relase Bouquet 1.5 x32,x64,pop64(SSE4.2)

[velmarin]

I relase Bouquet 1.5 x32,x64,pop64(SSE4.2)


Download
https://sites.google.com/site/chessbouq ... ess-engine



libraries (if needed):
https://sites.google.com/site/chessbouq ... -microsoft

Code cleanup, some things did not work.
Some things added.
Tests, many tests.

Greetings, Chema.

[geots]

I relase Bouquet 1.5 x32,x64,pop64(SSE4.2)


Download
https://sites.google.com/site/chessbouq ... ess-engine



libraries (if needed):
https://sites.google.com/site/chessbouq ... -microsoft

Code cleanup, some things did not work.
Some things added.
Tests, many tests.

Greetings, Chema.

Thanks, my friend. Look- don't let me confuse you here. I am not talking about 1.5 that you are releasing here. I am speaking of 1.4, that has been out a while.

I have this intel i5, so naturally I use the 64bit popcnt for 1.4 and 1.5. However- follow me here- I also have another quad. It is an Intel XP- but it cannot use sse and popcnt. I am talking only about version 1.4. You have a 64bit exe for 1.4 that is not popcnt, and it will run in the xp system. BUT then you also have a zip file that has a Visual C 64bit exe that you say is for XP systems. Well it will run in my XP system also, but so will the other 64bit non-popcnt I mentioned. I ran them in console mode and the nps is almost the same between the 2.

So for 1.4, in the xp system, do I use the Visual C 64bit exe or do I use the 64bit exe that is in the folder with the popcnt exe. Or does it matter?


Thanks,

george

[Ajedrecista]

Hi Chema!

I relase Bouquet 1.5 x32,x64,pop64(SSE4.2)


Download
https://sites.google.com/site/chessbouq ... ess-engine



libraries (if needed):
https://sites.google.com/site/chessbouq ... -microsoft

Code cleanup, some things did not work.
Some things added.
Tests, many tests.

Greetings, Chema.

What have you done to Bouquet? Please take a look on Bob's test at Chess2U Forum:

http://www.chess2u.com/t5876-bouquet-new-engine#37027

Code: Select all

Games Completed = 1000 of 1000 (Avg game length = 16.116 sec)Settings = RR/32MB/2000ms+100ms/M 700000cp for 7 moves, D 150 moves/EPD:E:\LittleBlitzer\openings.epd(4000)Time = 4449 sec elapsed, 0 sec remaining

1.  Bouquet _1.5 x64            660.5/1000   440-119-441     (L: m=119 t=0 i=0 a=0)   (D: r=276 i=77 f=54 s=1 a=33)   (tpm=110.8 d=13.03 nps=1427017)

2.  Bouquet _1.4 x64            339.5/1000   119-440-441     (L: m=440 t=0 i=0 a=0)   (D: r=276 i=77 f=54 s=1 a=33)   (tpm=115.8 d=12.43 nps=1284123)

User Bob usually reports very trustable tests (for example, giving +10 to SF 2.3 over SF 2.2.2):

http://www.chess2u.com/t4410p375-stockfish#36955

Code: Select all

Games Completed = 1000 of 1000 (Avg game length = 15.186 sec)Settings = RR/32MB/2000ms+100ms/M 700000cp for 7 moves, D 150 moves/EPD:E:\LittleBlitzer\openings.epd(4000)Time = 4009 sec elapsed, 0 sec remaining

1.  Stockfish 2.3 JA 64bit      515.0/1000   267-237-496     (L: m=237 t=0 i=0 a=0)   (D: r=380 i=67 f=26 s=6 a=17)   (tpm=113.3 d=15.57 nps=1348166)

2.  Stockfish 2.2.2 JA 64bit    485.0/1000   237-267-496     (L: m=267 t=0 i=0 a=0)   (D: r=380 i=67 f=26 s=6 a=17)   (tpm=113.6 d=15.29 nps=1373859)

Running my own statistical Fortran 95 programme LOS_and_Elo_uncertainties_calculator:

Code: Select all

LOS_and_Elo_uncertainties_calculator, ® 2012.

----------------------------------------------------------------
Calculation of Elo uncertainties in a match between two engines:
----------------------------------------------------------------

(The input and output data is referred to the first engine).

Please write down non-negative integers.

Maximum number of games supported: 2147483647.

Write down the number of wins (up to 1825361100):

440

Write down the number of loses (up to 1825361100):

119

Write down the number of draws (up to 2147483088):

441

 Write down the confidence level (in percentage) between 65% and 99.9% (it will be rounded up to 0.01%):

95

Write down the clock rate of the CPU (in GHz), only for timing the elapsed time of the calculations:

3

---------------------------------------
Elo interval for 95.00 % confidence:

Elo rating difference:    115.61 Elo

Lower rating difference:   99.63 Elo
Upper rating difference:  132.08 Elo

Lower bound uncertainty:  -15.98 Elo
Upper bound uncertainty:   16.47 Elo
Average error:        +/-  16.23 Elo

K = (average error)*[sqrt(n)] =  513.13

Elo interval: ]  99.63,  132.08[
---------------------------------------

Number of games of the match:      1000
Score: 66.05 %
Elo rating difference:  115.61 Elo
Draw ratio: 44.10 %

*********************************************************
Standard deviation:  2.0926 % of the points of the match.
*********************************************************

 Error bars were calculated with two-sided tests; values are rounded up to 0.01 Elo, or 0.01 in the case of K.

