| | Ivanhoe v68 (32 + 64 bit): yes another new one | |
| | Author | Message |
---|
me4u Admin
Posts : 819 Reputation : 2737 Join date : 2010-01-05 Location : Pugatory
| Subject: Ivanhoe v68 (32 + 64 bit): yes another new one Sat Jan 30, 2010 9:27 pm | |
| | |
| | | "B" Admin
Posts : 1201 Reputation : 4018 Join date : 2010-01-17 Location : undetermined
| Subject: Re: Ivanhoe v68 (32 + 64 bit): yes another new one Sat Jan 30, 2010 9:34 pm | |
| | |
| | | "B" Admin
Posts : 1201 Reputation : 4018 Join date : 2010-01-17 Location : undetermined
| Subject: Re: Ivanhoe v68 (32 + 64 bit): yes another new one Sat Jan 30, 2010 11:42 pm | |
| the V68_JR = RUBBlSH ,,WEAKEST IvanHoe yet, no idea about peter pans version as there are several apllicationx and no instructions as to whats for what any help appreciated....preety please :cherry: | |
| | | Angels77 Admin
Posts : 548 Reputation : 1719 Join date : 2009-11-24 Location : awaiting land
| Subject: experimental patch to v68 ???? Sun Jan 31, 2010 12:36 am | |
| experimental patch to v68 ... re SPLIT_ABISSO 1 - 1 of 1 row selected - rows selected - clear theprez theprez 15 minutes ago Adds SplitDepth ("SPLIT_ABISSO") to UCI.
patch text below the cut
-- diff -rupN 68/all_node.c 68.2/all_node.c --- 68/all_node.c 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/all_node.c 2010-01-30 17:37:26.865252483 -0600 @@ -25,7 +25,7 @@ int MyAll (typePOS* POSITION, int VALUE, uint64 zob = POSITION->DYN->HASH; int to, fr; boolean SPLIT; - int SPLIT_ABISSO; + int SPLIT_ABISSO; YUSUF_DECLARE ();
if (VALUE < -VALUE_MATE + 1) @@ -129,12 +129,12 @@ int MyAll (typePOS* POSITION, int VALUE, NextMove->exclude = 0; v = VALUE; SPLIT = FALSE; - SPLIT_ABISSO = 14; + SPLIT_ABISSO = 14;
while (TRUE) { #if 1 && defined (YUSUF_MULTICORE) - if (TITANIC_MODE && SMP_FREE != 0 && depth >= SPLIT_ABISSO && + if (TITANIC_MODE && SMP_FREE != 0 && depth >= SPLIT_DEPTH && !SPLIT && NextMove->phase != TRANS && cnt >= 1 && NextMove->phase <= ORDINARY_MOVES) { diff -rupN 68/cut_node.c 68.2/cut_node.c --- 68/cut_node.c 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/cut_node.c 2010-01-30 17:47:50.757253067 -0600 @@ -24,7 +24,7 @@ int MyCut (typePOS* POSITION, int VALUE, uint64 zob = POSITION->DYN->HASH; int to, fr; boolean SPLIT; - int SPLIT_ABISSO; + int SPLIT_ABISSO; YUSUF_DECLARE ();
if (VALUE < -VALUE_MATE + 1) @@ -162,12 +162,12 @@ int MyCut (typePOS* POSITION, int VALUE, v = VALUE;
SPLIT = FALSE; - SPLIT_ABISSO = 16; + SPLIT_ABISSO = 16;
while (TRUE) { #if 1 && defined (YUSUF_MULTICORE) - if (TITANIC_MODE && SMP_FREE != 0 && depth >= SPLIT_ABISSO && + if (TITANIC_MODE && SMP_FREE != 0 && depth >= SPLIT_DEPTH && !SPLIT && NextMove->phase != TRANS && cnt >= 1 && NextMove->phase <= ORDINARY_MOVES) { diff -rupN 68/exclude_node.c 68.2/exclude_node.c --- 68/exclude_node.c 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/exclude_node.c 2010-01-30 17:52:48.522252926 -0600 @@ -25,7 +25,7 @@ int MyExclude (typePOS* POSITION, int VA int to, fr; int REDUCTION; boolean SPLIT; - int SPLIT_ABISSO; + int SPLIT_ABISSO; YUSUF_DECLARE ();
