworks-ign).
At the same time Igel is open source and anyone can do anything with it as long as GPL3 license is not violatedPlease reach out to me via PM to explain in more details what you are trying to do
, perhaps I can help.
Introducing Igel chess engine
Moderator: Ras
-
Gabor Szots
- Posts: 1594
- Joined: Sat Jul 21, 2018 7:43 am
- Location: Budapest, Hungary
- Full name: Gabor Szots
Re: Introducing Igel chess engine
Gabor Szots
CCRL testing group
CCRL testing group
-
voffka
- Posts: 297
- Joined: Sat Jun 30, 2018 10:58 pm
- Location: Ukraine
- Full name: Volodymyr Shcherbyna
Re: Introducing Igel chess engine
Hello Gabor,
1. Download another network and modify it's path inside makefile: https://github.com/vshcherbyna/igel/blo ... kefile#L12
2. Or you download the other network, and rename it as ign-1-d593efbd and put inside weights folder. You can avoid modification to makefile in this case
Then it is very simple, you can either:Gabor Szots wrote: ↑Wed Apr 28, 2021 11:07 am I never intended to use other networks. I just wanted a compiling experiment for fun.
1. Download another network and modify it's path inside makefile: https://github.com/vshcherbyna/igel/blo ... kefile#L12
2. Or you download the other network, and rename it as ign-1-d593efbd and put inside weights folder. You can avoid modification to makefile in this case
-
voffka
- Posts: 297
- Joined: Sat Jun 30, 2018 10:58 pm
- Location: Ukraine
- Full name: Volodymyr Shcherbyna
Re: Introducing Igel chess engine
Hello Ed,
Regarding the old compiles, I do not generate them intentionally - I believe AVX2 covers vast majority of machines produced in last 9 years and it is fast therefore good to be used in rating lists. If there are enthusiasts or rating lists who want a custom old compile - I support them too and use gcc march option to provide them the fastest compile for their cpu architecture.
Work on Igel is basically going 24/7 in parallel tracks to improve search and eval with network data generation happening around the clock on OVH rented hardware.
Regarding the old compiles, I do not generate them intentionally - I believe AVX2 covers vast majority of machines produced in last 9 years and it is fast therefore good to be used in rating lists. If there are enthusiasts or rating lists who want a custom old compile - I support them too and use gcc march option to provide them the fastest compile for their cpu architecture.
-
Gabor Szots
- Posts: 1594
- Joined: Sat Jul 21, 2018 7:43 am
- Location: Budapest, Hungary
- Full name: Gabor Szots
Re: Introducing Igel chess engine
Ah, so I can't make the network choice a UCI parameter (that's what I wanted to do).voffka wrote: ↑Wed Apr 28, 2021 11:24 am Hello Gabor,
Then it is very simple, you can either:Gabor Szots wrote: ↑Wed Apr 28, 2021 11:07 am I never intended to use other networks. I just wanted a compiling experiment for fun.
1. Download another network and modify it's path inside makefile: https://github.com/vshcherbyna/igel/blo ... kefile#L12
2. Or you download the other network, and rename it as ign-1-d593efbd and put inside weights folder. You can avoid modification to makefile in this case
Gabor Szots
CCRL testing group
CCRL testing group
-
voffka
- Posts: 297
- Joined: Sat Jun 30, 2018 10:58 pm
- Location: Ukraine
- Full name: Volodymyr Shcherbyna
Re: Introducing Igel chess engine
Hello Gabor,
Since Igel is open source enthusiast people can compile custom network in as well via makefile or cmake option like it is explained on the main github page.
I removed the EvalOption in Igel 3.0.0 as seen in the change logs https://github.com/vshcherbyna/igel/releases/tag/3.0.0. The main reason to remove it was because many people had all sort of issues with it when configuring in the UI so I wanted something simpler and basically have an Igel exe that just simply works without any configs.Gabor Szots wrote: ↑Wed Apr 28, 2021 11:47 am Ah, so I can't make the network choice a UCI parameter (that's what I wanted to do).