-------------------------------------------------------------------
Calculation of likelihood of superiority (LOS) in a one-sided test:
-------------------------------------------------------------------

LOS (taking into account draws) is always calculated, if possible.

LOS &#40;not taking into account draws&#41; is only calculated if wins + loses < 16001.

LOS &#40;average value&#41; is calculated only when LOS &#40;not taking into account draws&#41; is calculated.
______________________________________________

LOS&#58; 100.00 % &#40;taking into account draws&#41;.
LOS&#58; 100.00 % &#40;not taking into account draws&#41;.
LOS&#58; 100.00 % &#40;average value&#41;.
______________________________________________

These values of LOS are rounded up to 0.01%

End of the calculations. Approximated elapsed time&#58;   75 ms.

Thanks for using LOS_and_Elo_uncertainties_calculator. Press Enter to exit.

Around +116 ± 16 with 95% confidence! Maybe this test is one of the few that are out of the 95% confidence interval? I mean, can it be one of the few tests that brings strange results? Or this huge Elo gain can be confirmed by others? In any case: CONGRATULATIONS! Sigue así.

Regards from Spain.

Ajedrecista.

[geots]

Don't need the answer- you were a step ahead of me as is usual. The regular 64bit was asking for a file- so obviously you knew that and that is why you so kindly put the visual C exe 64bit in there for me.

And the non-popcnt 64bit exe for 1.5 works fine in the XP box, and of course the popcnt works fine as always in my Intel i5.

You are too good to us with this great program- I appreciate you and your work.

You know, I may just get me an Intel i7 with 6 cores sooner than I think. I will network it to this 4core i5, with 10 cores running off the same monitor. Or get a bigger one than 24 widescreen that shows 4 cores at once.

I feel like spending some of that 2 million bucks my wife got out of a settlement 2 weeks ago. And tax free on top of that.

I guess I could also keep using this XP quad- it's a speed demon. And that would give me 14 cores to play with.

Nitey- day my friend. Maybe I am after sleep now. Got 5 RRs running. BTW, when yours comes up I assume you want me to go with the latest version.

A g- A t - an s, and you have gts

[geots]

Hi Chema!

I relase Bouquet 1.5 x32,x64,pop64(SSE4.2)


Download
https://sites.google.com/site/chessbouq ... ess-engine



libraries (if needed):
https://sites.google.com/site/chessbouq ... -microsoft

Code cleanup, some things did not work.
Some things added.
Tests, many tests.

Greetings, Chema.

What have you done to Bouquet? Please take a look on Bob's test at Chess2U Forum:

http://www.chess2u.com/t5876-bouquet-new-engine#37027

Code: Select all

Games Completed = 1000 of 1000 (Avg game length = 16.116 sec)Settings = RR/32MB/2000ms+100ms/M 700000cp for 7 moves, D 150 moves/EPD:E:\LittleBlitzer\openings.epd(4000)Time = 4449 sec elapsed, 0 sec remaining

1.  Bouquet _1.5 x64            660.5/1000   440-119-441     (L: m=119 t=0 i=0 a=0)   (D: r=276 i=77 f=54 s=1 a=33)   (tpm=110.8 d=13.03 nps=1427017)

2.  Bouquet _1.4 x64            339.5/1000   119-440-441     (L: m=440 t=0 i=0 a=0)   (D: r=276 i=77 f=54 s=1 a=33)   (tpm=115.8 d=12.43 nps=1284123)

User Bob usually reports very trustable tests (for example, giving +10 to SF 2.3 over SF 2.2.2):

http://www.chess2u.com/t4410p375-stockfish#36955

Code: Select all

Games Completed = 1000 of 1000 (Avg game length = 15.186 sec)Settings = RR/32MB/2000ms+100ms/M 700000cp for 7 moves, D 150 moves/EPD:E:\LittleBlitzer\openings.epd(4000)Time = 4009 sec elapsed, 0 sec remaining

1.  Stockfish 2.3 JA 64bit      515.0/1000   267-237-496     (L: m=237 t=0 i=0 a=0)   (D: r=380 i=67 f=26 s=6 a=17)   (tpm=113.3 d=15.57 nps=1348166)

2.  Stockfish 2.2.2 JA 64bit    485.0/1000   237-267-496     (L: m=267 t=0 i=0 a=0)   (D: r=380 i=67 f=26 s=6 a=17)   (tpm=113.6 d=15.29 nps=1373859)

Running my own statistical Fortran 95 programme LOS_and_Elo_uncertainties_calculator:

Code: Select all

LOS_and_Elo_uncertainties_calculator, ® 2012.

