Introducing Igel chess engine

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

voffka
Posts: 288
Joined: Sat Jun 30, 2018 10:58 pm
Location: Ukraine
Full name: Volodymyr Shcherbyna

Re: Introducing Igel chess engine

Post by voffka »

Hello OliverBr,
OliverBr wrote: Sat Oct 10, 2020 4:01 pm I am having problems running igel on an AMD EPYC 7502P 32-Core Processor with Linux:

Code: Select all

ch@AR161:~/engines/igel/src$ make
g++ -Wall -std=c++17 -mpopcnt -pthread -msse -msse2 -msse3 -msse4.1 -mavx2 -O3 -march=native -flto *.cpp nnue/*.cpp nnue/features/*.cpp -DNDEBUG -DEVAL_NNUE=1 -D_BTYPE=0 -DUSE_SSE2=1 -DUSE_SSSE3=1 -DUSE_SSE41=1 -DUSE_AVX2=1 -o igel
ch@AR161:~/engines/igel/src$ ./igel 
Igel 2.8.0 64 POPCNT by V. Medvedev, V. Shcherbyna
go
Segmentation fault
I think for such a modern CPU you need to use AVX2 switch, e.g.,

(BMI2 compatible build):

Code: Select all

-DEVAL_NNUE=1 -DUSE_AVX2=1 -D_BTYPE=1 -DSYZYGY_SUPPORT=TRUE
(POPCNT compatible build given the fact that his is an AMD):

Code: Select all

-DEVAL_NNUE=1 -DUSE_AVX2=1 -D_BTYPE=0 -DSYZYGY_SUPPORT=TRUE
For the segmentation fault, I think the problem is that you are not specifying a NNUE file via cmd line as a parameter. For example,

Code: Select all

setoption name Threads value 1
setoption name Hash value 256
setoption name EvalFile value C:\Users\volodymyr\Downloads\Igel_2_7_0.nnue
go infinite
bastiball
Posts: 5212
Joined: Tue Oct 20, 2020 4:18 am
Full name: Basti Dangca

Re: Introducing Igel chess engine

Post by bastiball »

Can I have some SSE 4.1 + popnt for my Intel Celeron windows 10 Thanks a lot
Basti Dangca
CCRL testing group
voffka
Posts: 288
Joined: Sat Jun 30, 2018 10:58 pm
Location: Ukraine
Full name: Volodymyr Shcherbyna

Re: Introducing Igel chess engine

Post by voffka »

Hey bastiball,
bastiball wrote: Fri Oct 30, 2020 4:11 pm Can I have some SSE 4.1 + popnt for my Intel Celeron windows 10 Thanks a lot
Let me know the exact cpu code since I need to figure out it's architecture to use gcc march for it).
voffka
Posts: 288
Joined: Sat Jun 30, 2018 10:58 pm
Location: Ukraine
Full name: Volodymyr Shcherbyna

Igel Generation Networks (IGN)

Post by voffka »

In late 2020 there have been a lot of discussions in the chess community about NNUE techology in general and NNUE networks in particular raising important questions about authenticity of networks.

I have decided to dedicate time and resources to create my own class of networks that will be used for future Igel releases: Igel Generation Network (IGN).

The IGN networks comply with the following mandatory requirements:

1. Source of network data: evaluation, search, pv line is generated solely using Igel chess engine. As a starting point Igel 2.6.0 is choosen as this was the last version of Igel featuring HCE (Hand Crafted Evaluation)

2. Both data and validation data generation must be following the rule #1

3. Two versions of Igel: 2.7.0 and 2.8.0 are excluded from network training/data generation process (both for data and for validation data) because they use Dietrich Kappe's Night Nurse network as thus violate the rule #1

4. Data generation and network training must be done by myself on my own (or rented by me) hardware in order to make sure the rule #1 is not violated

5. Each generation of network must contain information (on github page) on training parameters for clarity of the source of data

6. Complete source data of network may be given to tournament organizers for validation purposes and it must be possible for external parties to train the network from scratch using the provided data to the same strength as submitted network (margin of +-10 elo)

First versions of IGN will participate in TCEC Cup7 and will be released with Igel 2.9.0 in Q4 2020.

