Hi Jim,
I've almost entirely stopped working on Myrddin, but I wouldn't complain if you made one of your magic compiles that are 10-20% faster than mine.
https://github.com/JVMerlino/Myrddin
Many thanks in advance, but certainly not expecting you to spend your time on a (nearly) dead engine.
jm
New J.A Compiles
Moderators: hgm, chrisw, Rebel
-
- Posts: 1385
- Joined: Wed Mar 08, 2006 10:15 pm
- Location: San Francisco, California
-
- Posts: 1792
- Joined: Fri Jul 14, 2006 7:56 am
- Location: London, England
- Full name: Jim Ablett
Re: New J.A Compiles
Hi John,JVMerlino wrote: ↑Tue Aug 06, 2024 9:14 pm Hi Jim,
I've almost entirely stopped working on Myrddin, but I wouldn't complain if you made one of your magic compiles that are 10-20% faster than mine.
https://github.com/JVMerlino/Myrddin
Many thanks in advance, but certainly not expecting you to spend your time on a (nearly) dead engine.
jm
Nice to see Myrddin still being worked on after all this time
https://github.com/JVMerlino/Myrddin
Myrddin 0.90 by John Merlino
Myrddin 0.90 64 JA
Windows & Linux clang profiled builds
https://drive.proton.me/urls/BHH42XH5QC#VAj09zBD1N08
The src code has been ported to Linux about 95% (smp is disabled for Linux build)
Jim.
-
- Posts: 48
- Joined: Mon Sep 12, 2022 3:50 am
- Location: Christchurch, NZ
- Full name: Ray Bongalon
Re: New J.A Compiles
Hi Jim. May I request the latest Nemorino engine, thanks.
-
- Posts: 1792
- Joined: Fri Jul 14, 2006 7:56 am
- Location: London, England
- Full name: Jim Ablett
Re: New J.A Compiles
https://bitbucket.org/christian_g_nther ... rc/master/
Nemorino 6.11 by Christian Guenther
Nemorino 6.11 64 JA
Windows/Linux & Android clang profiled builds
https://drive.proton.me/urls/3YC9JPF0TR#N6suNAZnjrfC
Jim.
-
- Posts: 1385
- Joined: Wed Mar 08, 2006 10:15 pm
- Location: San Francisco, California
Re: New J.A Compiles
My goodness you did a LOT of stuff. I'll run a gauntlet and let you know the results.Jim Ablett wrote: ↑Wed Aug 07, 2024 11:43 pm Hi John,
Nice to see Myrddin still being worked on after all this time
https://github.com/JVMerlino/Myrddin
Myrddin 0.90 by John Merlino
Myrddin 0.90 64 JA
Windows & Linux clang profiled builds
https://drive.proton.me/urls/BHH42XH5QC#VAj09zBD1N08
The src code has been ported to Linux about 95% (smp is disabled for Linux build)
Jim.
Thank you SO MUCH, Jim. You're a wonder....
jm
-
- Posts: 1792
- Joined: Fri Jul 14, 2006 7:56 am
- Location: London, England
- Full name: Jim Ablett
Re: New J.A Compiles
https://github.com/ppipelin/radiance
Radiance 3.1.1 by Paul-Elie Pipelin
Radiance 3.1.1 64 JA
Windows/Linux & Android builds.
https://drive.proton.me/urls/VP175REEV4#4CYFkM589RrA
Jim.
-
- Posts: 48
- Joined: Mon Sep 12, 2022 3:50 am
- Location: Christchurch, NZ
- Full name: Ray Bongalon
Re: New J.A Compiles
Wonderful. Thanks again, Jim.Jim Ablett wrote: ↑Thu Aug 08, 2024 5:19 pm
https://bitbucket.org/christian_g_nther ... rc/master/
Nemorino 6.11 by Christian Guenther
Nemorino 6.11 64 JA
Windows/Linux & Android clang profiled builds
https://drive.proton.me/urls/3YC9JPF0TR#N6suNAZnjrfC
Jim.