if (VALUE < -VALUE_MATE + 1) @@ -127,7 +127,7 @@ int MyExclude (typePOS* POSITION, int VA NextMove->exclude = MOVE & 0x7fff; v = VALUE; SPLIT = FALSE; - SPLIT_ABISSO = 14; + SPLIT_ABISSO = 14;
while (TRUE) { diff -rupN 68/input.c 68.2/input.c --- 68/input.c 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/input.c 2010-01-30 17:56:02.293252466 -0600 @@ -113,6 +113,17 @@ static void SetOption (typePOS* POSITION } }
+ if ( !memcmp ( string, "SplitDepth", 10 ) ) { + v = strstr ( string, "value" ); + + if ( v ) { + Value = atoi ( v + 6 ); + SPLIT_DEPTH = Value; + + SEND ( "Split Depth %d\n", Value ); + } + } + #ifdef CON_ROBBO_BUILD if (!memcmp (string, "RobboTripleBaseDirectory", 24)) { @@ -271,6 +282,7 @@ static void uci () SEND ("option name TitanicCPUsSizings (Capitalist!) type spin min 1 max %d default 1\n", MAX_CPUS); SEND ("option name Ponder type check default false\n"); SEND ("option name VerifyNullMove type check default true\n"); + SEND ("option name SplitDepth type spin min 1 max 30 default 14\n"); SEND ("option name AlternativeTimeUsage (beta) type check default false\n"); SEND ("option name BufferTime (milliseconds) type spin min 0 max 10000 default 0\n"); #ifdef CON_ROBBO_BUILD diff -rupN 68/main.c 68.2/main.c --- 68/main.c 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/main.c 2010-01-30 17:58:04.428502282 -0600 @@ -97,7 +97,7 @@ int main () InitMaterialValue (); InitStatic (); NewGame (ROOT_POSIZIONE0, TRUE); - + SPLIT_DEPTH = 14; DEFAULT SEND_CURR_MOVE = FALSE; USE_ALTERNATIVE_TIME = FALSE; BUFFER_TIME = 0; diff -rupN 68/pv_node.c 68.2/pv_node.c --- 68/pv_node.c 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/pv_node.c 2010-01-30 18:01:50.605502473 -0600 @@ -22,7 +22,7 @@ int MyPV (typePOS* POSITION, int ALPHA, typeDYNAMIC* POS0 = POSITION->DYN; int SINGULAR; boolean SPLIT; - int SPLIT_ABISSO; + int SPLIT_ABISSO; YUSUF_DECLARE ();
if (BETA < -VALUE_MATE) @@ -197,14 +197,14 @@ int MyPV (typePOS* POSITION, int ALPHA, good_move = 0;
SPLIT = FALSE; - SPLIT_ABISSO = 14; + //SPLIT_ABISSO = 14; while (TRUE) { #if 1 && defined (YUSUF_MULTICORE) /* nein check ? */ /* depth ya 10 ? */ /* Personal Opinion: with the *increase* for the depth ! */ /* now what is the best? decembrist testing for best params to find it! */ if (TITANIC_MODE && SMP_FREE != 0 && !check && - depth >= SPLIT_ABISSO && !SPLIT && best_value != -VALUE_INFINITY) + depth >= SPLIT_DEPTH && !SPLIT && best_value != -VALUE_INFINITY) { int r; boolean b; diff -rupN 68/RobboLito.h 68.2/RobboLito.h --- 68/RobboLito.h 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/RobboLito.h 2010-01-30 18:02:40.405252575 -0600 @@ -129,6 +129,7 @@ typePawnEval *PawnHash; boolean TITANIC_MODE; uint64 volatile CPUS_SIZE; uint64 NODE_CHECK; +uint64 volatile SPLIT_DEPTH;
typedef struct {uint64 h1, h2;} YUSUF128; #define YUSUF_DECLARE() /* */ permalink _________________ nothing is ever truly lost , just misplaced and awaiting us
| |
| | | Angels77 Admin
Posts : 548 Reputation : 1719 Join date : 2009-11-24 Location : awaiting land
