Stockfish 1.9 JA update available

[NATIONAL12]

i know i am being greedy here but any chance of 12 or 16 compile like JA did for 1.7.1

[alpha123]

i know i am being greedy here but any chance of 12 or 16 compile like JA did for 1.7.1

For my Gaviota Tablebases version I upped the max cores to 24, mostly cause of you.

As soon as I can manage to compile it on Windows, I'll release it.

Peter

[swami]

Stockfish 1.9:
Single Core




https://sites.google.com/site/strategic ... sts-1---12
1200 Positions
10 seconds per position
Hardware: Q6600, 32 bits, 2 GB RAM, 2.4 GHZ. Arena 2.01 GUI.

• Huge Improvement in STS 7 -- "Offer of Simplification" (+11)
  Great Improvement in STS 9 -- "Advancement of a/b/c Pawns" (+9)
  Some Improvement in STS 3 -- "Knight Outposts/Centralization/Re-positioning" (+4)

Stockfish 1.8:
Single core



Best wishes,
Swami

[mcostalba]

Thanks Swami !

Actually we also have some small decreases:

Code: Select all

STS  6 -->  -1
STS  8 -->  -2
STS 10 -->  -3

But I guess difference is just below noise level and means nothing...

[swami]

Thanks Swami !

Actually we also have some small decreases:

Code: Select all

STS  6 -->  -1
STS  8 -->  -2
STS 10 -->  -3

But I guess difference is just below noise level and means nothing...

Yes, the difference of <= 3 in each epd should be ignored unless it comes in several coloumns. In this case, it doesn't so it can therefore be ignored.

My criteria to determine whether the engine has improved in certain area is: atleast +4 in any of the epd's as well as the increase in total scores (out of 1000).

The 1.9 plays much better than 1.8 in the games also, I have tested it a bit.

Good luck, and thanks!

Best wishes,
Swami

[peter]

[

• Huge Improvement in STS 7 -- "Offer of Simplification" (+11)
  Great Improvement in STS 9 -- "Advancement of a/b/c Pawns" (+9)
  Some Improvement in STS 3 -- "Knight Outposts/Centralization/Re-positioning" (+4)

Hi Swami!

Thanks for documentation!
Could you please give us an "official" Elo- rating from this too?

Not that I'd consider that of any importance but some folks do and for my personal point of view Elo isn't sacro sanct neither, so different ways of testing may well lead to different ratings.
All we'd need was an ability to see relationships between different strengths and weaknesses of different engines, more than a one and only measurement that changes anyhow with every new development of hard- and software.
P.S. Any news about Philippe Gailhac's tool, Stefan Meyer-Kahlen did contact you or him, didn't he?

Ceterum censeo...

[swami]

[

• Huge Improvement in STS 7 -- "Offer of Simplification" (+11)
  Great Improvement in STS 9 -- "Advancement of a/b/c Pawns" (+9)
  Some Improvement in STS 3 -- "Knight Outposts/Centralization/Re-positioning" (+4)

Hi Swami!

Thanks for documentation!
Could you please give us an "official" Elo- rating from this too?

Not that I'd consider that of any importance but some folks do and for my personal point of view Elo isn't sacro sanct neither, so different ways of testing may well lead to different ratings.
All we'd need was an ability to see relationships between different strengths and weaknesses of different engines, more than a one and only measurement that changes anyhow with every new development of hard- and software.
P.S. Any news about Philippe Gailhac's tool, Stefan Meyer-Kahlen did contact you or him, didn't he?

Ceterum censeo...

Hi Peter,

I doubt STS Elo estimate 1.1 could calculate the ratings based on STS scores 1 to 12. It was designed to only calculate for STS 1 to 10. With the new version, it certainly will. I'm also interested to know how much elo this translates to.

When I get back home, I will enter the STS 1 to 10 datas into STS Elo estimate 1.1 and see what the change is.

I haven't heard anything lately from Gailhac. I think he must be busy. I didn't receive any messages from Stefan either. Perhaps Gailhac might have received it.

I agree that there are several ways to measure the elo's.

Best regards,
Swami

[swami]

Hi Peter,

My guess is that it has improved by over 60 Elo. STS Elo Estimate certainly showed more than 60 elo change between STS Total scores of Stockfish 1.8 (3124) and Stockfish 1.9 (3210). It could be anywhere between 40 and 60 elo... It could even be more than 60 though. We could say 40-50 just to be on safe side.

Best regards,
Swami

[gerold]

Thanks Swami !

Actually we also have some small decreases:

Code: Select all

STS  6 -->  -1
STS  8 -->  -2
STS 10 -->  -3

But I guess difference is just below noise level and means nothing...

Thanks again for the new Stockfish. M.C.
After 52 games at 12sec.move. Stockfish 1.9 is 15 elo higher than
Stockfish 1.8. To few games yet. I will test at longer TC now.

Best,
Gerold.

[Don]

Hi Peter,

My guess is that it has improved by over 60 Elo. STS Elo Estimate certainly showed more than 60 elo change between STS Total scores of Stockfish 1.8 (3124) and Stockfish 1.9 (3210). It could be anywhere between 40 and 60 elo... It could even be more than 60 though. We could say 40-50 just to be on safe side.

Best regards,
Swami

Over 60 ELO puts stockfish above Rybka 4! If this is true, congratulations! Stockfish now is stronger than all the clones.