Since Igel is open source enthusiast people can compile custom network in as well via makefile or cmake option like it is explained on the main github page.
-
cpeters
- Posts: 254
- Joined: Wed Feb 17, 2021 7:44 pm
- Full name: Christian Petersen
Re: Introducing Igel chess engine
Hello voffka!
Do you plan to implement the 'knowledge' of the wrong bishop in endgames for the engine itself or will it rely on bases solely?
Greetings
Do you plan to implement the 'knowledge' of the wrong bishop in endgames for the engine itself or will it rely on bases solely?
Greetings
-
voffka
- Posts: 297
- Joined: Sat Jun 30, 2018 10:58 pm
- Location: Ukraine
- Full name: Volodymyr Shcherbyna
Re: Introducing Igel chess engine
Hello Christian,
Thanks for reaching out. I may look into it, but most of the tournaments are using tbs, so there is little justification I see on how important this is. Can you please tell your use-case? You are playing Igel on a phone device or without tbs and you can draw it?
-
AlexChess
- Posts: 1563
- Joined: Sat Feb 06, 2021 8:06 am
- Full name: Alex Morales
Re: Introducing Igel chess engine
Hi Voffka!
Just completed test with Igel 3.0.0 and 3.0.5 is out
Please, compile it also for Windows 10 ARM64

Regards, AlexChess
Just completed test with Igel 3.0.0 and 3.0.5 is out
Please, compile it also for Windows 10 ARM64

Regards, AlexChess
Chess engines and dedicated chess computers fan since 1981
macOS Sequoia 16GB-512GB, Windows 11 & Ubuntu ARM64.
ProteusSF Dev Forum
ProteusSF Dev Forum
-
Rebel
- Posts: 7603
- Joined: Thu Aug 18, 2011 12:04 pm
- Full name: Ed Schröder
Re: Igel 3.0.5
Seems about right.voffka wrote: ↑Wed Apr 28, 2021 2:15 am
Igel 3.0.5 at https://github.com/vshcherbyna/igel/releases/tag/3.0.5 (official executable binaries for Windows and IGN net).
This release brings significant improvements in search, time management and evaluation. I accidentally found a strong patch in search which brought around 12 elo in LTC in one shot. Overall strength increase due to search/time management improvements is around 20 elo, strength increase due to new network is 20 elo more. There are also some improvements in SMP mode that were quite good in short time controls: 15 elo, but hard to test on long time control as it will require a lot of computational power.
Code: Select all
No. Engine 1 2 3 4 5 6 7 8 Score Games Perc
------------------------------------------------------------------------------------------
1 Stockfish 12 xxxx 62.0 79.5 84.5 81.5 86.5 88.5 87.0 569.5 / 700 (81.36%)
2 Komodo Dragon 38.0 xxxx 69.0 81.0 76.5 78.0 77.0 80.0 499.5 / 700 (71.36%)
3 Komodo 14 20.5 31.0 xxxx 62.5 58.0 58.5 62.0 62.0 354.5 / 700 (50.64%)
4 RubiChess 2.1 15.5 19.0 37.5 xxxx 54.0 53.0 56.0 58.0 293.0 / 700 (41.86%)
5 SlowChess 2.5 18.5 23.5 42.0 46.0 xxxx 53.5 46.0 55.5 285.0 / 700 (40.71%)
6 Igel 3.0.5 13.5 22.0 41.5 47.0 46.5 xxxx 49.5 54.5 274.5 / 700 (39.21%)
7 Ethereal 12.75 11.5 23.0 38.0 44.0 54.0 50.5 xxxx 50.5 271.5 / 700 (38.79%)
8 Igel 3.0.0 13.0 20.0 38.0 42.0 44.5 45.5 49.5 xxxx 252.5 / 700 (36.07%)90% of coding is debugging, the other 10% is writing bugs.
-
cpeters
- Posts: 254
- Joined: Wed Feb 17, 2021 7:44 pm
- Full name: Christian Petersen