| Subject: WTF !!!!! Sun Jan 31, 2010 12:42 am | |
| [img] [/img] _________________ nothing is ever truly lost , just misplaced and awaiting us
| |
| | | Angels77 Admin
Posts : 548 Reputation : 1719 Join date : 2009-11-24 Location : awaiting land
| Subject: Re: Ivanhoe v68 (32 + 64 bit): yes another new one Sun Jan 31, 2010 1:59 am | |
| WlTHOUT ROBBO BASES PETER PANS VERSlON = CRAP 10 GAMES VRS RYBKA 3 (965 PHENOM BLACK ) same books +1 =5 -4 = CRAP LAST 2 VERSlONS DlD NOT NEED ROBBO BASES TO EQUAL DEEP RYBKA3 ERGO V68 = CRAP _________________ nothing is ever truly lost , just misplaced and awaiting us
| |
| | | Angels77 Admin
Posts : 548 Reputation : 1719 Join date : 2009-11-24 Location : awaiting land
| Subject: 2ND PATCH ???? Sun Jan 31, 2010 2:10 am | |
| Adds SplitDepth ("SPLIT_ABISSO") to UCI.
patch text below the cut
-- diff -rupN 68/all_node.c 68.2/all_node.c --- 68/all_node.c 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/all_node.c 2010-01-30 17:37:26.865252483 -0600 @@ -25,7 +25,7 @@ int MyAll (typePOS* POSITION, int VALUE, uint64 zob = POSITION->DYN->HASH; int to, fr; boolean SPLIT; - int SPLIT_ABISSO; + int SPLIT_ABISSO; YUSUF_DECLARE ();
if (VALUE < -VALUE_MATE + 1) @@ -129,12 +129,12 @@ int MyAll (typePOS* POSITION, int VALUE, NextMove->exclude = 0; v = VALUE; SPLIT = FALSE; - SPLIT_ABISSO = 14; + SPLIT_ABISSO = 14;
while (TRUE) { #if 1 && defined (YUSUF_MULTICORE) - if (TITANIC_MODE && SMP_FREE != 0 && depth >= SPLIT_ABISSO && + if (TITANIC_MODE && SMP_FREE != 0 && depth >= SPLIT_DEPTH && !SPLIT && NextMove->phase != TRANS && cnt >= 1 && NextMove->phase <= ORDINARY_MOVES) { diff -rupN 68/cut_node.c 68.2/cut_node.c --- 68/cut_node.c 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/cut_node.c 2010-01-30 17:47:50.757253067 -0600 @@ -24,7 +24,7 @@ int MyCut (typePOS* POSITION, int VALUE, uint64 zob = POSITION->DYN->HASH; int to, fr; boolean SPLIT; - int SPLIT_ABISSO; + int SPLIT_ABISSO; YUSUF_DECLARE ();
if (VALUE < -VALUE_MATE + 1) @@ -162,12 +162,12 @@ int MyCut (typePOS* POSITION, int VALUE, v = VALUE;
SPLIT = FALSE; - SPLIT_ABISSO = 16; + SPLIT_ABISSO = 16;
while (TRUE) { #if 1 && defined (YUSUF_MULTICORE) - if (TITANIC_MODE && SMP_FREE != 0 && depth >= SPLIT_ABISSO && + if (TITANIC_MODE && SMP_FREE != 0 && depth >= SPLIT_DEPTH && !SPLIT && NextMove->phase != TRANS && cnt >= 1 && NextMove->phase <= ORDINARY_MOVES) { diff -rupN 68/exclude_node.c 68.2/exclude_node.c --- 68/exclude_node.c 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/exclude_node.c 2010-01-30 17:52:48.522252926 -0600 @@ -25,7 +25,7 @@ int MyExclude (typePOS* POSITION, int VA int to, fr; int REDUCTION; boolean SPLIT; - int SPLIT_ABISSO; + int SPLIT_ABISSO; YUSUF_DECLARE ();
if (VALUE < -VALUE_MATE + 1) @@ -127,7 +127,7 @@ int MyExclude (typePOS* POSITION, int VA NextMove->exclude = MOVE & 0x7fff; v = VALUE; SPLIT = FALSE; - SPLIT_ABISSO = 14; + SPLIT_ABISSO = 14;
while (TRUE) { diff -rupN 68/input.c 68.2/input.c --- 68/input.c 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/input.c 2010-01-30 17:56:02.293252466 -0600 @@ -113,6 +113,17 @@ static void SetOption (typePOS* POSITION } }