More technical info at https://github.com/vshcherbyna/igel#ige ... tworks-ign
bastiball
Posts: 5212
Joined: Tue Oct 20, 2020 4:18 am
Full name: Basti Dangca

Re: Introducing Igel chess engine

Post by bastiball »

voffka wrote: Fri Oct 30, 2020 6:32 pm Hey bastiball,
bastiball wrote: Fri Oct 30, 2020 4:11 pm Can I have some SSE 4.1 + popnt for my Intel Celeron windows 10 Thanks a lot
Let me know the exact cpu code since I need to figure out it's architecture to use gcc march for it).
Intel(R) Celeron(R) N4100 CPU @ 1.10Ghz
Basti Dangca
CCRL testing group
bastiball
Posts: 5212
Joined: Tue Oct 20, 2020 4:18 am
Full name: Basti Dangca

Re: Introducing Igel chess engine

Post by bastiball »

bastiball wrote: Sat Oct 31, 2020 1:28 pm
voffka wrote: Fri Oct 30, 2020 6:32 pm Hey bastiball,
bastiball wrote: Fri Oct 30, 2020 4:11 pm Can I have some SSE 4.1 + popnt for my Intel Celeron windows 10 Thanks a lot
Let me know the exact cpu code since I need to figure out it's architecture to use gcc march for it).
Intel(R) Celeron(R) N4100 CPU @ 1.10Ghz SSE4.1-popcnt windows 10 64 bit
Gemini Lake
Basti Dangca
CCRL testing group
peter
Posts: 3186
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: Igel Generation Networks (IGN)

Post by peter »

voffka wrote: Fri Oct 30, 2020 11:49 pm In late 2020 there have been a lot of discussions in the chess community about NNUE techology in general and NNUE networks in particular raising important questions about authenticity of networks.

I have decided to dedicate time and resources to create my own class of networks that will be used for future Igel releases: Igel Generation Network (IGN).

The IGN networks comply with the following mandatory requirements:

1. Source of network data: evaluation, search, pv line is generated solely using Igel chess engine. As a starting point Igel 2.6.0 is choosen as this was the last version of Igel featuring HCE (Hand Crafted Evaluation)

2. Both data and validation data generation must be following the rule #1

3. Two versions of Igel: 2.7.0 and 2.8.0 are excluded from network training/data generation process (both for data and for validation data) because they use Dietrich Kappe's Night Nurse network as thus violate the rule #1

4. Data generation and network training must be done by myself on my own (or rented by me) hardware in order to make sure the rule #1 is not violated

5. Each generation of network must contain information (on github page) on training parameters for clarity of the source of data

6. Complete source data of network may be given to tournament organizers for validation purposes and it must be possible for external parties to train the network from scratch using the provided data to the same strength as submitted network (margin of +-10 elo)

First versions of IGN will participate in TCEC Cup7 and will be released with Igel 2.9.0 in Q4 2020.

More technical info at https://github.com/vshcherbyna/igel#ige ... tworks-ign
Great news, Volodymyr!
Is the version playing in TCEC Cup 7 with dkappe's net still or already with IGN?
Looking forward to the match against Stoofvlees in next round.
Go Igel, go!
:) regards
Peter.
voffka
Posts: 288
Joined: Sat Jun 30, 2018 10:58 pm
Location: Ukraine
Full name: Volodymyr Shcherbyna

Re: Igel Generation Networks (IGN)

Post by voffka »

Hello peter,
peter wrote: Mon Nov 02, 2020 10:24 am
Great news, Volodymyr!
Is the version playing in TCEC Cup 7 with dkappe's net still or already with IGN?
Looking forward to the match against Stoofvlees in next round.
Go Igel, go!
:) regards
Thanks. The one playing in TCEC Cup7 is ign-0 network: https://github.com/vshcherbyna/igel#ign-0
Jamal Bubker
Posts: 326
Joined: Mon May 24, 2010 4:32 pm

Re: Introducing Igel chess engine

Post by Jamal Bubker »

Hi !
Please someone could provide me a windows binary of Igel v2.8.0 designed for westemere architecture ?
Thanks
Greetings,
Jamal
supersharp77
Posts: 1242
Joined: Sat Jul 05, 2014 7:54 am
Location: Southwest USA

Re: Introducing Igel chess engine

Post by supersharp77 »

voffka wrote: Fri Oct 30, 2020 6:32 pm Hey bastiball,
bastiball wrote: Fri Oct 30, 2020 4:11 pm Can I have some SSE 4.1 + popnt for my Intel Celeron windows 10 Thanks a lot
Let me know the exact cpu code since I need to figure out it's architecture to use gcc march for it).
Igel v2.8 NNUE nonpopcnt x64.....Thx AR :) :wink: