Test between SV 384x30-T60-3010 and Stockfish2020.06.24 popc

[corres]

I made a test between SV-384x30-T60-3010 and Stockfish-dev.2020.06.24. popcount.
One of aim of the test was to check the stability of my new system.
My system is:
OS - Windows 10 64 bits
Processor - Ryzen 9 3950x 16 x 4.0 GHz, no power boost, no SMT(=HT)
RAM - 16 GB 3000 MHz XMP level 2
kTRIAD - 1 x RTX 2080 Ti Turbo OC + 2 x RTX 2060 OC
Used Leela - lcv0-ver.0.25.1 (Kiudee params)
Endgame database - 6 men SYZYGY and 5 men Nalimov (for chess GUI only)
Chess GUI - Fritz 14
Opening book - My MiniBook with 100 opening position
Time 1 min + 2 sec / move
Games started from positions with alternated colors.

Result
kTRIAD Leela ver.0.25.1 SV 384x30-T60-3010 : Stockfish20200624 popc = 21 : 10 (169 draw) 200 games.

The test also prove for Leela the closed openings are better even for Stockfish the Open and (mainly) the Half-Open openings are the better,

[perejaslav]

Can you upload the played games?

[Milos]

I made a test between SV-384x30-T60-3010 and Stockfish-dev.2020.06.24. popcount.
One of aim of the test was to check the stability of my new system.
My system is:
OS - Windows 10 64 bits
Processor - Ryzen 9 3950x 16 x 4.0 GHz, no power boost, no SMT(=HT)
RAM - 16 GB 3000 MHz XMP level 2
kTRIAD - 1 x RTX 2080 Ti Turbo OC + 2 x RTX 2060 OC
Used Leela - lcv0-ver.0.25.1 (Kiudee params)
Endgame database - 6 men SYZYGY and 5 men Nalimov (for chess GUI only)
Chess GUI - Fritz 14
Opening book - My MiniBook with 100 opening position
Time 1 min + 2 sec / move
Games started from positions with alternated colors.

Result
kTRIAD Leela ver.0.25.1 SV 384x30-T60-3010 : Stockfish20200624 popc = 21 : 10 (169 draw) 200 games.

The test also prove for Leela the closed openings are better even for Stockfish the Open and (mainly) the Half-Open openings are the better,

By disallowing SF to use HT on your particular machine you basically reduce its strength for at least 20Elo.
What is power boost? Why is all core frequency only 4GHz, when 3950 could easily run on 4.25GHz with a decent cooling?

[corres]

I made a test between SV-384x30-T60-3010 and Stockfish-dev.2020.06.24. popcount.
One of aim of the test was to check the stability of my new system.
My system is:
OS - Windows 10 64 bits
Processor - Ryzen 9 3950x 16 x 4.0 GHz, no power boost, no SMT(=HT)
RAM - 16 GB 3000 MHz XMP level 2
kTRIAD - 1 x RTX 2080 Ti Turbo OC + 2 x RTX 2060 OC
Used Leela - lcv0-ver.0.25.1 (Kiudee params)
Endgame database - 6 men SYZYGY and 5 men Nalimov (for chess GUI only)
Chess GUI - Fritz 14
Opening book - My MiniBook with 100 opening position
Time 1 min + 2 sec / move
Games started from positions with alternated colors.

Result
kTRIAD Leela ver.0.25.1 SV 384x30-T60-3010 : Stockfish20200624 popc = 21 : 10 (169 draw) 200 games.

The test also prove for Leela the closed openings are better even for Stockfish the Open and (mainly) the Half-Open openings are the better,

By disallowing SF to use HT on your particular machine you basically reduce its strength for at least 20Elo.
What is power boost? Why is all core frequency only 4GHz, when 3950 could easily run on 4.25GHz with a decent cooling?