+ if ( !memcmp ( string, "SplitDepth", 10 ) ) { + v = strstr ( string, "value" ); + + if ( v ) { + Value = atoi ( v + 6 ); + SPLIT_DEPTH = Value; + + SEND ( "Split Depth %d\n", Value ); + } + } + #ifdef CON_ROBBO_BUILD if (!memcmp (string, "RobboTripleBaseDirectory", 24)) { @@ -271,6 +282,7 @@ static void uci () SEND ("option name TitanicCPUsSizings (Capitalist!) type spin min 1 max %d default 1\n", MAX_CPUS); SEND ("option name Ponder type check default false\n"); SEND ("option name VerifyNullMove type check default true\n"); + SEND ("option name SplitDepth type spin min 1 max 30 default 14\n"); SEND ("option name AlternativeTimeUsage (beta) type check default false\n"); SEND ("option name BufferTime (milliseconds) type spin min 0 max 10000 default 0\n"); #ifdef CON_ROBBO_BUILD diff -rupN 68/main.c 68.2/main.c --- 68/main.c 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/main.c 2010-01-30 17:58:04.428502282 -0600 @@ -97,7 +97,7 @@ int main () InitMaterialValue (); InitStatic (); NewGame (ROOT_POSIZIONE0, TRUE); - + SPLIT_DEPTH = 14; DEFAULT SEND_CURR_MOVE = FALSE; USE_ALTERNATIVE_TIME = FALSE; BUFFER_TIME = 0; diff -rupN 68/pv_node.c 68.2/pv_node.c --- 68/pv_node.c 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/pv_node.c 2010-01-30 18:01:50.605502473 -0600 @@ -22,7 +22,7 @@ int MyPV (typePOS* POSITION, int ALPHA, typeDYNAMIC* POS0 = POSITION->DYN; int SINGULAR; boolean SPLIT; - int SPLIT_ABISSO; + int SPLIT_ABISSO; YUSUF_DECLARE ();
if (BETA < -VALUE_MATE) @@ -197,14 +197,14 @@ int MyPV (typePOS* POSITION, int ALPHA, good_move = 0;
SPLIT = FALSE; - SPLIT_ABISSO = 14; + //SPLIT_ABISSO = 14; while (TRUE) { #if 1 && defined (YUSUF_MULTICORE) /* nein check ? */ /* depth ya 10 ? */ /* Personal Opinion: with the *increase* for the depth ! */ /* now what is the best? decembrist testing for best params to find it! */ if (TITANIC_MODE && SMP_FREE != 0 && !check && - depth >= SPLIT_ABISSO && !SPLIT && best_value != -VALUE_INFINITY) + depth >= SPLIT_DEPTH && !SPLIT && best_value != -VALUE_INFINITY) { int r; boolean b; diff -rupN 68/RobboLito.h 68.2/RobboLito.h --- 68/RobboLito.h 2010-01-29 07:05:00.000000000 -0600 +++ 68.2/RobboLito.h 2010-01-30 18:02:40.405252575 -0600 @@ -129,6 +129,7 @@ typePawnEval *PawnHash; boolean TITANIC_MODE; uint64 volatile CPUS_SIZE; uint64 NODE_CHECK; +uint64 volatile SPLIT_DEPTH;
typedef struct {uint64 h1, h2;} YUSUF128; #define YUSUF_DECLARE() /* */ _________________ nothing is ever truly lost , just misplaced and awaiting us
| |
| | | Angels77 Admin
Posts : 548 Reputation : 1719 Join date : 2009-11-24 Location : awaiting land
| Subject: Re: Ivanhoe v68 (32 + 64 bit): yes another new one Sun Jan 31, 2010 2:16 am | |
| _________________ nothing is ever truly lost , just misplaced and awaiting us
| |
| | | Angels77 Admin
Posts : 548 Reputation : 1719 Join date : 2009-11-24 Location : awaiting land
| Subject: Re: Ivanhoe v68 (32 + 64 bit): yes another new one Sun Jan 31, 2010 7:07 am | |
| - jazztime wrote:
