Moderator: Ras
Neat! Thank you. I've posted my first game against version 4.0 on r.g.c.*. Oooph, it was hard work for me to play against your smallest engine. I got my K+R+B+1p against K+B+4p of micro-Max 4.0 in the ending.hgm wrote:These are the non-WinBoard versions of micro-Max, and the protocol they use is 'stateless' w.r.t. the side they play. You have to press an extra <enter> (i.e. input an empty line) to make them do a move, and you can do so in any situation.
This makes it possible to setup a position (by just entering moves, as in WB 'force' mode), and play white or black. You can even have the computer play itself by just pressing <enter> all the time.
I won just barely. On rgc* I have annotated the game very lightly (almost not at all) but I had to compute the ending sequence to a move, when I made the final exchange of bishops. My position was not easy (true, here and there I played a bit shaky-flaky, but then we are not talking about a superGM tournament). Actually, I let the program take the central pawn e4 in the early part of the game because I lost my concentration. I don't know if it were an error on my part, I got a recompense on the Queen side, but I was worried about the central pawn mass of black. Micro-Chess 4.0 didn't get it rolling.hgm wrote:In the game you can see the main weak spot of uMax 4.0: it very naively lets its opponent's passers advance to 7th rank, and only then discovers it is too late to top them without significant material loss.
Micro-Max 4.8 is clearly stronger than me, at least at this time. I've already lost 3 games. Two of them were not real encounters, but the third one was a good fight and I lost. I risked a bit by castling long, tempted by an immediate pressure due to my Rd1 opposing black Qd8, and I was ahead with the development at the time. I allowed black to have two bishops and a very open position. Micro-Max 4.8 attacked my king very well. It played a weird move too but it was a good oneIn uMax 4.8 I fixed this, and the program already starts to vehemently oppose any attempts to get a Pawn on the 6th rank. This caused a real breakthrough in playing strength against other engines.
I tried a desperation counter-attack but it didn't work. It was interesting though.Micro-Max 4.8 is slower the 4.0 but still very reasonable. Would setting 1e7 be still within a minute to 90 seconds per ply? Well, I might find out.Note that you can set the level of the stand-alone versions of uMax only by recompiling with a different bound to the number of nodes (N) it is allowed to search. The standard version says N<1e6, which, on modern computers, makes it think only a few seconds per move. (It does about 1 million nodes per second on my 2.4GHz Core 2 Duo.)
).It takes a little more than a few seconds for 4.8 on my Mac, but it is still quite fast. I tried to replace 1e6 by 1e7 but the result was too slow for me. So I recompiled with 2e6. Now uM4.0 played fast again; it moved 1... Nc6 to my 1.d4, just like standard uM4.8 does. Then it played differently from 4.8. I lost quickly. (I'll have to play this opening in a more solid fashionhgm wrote:Note that you can set the level of the stand-alone versions of uMax only by recompiling with a different bound to the number of nodes (N) it is allowed to search. The standard version says N<1e6, which, on modern computers, makes it think only a few seconds per move. (It does about 1 million nodes per second on my 2.4GHz Core 2 Duo.)
-- 1d4 Nc6 2.d5 Ne5 3.Bf4 with eventual 3...Ng6 Qd2, was my attempt at fast development, but it's not too good). No, QS only considers captures, and in 4.0 even only recaptures (= capturing the piece that moved on the previous ply).AndrewShort wrote:If I understand you correctly, your Quies() function considers pawn moves to the 6th rank (or higher), even if they are not captures nor checks, and even if the moving pawn is not passed. Or do you only do this for passed pawns?