From: Marco Costalba Date: Sat, 8 Oct 2011 09:51:39 +0000 (+0100) Subject: Better document how MultiPV search works X-Git-Url: https://git.sesse.net/?p=stockfish;a=commitdiff_plain;h=fec623d68ddc3776257e7421109afdf61cf06ca4 Better document how MultiPV search works No functional change. Signed-off-by: Marco Costalba --- diff --git a/src/search.cpp b/src/search.cpp index 4bb701a2..b893fdce 100644 --- a/src/search.cpp +++ b/src/search.cpp @@ -159,7 +159,7 @@ namespace { RootMoveList Rml; // MultiPV mode - int MultiPV, UCIMultiPV, MultiPVIteration; + int MultiPV, UCIMultiPV, MultiPVIdx; // Time management variables bool StopOnPonderhit, FirstRootMove, StopRequest, QuitRequest, AspirationFailLow; @@ -527,18 +527,17 @@ namespace { // Iterative deepening loop until requested to stop or target depth reached while (!StopRequest && ++depth <= PLY_MAX && (!Limits.maxDepth || depth <= Limits.maxDepth)) { - // Save last iteration's scores, this needs to be done now, because in - // the following MultiPV loop Rml moves could be reordered. + // Save now last iteration's scores, before Rml moves are reordered for (size_t i = 0; i < Rml.size(); i++) Rml[i].prevScore = Rml[i].score; Rml.bestMoveChanges = 0; - // MultiPV iteration loop - for (MultiPVIteration = 0; MultiPVIteration < Min(MultiPV, (int)Rml.size()); MultiPVIteration++) + // MultiPV loop. We perform a full root search for each PV line + for (MultiPVIdx = 0; MultiPVIdx < Min(MultiPV, (int)Rml.size()); MultiPVIdx++) { // Calculate dynamic aspiration window based on previous iterations - if (depth >= 5 && abs(Rml[MultiPVIteration].prevScore) < VALUE_KNOWN_WIN) + if (depth >= 5 && abs(Rml[MultiPVIdx].prevScore) < VALUE_KNOWN_WIN) { int prevDelta1 = bestValues[depth - 1] - bestValues[depth - 2]; int prevDelta2 = bestValues[depth - 2] - bestValues[depth - 3]; @@ -546,8 +545,8 @@ namespace { aspirationDelta = Min(Max(abs(prevDelta1) + abs(prevDelta2) / 2, 16), 24); aspirationDelta = (aspirationDelta + 7) / 8 * 8; // Round to match grainSize - alpha = Max(Rml[MultiPVIteration].prevScore - aspirationDelta, -VALUE_INFINITE); - beta = Min(Rml[MultiPVIteration].prevScore + aspirationDelta, VALUE_INFINITE); + alpha = Max(Rml[MultiPVIdx].prevScore - aspirationDelta, -VALUE_INFINITE); + beta = Min(Rml[MultiPVIdx].prevScore + aspirationDelta, VALUE_INFINITE); } else { @@ -558,39 +557,44 @@ namespace { // Start with a small aspiration window and, in case of fail high/low, // research with bigger window until not failing high/low anymore. do { - // Search starting from ss+1 to allow referencing (ss-1). This is + // Search starts from ss+1 to allow referencing (ss-1). This is // needed by update_gains() and ss copy when splitting at Root. value = search(pos, ss+1, alpha, beta, depth * ONE_PLY); - // It is critical that sorting is done with a stable algorithm - // because all the values but the first are usually set to - // -VALUE_INFINITE and we want to keep the same order for all - // the moves but the new PV that goes to head. - sort(Rml.begin() + MultiPVIteration, Rml.end()); - - // In case we have found an exact score reorder the PV moves - // before leaving the fail high/low loop, otherwise leave the - // last PV move in its position so to be searched again. - if (value > alpha && value < beta) - sort(Rml.begin(), Rml.begin() + MultiPVIteration); + // Bring to front the best move. It is critical that sorting is + // done with a stable algorithm because all the values but the first + // and eventually the new best one are set to -VALUE_INFINITE and + // we want to keep the same order for all the moves but the new + // PV that goes to the front. Note that in case of MultiPV search + // the already searched PV lines are preserved. + sort(Rml.begin() + MultiPVIdx, Rml.end()); + + // In case we have found an exact score and we are going to leave + // the fail high/low loop then reorder the PV moves, otherwise + // leave the last PV move in its position so to be searched again. + // Of course this is needed only in MultiPV search. + if (MultiPVIdx && value > alpha && value < beta) + sort(Rml.begin(), Rml.begin() + MultiPVIdx); // Write PV back to transposition table in case the relevant entries // have been overwritten during the search. - for (int i = 0; i <= MultiPVIteration; i++) + for (int i = 0; i <= MultiPVIdx; i++) Rml[i].insert_pv_in_tt(pos); - // Value cannot be trusted. Break out immediately! + // If search has been stopped exit the aspiration window loop, + // note that sorting and writing PV back to TT