- Admin i don't know what is wrong with your PC and what time control settings you are using.
I also don't know whether you enable AlternativeTimeusage or not.
But obviously you don't know what you are doing.
I have tested Peterpans and Jr 's versions and the are BOTH doing very well against Rybka 3.
Please recheck your system settings or buy a new pc :) watch you mouth or look to another forum.. sfa wrong wih our pc lve just creamed 2 top gms wih it ones an i7 the others an overclocked xeon, PLUS we watch to make sure the engine games are not lost by time problems and weve tested all the v68 versions now .. (NOT with Robbo bases ) and only one managed to win a single game vrs deep rykba3 we then retested v81 and v70 vrs deep rykba3 both won lost and drew the same amount of games and finished equal to rykba all 2+1 games same books ** and several others have also messaged there disgust at v68 :geek: _________________ nothing is ever truly lost , just misplaced and awaiting us
| |
| | | Bettina Admin
Posts : 914 Reputation : 3742 Join date : 2009-11-27 Location : Bettina ( Sobel ? ) speakeasy 77>> Up All Night 77 >> Chelsea 77 >> little venice, maida vale 77 >> ???? Wendy D Green & Adnan Khashoggi know .... DO YOU ? £10,000.00 reward for information leading to me . but i am NOT here !
| Subject: Re: Ivanhoe v68 (32 + 64 bit): yes another new one Sun Jan 31, 2010 7:25 am | |
| _________________ Bettina (Sobel ? ) speakeasy 77>> Up all Night 77 >>Chelsea 77 >> little venice 77 /Maida Vale >>> ???? Wendy D Green & Adnan Kashoggi know .... DO YOU ? £2,000.00 reward for information leading to me . but i am NOT here ! | |
| | | pernicus
Posts : 9 Reputation : 5 Join date : 2010-01-22 Location : uranus
| Subject: Re: Ivanhoe v68 (32 + 64 bit): yes another new one Sun Jan 31, 2010 6:22 pm | |
| - Admin wrote:
- WlTHOUT ROBBO BASES PETER PANS VERSlON = CRAP
10 GAMES VRS RYBKA 3 (965 PHENOM BLACK ) same books +1 =5 -4 = CRAP
LAST 2 VERSlONS DlD NOT NEED ROBBO BASES TO EQUAL DEEP RYBKA3 ERGO V68 = CRAP
10 games is no where near enough to give a true statistical difference between two engines that are even on remotely similar playing fields | |
| | | Angels77 Admin
Posts : 548 Reputation : 1719 Join date : 2009-11-24 Location : awaiting land
| Subject: Re: Ivanhoe v68 (32 + 64 bit): yes another new one Mon Feb 01, 2010 2:11 am | |
| - pernicus wrote:
- Admin wrote:
- WlTHOUT ROBBO BASES PETER PANS VERSlON = CRAP
10 GAMES VRS RYBKA 3 (965 PHENOM BLACK ) same books +1 =5 -4 = CRAP
LAST 2 VERSlONS DlD NOT NEED ROBBO BASES TO EQUAL DEEP RYBKA3 ERGO V68 = CRAP
10 games is no where near enough to give a true statistical difference between two engines that are even on remotely similar playing fields when an engine loses 10 -0 that engine is not going to win next 10 games ,and i suggest you read all weve posted here ..weve hested each version of v68 and over very many concluded that WlTHOUT Robbo base v68 is the WEAKEST of ALL Ippolit , RobboLito IvanHoe ans Fire Bird engines . _________________ nothing is ever truly lost , just misplaced and awaiting us
| |
| | | me4u Admin
