Jazz 640 released

[mar]

Almost missed this one! Thanks Evert!

[Jim Ablett]

Sure. I've applied the following patch:

Code: Select all

Index: src/computer/computer.c
===================================================================
--- src/computer/computer.c	(revision 648)
+++ src/computer/computer.c	(working copy)
@@ -687,6 +687,8 @@
       print_iter("\n");
 #endif
 
+      uci("info nps %lld\n", 1000000ll*moves_searched / (get_timer() - 1 - start_time));
+
       uci("info depth %d score cp %d nodes %d time %d hashfull %d ", depth+1,
          score[best_move], moves_searched, peek_timer(game), min(1000, game->transposition_table->write_count*1000/game->transposition_table->number_of_elements));
 

Thanks!

Thanks Evert,

Just one more thing I found.
'go infinite' isn't . It stops with a bestmove at depth 10.

Jim.

[Evert]

Just one more thing I found.
'go infinite' isn't . It stops with a bestmove at depth 10.

Hmm... this is not so obvious. I can confirm here that it exits early (at depth 13 ) but the code that launches the search sets the thinking time to infinite and the maximum depth to 60 (which is a hard limit due to the size of killer tables and things like that). I know it works correctly in the XBoard version; it's possible I've changed something in the root-node search and did update the XBoard interface but not UCI. I'll add it to my list of things to check out.

Thanks!

[geots]

Just one more thing I found.
'go infinite' isn't . It stops with a bestmove at depth 10.

Hmm... this is not so obvious. I can confirm here that it exits early (at depth 13 ) but the code that launches the search sets the thinking time to infinite and the maximum depth to 60 (which is a hard limit due to the size of killer tables and things like that). I know it works correctly in the XBoard version; it's possible I've changed something in the root-node search and did update the XBoard interface but not UCI. I'll add it to my list of things to check out.

Thanks!

Between your fixes and Jim's compiles, Jazz now shows the evaluations in the thinking lines and the kN/s. The only other thing I see is that you still cannot load it on the main board in Fritz and play it yourself. Makes only one move- as I explained earlier. But that does not concern me- because it seems to play fine on the other board against another engine.

I hate that your earlier version missed out on the Freeware RR. I owe you one there. But if it makes you feel any better- Capivara won Group 12 with ease, and that is the group Jazz would have played in based on elo at that time. Not many games run- just a few- but I have a gut feeling that it is highly likely that this Jazz 6.40 would now be too strong to play in Group 12.


Best and thanks for your work-

george


PS: Now I need to know something from you. This depth stoppage issue that you are going to look at- should I expect a fix that will be put in 6.40 and a re-compile, or do I go ahead and use it "as is" and expect that to be fixed in the next release?

[Evert]

PS: Now I need to know something from you. This depth stoppage issue that you are going to look at- should I expect a fix that will be put in 6.40 and a re-compile, or do I go ahead and use it "as is" and expect that to be fixed in the next release?

I don't know what the issue is. One thought that came to mind is that a special mode where the search aborts early if the score is high and the best move is kept for "long enough" (mainly intended for test suites) was somehow activated, but that isn't it.

If, as it seems, this affects infinite-time searches but not regular time-controls, it shouldn't affect playing strength in a tournament. So go ahead and use the current version, I'll ry to have a fix in the next version (which won't be 640).

[geots]

PS: Now I need to know something from you. This depth stoppage issue that you are going to look at- should I expect a fix that will be put in 6.40 and a re-compile, or do I go ahead and use it "as is" and expect that to be fixed in the next release?

I don't know what the issue is. One thought that came to mind is that a special mode where the search aborts early if the score is high and the best move is kept for "long enough" (mainly intended for test suites) was somehow activated, but that isn't it.

If, as it seems, this affects infinite-time searches but not regular time-controls, it shouldn't affect playing strength in a tournament. So go ahead and use the current version, I'll ry to have a fix in the next version (which won't be 640).

Thanks Evert,

george