is safe becuase + // Rml is still valid, although refers to the previous iteration. if (StopRequest) break; // Send full PV info to GUI if we are going to leave the loop or - // if we have a fail high/low and we are deep in the search. Note - // that UCI protol requires to send all the PV lines also if are - // still to be searched and so refer to the previous search's score. - if ((value > alpha && value < beta) || current_search_time() > 5000) + // if we have a fail high/low and we are deep in the search. UCI + // protocol requires to send all the PV lines also if are still + // to be searched and so refer to the previous search's score. + if ((value > alpha && value < beta) || current_search_time() > 2000) for (int i = 0; i < Min(UCIMultiPV, (int)Rml.size()); i++) { - bool updated = (i <= MultiPVIteration); + bool updated = (i <= MultiPVIdx); if (depth == 1 && !updated) continue; @@ -600,14 +604,14 @@ namespace { cout << "info" << depth_to_uci(d) - << (i == MultiPVIteration ? score_to_uci(s, alpha, beta) : score_to_uci(s)) + << (i == MultiPVIdx ? score_to_uci(s, alpha, beta) : score_to_uci(s)) << speed_to_uci(pos.nodes_searched()) << pv_to_uci(&Rml[i].pv[0], i + 1, pos.is_chess960()) << endl; } - // In case of failing high/low increase aspiration window and research, - // otherwise exit the fail high/low loop. + // In case of failing high/low increase aspiration window and + // research, otherwise exit the fail high/low loop. if (value >= beta) { beta = Min(beta + aspirationDelta, VALUE_INFINITE); @@ -633,14 +637,14 @@ namespace { bestValues[depth] = value; bestMoveChanges[depth] = Rml.bestMoveChanges; - // Do we need to pick now the best and the ponder moves ? + // Skills: Do we need to pick now the best and the ponder moves ? if (SkillLevelEnabled && depth == 1 + SkillLevel) do_skill_level(&skillBest, &skillPonder); if (LogFile.is_open()) LogFile << pretty_pv(pos, depth, value, current_search_time(), &Rml[0].pv[0]) << endl; - // Init easyMove after first iteration or drop if differs from the best move + // Init easyMove at first iteration or drop it if differs from the best move if (depth == 1 && (Rml.size() == 1 || Rml[0].score > Rml[1].score + EasyMoveMargin)) easyMove = bestMove; else if (bestMove != easyMove) @@ -649,9 +653,9 @@ namespace { // Check for some early stop condition if (!StopRequest && Limits.useTimeManagement()) { - // Stop search early if one move seems to be much better than the - // others or if there is only a single legal move. Also in the latter - // case we search up to some depth anyway to get a proper score. + // Easy move: Stop search early if one move seems to be much better + // than the others or if there is only a single legal move. Also in + // the latter case search to some depth anyway to get a proper score. if ( depth >= 7 && easyMove == bestMove && ( Rml.size() == 1 @@ -779,7 +783,7 @@ namespace { excludedMove = ss->excludedMove; posKey = excludedMove ? pos.get_exclusion_key() : pos.get_key(); tte = TT.probe(posKey); - ttMove = RootNode ? Rml[MultiPVIteration].pv[0] : tte ? tte->move() : MOVE_NONE; + ttMove = RootNode ? Rml[MultiPVIdx].pv[0] : tte ? tte->move() : MOVE_NONE; // At PV nodes we check for exact scores, while at non-PV nodes we check for // a fail high/low. Biggest advantage at probing at PV nodes is to have a @@ -988,10 +992,10 @@ split_point_start: // At split points actual search starts from here if (move == excludedMove) continue; - // At root obey the "searchmoves" option and skip moves not listed in Root Move List. - // Also in MultiPV mode we skip moves which already have got an exact score - // in previous MultiPV Iteration. Finally any illegal move is skipped here. - if (RootNode && !Rml.find(move, MultiPVIteration)) + // At root obey the "searchmoves" option and skip moves not listed in Root + // Move List, as a consequence any illegal move is also skipped. In MultiPV + // mode we also skip PV moves which have been already searched. + if (RootNode && !Rml.find(move, MultiPVIdx)) continue; // At PV and SpNode nodes we want all moves to be legal since the beginning @@ -1018,7 +1022,7 @@ split_point_start: // At split points actual search starts from here if (pos.thread() == 0 && current_search_time() > 2000) cout << "info" << depth_to_uci(depth) << " currmove " << move - << " currmovenumber " << moveCount + MultiPVIteration << endl; + << " currmovenumber " << moveCount + MultiPVIdx << endl; } // At Root and at first iteration do a PV search on all the moves to score root moves