-
- Posts: 1792
- Joined: Fri Jul 14, 2006 7:56 am
- Location: London, England
- Full name: Jim Ablett
Re: New J.A Compiles
https://github.com/PGG106/Alexandria
Alexandria 7.0.14 by Guiseppe Pagano
Alexandria 7.0.14 64 JA
Windows/Linux & Android clang profiled builds
https://drive.proton.me/urls/DWE8V6PY04#hfXXGgXIauGf
Jim.
-
- Posts: 1385
- Joined: Wed Mar 08, 2006 10:15 pm
- Location: San Francisco, California
Re: New J.A Compiles
Hi again, Jim,
Your build is definitely faster than mine. For example, from the initial position, with one core, your build gets to depth 20 in 43.98 seconds, getting 2.27M NPS. My build needs 51.64 seconds, getting only 1.97M NPS. Clear improvement!
However, and I hate to say it, and I'm still trying to track down the issue, I think there's a problem with Gaviota egtb access.
For this position, for example:
5k2/pp3N2/5K2/5B2/8/8/8/8 w - - 15 1
My build gets:
Your build gets:
Huge slowdown!
Any ideas?
Thanks again for all of your work for Myrddin, and the community!
jm
Your build is definitely faster than mine. For example, from the initial position, with one core, your build gets to depth 20 in 43.98 seconds, getting 2.27M NPS. My build needs 51.64 seconds, getting only 1.97M NPS. Clear improvement!
However, and I hate to say it, and I'm still trying to track down the issue, I think there's a problem with Gaviota egtb access.
For this position, for example:
5k2/pp3N2/5K2/5B2/8/8/8/8 w - - 15 1
My build gets:
Code: Select all
1 470 0 2 f6e6
1 496 0 21 f7d6
1 508 0 24 f7e5
2 492 0 33 f7e5?
2 488 0 98 f7e5 a7a5
3 490 1 174 f7e5 a7a5 f6e6 (10 KNPS)
3 495 1 451 f5e4 b7b5 f7e5 (28 KNPS)
3 504 1 626 f7d6! (39 KNPS)
3 508 1 659 f7d6 b7b6 f6e5 (41 KNPS)
4 524 1 1103 f7d6! (68 KNPS)
4 524 1 1199 f7d6 a7a5 d6b7 a5a4 (74 KNPS)
5 540 1 1942 f7d6! (121 KNPS)
5 544 1 2656 f7d6 b7b6 d6c8 b6b5 c8a7 (166 KNPS)
6 528 1 6012 f7d6? (375 KNPS)
6 516 1 14484 f7d6 b7b6 d6c8 b6b5 c8a7 b5b4 (905 KNPS)
7 532 3 21944 f7d6! (707 KNPS)
7 548 3 21991 f7d6! (709 KNPS)
7 564 3 23321 f7d6 b7b6 d6c8 b6b5 c8a7 b5b4 a7c6 (752 KNPS)
8 580 3 35530 f7d6! (1146 KNPS)
8 596 3 35560 f7d6! (1147 KNPS)
8 32720 3 35830 f7d6 b7b6 d6c8 b6b5 c8a7 (1155 KNPS)(Mate in 24)
9 32720 3 38397 f7d6 b7b6 d6c8 b6b5 c8a7 (1238 KNPS)(Mate in 24)
10 32720 4 44037 f7d6 b7b6 d6c8 b6b5 c8a7 (936 KNPS)(Mate in 24)
11 32720 4 58315 f7d6 b7b6 d6c8 b6b5 c8a7 (1240 KNPS)(Mate in 24)
12 32720 6 86955 f7d6 b7b6 d6c8 b6b5 c8a7 (1380 KNPS)(Mate in 24)
13 32720 11 156548 f7d6 b7b6 d6c8 b6b5 c8a7 (1423 KNPS)(Mate in 24)
14 32720 17 305376 f7d6 b7b6 d6c8 b6b5 c8a7 (1775 KNPS)(Mate in 24)
15 32720 28 603081 f7d6 b7b6 d6c8 b6b5 c8a7 (2146 KNPS)(Mate in 24)
16 32720 46 1182560 f7d6 b7b6 d6c8 b6b5 c8a7 (2521 KNPS)(Mate in 24)
17 32732 81 2294648 f7d6 a7a5 f5e6 a5a4 e6f7 a4a3 d6b5 a3a2 f7a2 (2822 KNPS)(Mate in 18)
Code: Select all
1 476 0 2 f6e6
1 502 0 21 f7d6
1 514 0 24 f7e5
2 498 0 32 f7e5?