Posts : 819 Reputation : 2737 Join date : 2010-01-05 Location : Pugatory
| Subject: Re: Ivanhoe v68 (32 + 64 bit): yes another new one Mon Feb 01, 2010 2:22 am | |
| sent from another forum
yes your correct ........ when looking to analyze its games many were clearly draws it had lost because of time ,i had allready seen it "lock uo" and "ponder " for tull 2 minutes of some games and some games were "loops " of 20 and 50 moves each . noticed a group of 8 of these and when checked all were draws it had lost on time , | |
| | | wajit@cfl.rr.com V.I.P.MEMBER
Posts : 199 Reputation : 366 Join date : 2010-01-06
| Subject: Re: Ivanhoe v68 (32 + 64 bit): yes another new one Mon Feb 01, 2010 4:46 am | |
| Yes there are reports of time loss issues with the new ivanhoe v68. This can be found on the immortal forums russian site. The problem i have is in recent testings show that pretty much all ivanhoe engines are basically no better than rybka3. Coming from a group that supposedly made an engine better than rybka 3. i.e. robbolito and ippolit. Which this engine is based on. But what i dont understand is why they changed the code so much that its now only just as good as Rybka3 and not better. The +50 elo beyond Rybka3 has vanished. Igorrit is a much better engine than Ivanhoe. Maybe because is was the same code that made robbolito so great. Anyway i wish they would stop putting out these weaker engines and work with what they already know to be strong and that is Igorrit. | |
| | | Angels77 Admin
Posts : 548 Reputation : 1719 Join date : 2009-11-24 Location : awaiting land
| Subject: Re: Ivanhoe v68 (32 + 64 bit): yes another new one Mon Feb 01, 2010 5:01 am | |
| - wajit@cfl.rr.com wrote:
- Yes there are reports of time loss issues with the new ivanhoe v68. This can be found on the immortal forums russian site. The problem i have is in recent testings show that pretty much all ivanhoe engines are basically no better than rybka3. Coming from a group that supposedly made an engine better than rybka 3. i.e. robbolito and ippolit. Which this engine is based on. But what i dont understand is why they changed the code so much that its now only just as good as Rybka3 and not better. The +50 elo beyond Rybka3 has vanished. Igorrit is a much better engine than Ivanhoe. Maybe because is was the same code that made robbolito so great.
Anyway i wish they would stop putting out these weaker engines and work with what they already know to be strong and that is Igorrit. we agree and would hope charie and others here might think to come together and make the engine what it can easily be if the " comrades " continue on these weaker engines that NEED to rely on Robbo bases , we believe that the Iggporit compiles are the "way" :) _________________ nothing is ever truly lost , just misplaced and awaiting us
| |
| | | Sponsored content
| Subject: Re: Ivanhoe v68 (32 + 64 bit): yes another new one | |
| |
| | | | Ivanhoe v68 (32 + 64 bit): yes another new one | |
|
Similar topics | |
|
| Permissions in this forum: | You cannot reply to topics in this forum
| |
| |
| |