----------------------------------------------------------------
Calculation of Elo uncertainties in a match between two engines:
----------------------------------------------------------------

(The input and output data is referred to the first engine).

Please write down non-negative integers.

Maximum number of games supported: 2147483647.

Write down the number of wins (up to 1825361100):

440

Write down the number of loses (up to 1825361100):

119

Write down the number of draws (up to 2147483088):

441

 Write down the confidence level (in percentage) between 65% and 99.9% (it will be rounded up to 0.01%):

95

Write down the clock rate of the CPU (in GHz), only for timing the elapsed time of the calculations:

3

---------------------------------------
Elo interval for 95.00 % confidence:

Elo rating difference:    115.61 Elo

Lower rating difference:   99.63 Elo
Upper rating difference:  132.08 Elo

Lower bound uncertainty:  -15.98 Elo
Upper bound uncertainty:   16.47 Elo
Average error:        +/-  16.23 Elo

K = (average error)*[sqrt(n)] =  513.13

Elo interval: ]  99.63,  132.08[
---------------------------------------

Number of games of the match:      1000
Score: 66.05 %
Elo rating difference:  115.61 Elo
Draw ratio: 44.10 %

*********************************************************
Standard deviation:  2.0926 % of the points of the match.
*********************************************************

 Error bars were calculated with two-sided tests; values are rounded up to 0.01 Elo, or 0.01 in the case of K.

-------------------------------------------------------------------
Calculation of likelihood of superiority (LOS) in a one-sided test:
-------------------------------------------------------------------

LOS (taking into account draws) is always calculated, if possible.

LOS &#40;not taking into account draws&#41; is only calculated if wins + loses < 16001.

LOS &#40;average value&#41; is calculated only when LOS &#40;not taking into account draws&#41; is calculated.
______________________________________________

LOS&#58; 100.00 % &#40;taking into account draws&#41;.
LOS&#58; 100.00 % &#40;not taking into account draws&#41;.
LOS&#58; 100.00 % &#40;average value&#41;.
______________________________________________

These values of LOS are rounded up to 0.01%

End of the calculations. Approximated elapsed time&#58;   75 ms.

Thanks for using LOS_and_Elo_uncertainties_calculator. Press Enter to exit.

Around +116 ± 16 with 95% confidence! Maybe this test is one of the few that are out of the 95% confidence interval? I mean, can it be one of the few tests that brings strange results? Or this huge Elo gain can be confirmed by others? In any case: CONGRATULATIONS! Sigue así.

Regards from Spain.

Ajedrecista.

Jesus, nothing you tell me Velmarin did would be a surprise to me. The guy makes hard tasks look simple. He can do it in his sleep.


Best,
a g a t and an s- you got gts

[velmarin]

Thank you both, I do not think the gain is much,
littlebitter is a great tool, I use it for testing from the same engine,
with different engines with these super fast test,
startup not all are equally fast.
Among different engines prefer Fritz to "10" seconds, Fritz starts with the two engines already loaded.

This relase is compiled with Intel's new tools Compiler Version 2013
I've done a lot of cleaning things wrong code and I added all possible intrinsic functions (or alternative code)

George, do not quite understand, my machine is an AMD Phenom II,
There I try the 64-bit executable popcnt.
Then I have two virtual machines with Windows XP, 32 and 64 bits, so I try the other executables before publishing.

Of all forms seem to put the title on the page changed, and this corrected.
Thanks, Chema.

[geots]

Thank you both, I do not think the gain is much,
littlebitter is a great tool, I use it for testing from the same engine,
with different engines with these super fast test,
startup not all are equally fast.
Among different engines prefer Fritz to "10" seconds, Fritz starts with the two engines already loaded.

This relase is compiled with Intel's new tools Compiler Version 2013
I've done a lot of cleaning things wrong code and I added all possible intrinsic functions (or alternative code)

George, do not quite understand, my machine is an AMD Phenom II,
There I try the 64-bit executable popcnt.
Then I have two virtual machines with Windows XP, 32 and 64 bits, so I try the other executables before publishing.


Nothing to worry about. I got popcnt running in the intel i5, and the non-popcnt running in the xp system. All is well- and sleepy time.


Best,

george

Of all forms seem to put the title on the page changed, and this corrected.
Thanks, Chema.

[WuShock]

Bouquet smelling sweeter all the time !!

Bouquet 1.5 x32,x64,pop64(SSE4.2) working well enough on my old i7 - 940 that it is giving Rybka 4.1 a run for it's money...

Only 26 games so far , Bouquet is + 1 , and just held a tough draw.......

Tom

[Bram Visser]

On my 32 bit AMD dual core & Arena 3.0 Bouquet 1.5 crashes ... .

[WuShock]

i7-940 @ 2.93 , 2 threads ea , klo_openings , no egtb , ponder off
Time Control : 1 min + 1 sec

[/img]