2 495 0 97 f7e5 a7a5
3 497 0 173 f7e5 a7a5 f6e6
3 501 1 450 f5e4 b7b5 f7e5 (30 KNPS)
3 511 1 625 f7d6! (41 KNPS)
3 515 1 658 f7d6 b7b6 f6e5 (43 KNPS)
4 524 1 1179 f7d6 a7a5 d6b7 a5a4 (78 KNPS)
5 540 1 1917 f7d6! (127 KNPS)
5 544 1 2632 f7d6 b7b6 d6c8 b6b5 c8a7 (175 KNPS)
6 528 1 6012 f7d6? (400 KNPS)
6 516 1 14699 f7d6 b7b6 d6c8 b6b5 c8a7 b5b4 (979 KNPS)
7 532 3 22156 f7d6! (714 KNPS)
7 548 3 22209 f7d6! (716 KNPS)
7 564 3 23623 f7d6 b7b6 d6c8 b6b5 c8a7 b5b4 a7c6 (762 KNPS)
8 580 4 36236 f7d6! (770 KNPS)
8 596 4 36266 f7d6! (771 KNPS)
8 32720 6 36536 f7d6 b7b6 d6c8 b6b5 c8a7 (589 KNPS)(Mate in 24)
9 32720 6 39107 f7d6 b7b6 d6c8 b6b5 c8a7 (630 KNPS)(Mate in 24)
10 32720 10 44793 f7d6 b7b6 d6c8 b6b5 c8a7 (410 KNPS)(Mate in 24)
11 32720 20 59192 f7d6 b7b6 d6c8 b6b5 c8a7 (291 KNPS)(Mate in 24)
12 32720 37 88076 f7d6 b7b6 d6c8 b6b5 c8a7 (225 KNPS)(Mate in 24)
13 32720 87 157746 f7d6 b7b6 d6c8 b6b5 c8a7 (180 KNPS)(Mate in 24)
14 32720 198 307330 f7d6 b7b6 d6c8 b6b5 c8a7 (154 KNPS)(Mate in 24)
15 32720 401 601717 f7d6 b7b6 d6c8 b6b5 c8a7 (149 KNPS)(Mate in 24)
16 32726 887 1203412 f7d6 b7b6 f5d7 a7a5 d6c8 f8g8 c8b6 (135 KNPS)(Mate in 21)
17 32732 1626 2127206 f7d6 a7a5 f5e6 a5a4 e6f7 a4a3 d6b5 a3a2 f7a2 (130 KNPS)(Mate in 18)
Any ideas?
Thanks again for all of your work for Myrddin, and the community!
jm
-
- Posts: 227
- Joined: Wed Sep 19, 2007 11:07 am
- Location: Singapore
Re: New J.A Compiles
Dear Mastero Jim,
Thank you for compiling Alexandria.
There seems to be quite a difference in size of your compiled 7.0.14 bmi2 (1,654 KB) vs the official compile 7.0 bmi2 (5,246 KB). I'm curious: is something amiss?
Best wishes,
Koh, Kah Huat
Thank you for compiling Alexandria.
There seems to be quite a difference in size of your compiled 7.0.14 bmi2 (1,654 KB) vs the official compile 7.0 bmi2 (5,246 KB). I'm curious: is something amiss?
Best wishes,
Koh, Kah Huat