I never use HT/SMT and power boost because these decrease the the reliability of the test.
I set the clock of CPU on 4.0 GHz because my former machine (2950x) also was set on 4.0 GHz and I wanted to make comparison between the two PCs, using CPU chess bench. About the bench result i wrote in an another post in which I report the enhancement of parameters of 3950x to 4.4 GHz base CPU clock.
If you are Interested in that post, please search for it.
http://talkchess.com/forum3/viewtopic.php?f=2&t=74435

[Milos]

I made a test between SV-384x30-T60-3010 and Stockfish-dev.2020.06.24. popcount.
One of aim of the test was to check the stability of my new system.
My system is:
OS - Windows 10 64 bits
Processor - Ryzen 9 3950x 16 x 4.0 GHz, no power boost, no SMT(=HT)
RAM - 16 GB 3000 MHz XMP level 2
kTRIAD - 1 x RTX 2080 Ti Turbo OC + 2 x RTX 2060 OC
Used Leela - lcv0-ver.0.25.1 (Kiudee params)
Endgame database - 6 men SYZYGY and 5 men Nalimov (for chess GUI only)
Chess GUI - Fritz 14
Opening book - My MiniBook with 100 opening position
Time 1 min + 2 sec / move
Games started from positions with alternated colors.

Result
kTRIAD Leela ver.0.25.1 SV 384x30-T60-3010 : Stockfish20200624 popc = 21 : 10 (169 draw) 200 games.

The test also prove for Leela the closed openings are better even for Stockfish the Open and (mainly) the Half-Open openings are the better,

By disallowing SF to use HT on your particular machine you basically reduce its strength for at least 20Elo.
What is power boost? Why is all core frequency only 4GHz, when 3950 could easily run on 4.25GHz with a decent cooling?

I never use HT/SMT and power boost because these decrease the the reliability of the test.
I set the clock of CPU on 4.0 GHz because my former machine (2950x) also was set on 4.0 GHz and I wanted to make comparison between the two PCs, using CPU chess bench. About the bench result i wrote in an another post in which I report the enhancement of parameters of 3950x to 4.4 GHz base CPU clock.
If you are Interested in that post, please search for it.
http://talkchess.com/forum3/viewtopic.php?f=2&t=74435

Ok I see, you wanted a hardware comparison test. What I was interested in, was whether strongest current SF (SFdev) on 3950x can be on par with strongest T60 based Lc0 on like 2x2080Ti, since your tri config is close to that.

[corres]

I made a test between SV-384x30-T60-3010 and Stockfish-dev.2020.06.24. popcount.
One of aim of the test was to check the stability of my new system.
My system is:
OS - Windows 10 64 bits
Processor - Ryzen 9 3950x 16 x 4.0 GHz, no power boost, no SMT(=HT)
RAM - 16 GB 3000 MHz XMP level 2
kTRIAD - 1 x RTX 2080 Ti Turbo OC + 2 x RTX 2060 OC
Used Leela - lcv0-ver.0.25.1 (Kiudee params)
Endgame database - 6 men SYZYGY and 5 men Nalimov (for chess GUI only)
Chess GUI - Fritz 14
Opening book - My MiniBook with 100 opening position
Time 1 min + 2 sec / move
Games started from positions with alternated colors.

Result
kTRIAD Leela ver.0.25.1 SV 384x30-T60-3010 : Stockfish20200624 popc = 21 : 10 (169 draw) 200 games.

The test also prove for Leela the closed openings are better even for Stockfish the Open and (mainly) the Half-Open openings are the better,

By disallowing SF to use HT on your particular machine you basically reduce its strength for at least 20Elo.
What is power boost? Why is all core frequency only 4GHz, when 3950 could easily run on 4.25GHz with a decent cooling?

I never use HT/SMT and power boost because these decrease the the reliability of the test.
I set the clock of CPU on 4.0 GHz because my former machine (2950x) also was set on 4.0 GHz and I wanted to make comparison between the two PCs, using CPU chess bench. About the bench result i wrote in an another post in which I report the enhancement of parameters of 3950x to 4.4 GHz base CPU clock.
If you are Interested in that post, please search for it.
http://talkchess.com/forum3/viewtopic.php?f=2&t=74435

Ok I see, you wanted a hardware comparison test. What I was interested in, was whether strongest current SF (SFdev) on 3950x can be on par with strongest T60 based Lc0 on like 2x2080Ti, since your tri config is close to that.

Yes, the 2 x RTX 2060 is stronger than 1 x RTX 2080 Ti about 10 % only.

[corres]

Can you upload the played games?

You can download it from
wikisend.com
File Id 845200
Password leela