8.Rb1!? In Game 9 Topalov employed another fresh setup  8.a3!? Nbd7 9.g3. The novelties pour out as from the horn of plenty. A little digression: back in 1981 in Merano Karpov against Korchnoi made a new move in the Open Variation of the Ruy Lopez. The Soviet press attached an endless amount of exclamation marks to this move and gave a lot of ecstatic comments. In another game Karpov introduced one more novelty in the same position and the reaction of the press was identical. I remember my perplexity  which one of the novelties was "correct"? (and I don't understand it still). OK let's not become like them. Further practice will determine the value of the Topalov team's ideas. 8...Nbd7 9.c5. Everything is clear  White prepares b4b5 allowing ...e5. At first sight the idea seemed to me extremely risky or even suspect from the strategic standpoint. Later White will find it very difficult to hold the critical d4pawn. Nevertheless the double pawns on the gfile might hamper the mobility of Black's pieces. 9...a5!? Securing the opening of the afile. 10.a3. In the event of 10.f4?! Black destroys White's centre from the other side  10...b6! and after the exchange on b6 he plays ...c6c5. 10...e5 11.b4 axb4 12.axb4. 12...Qc7. Attacking the h2 pawn. In principle the advance of the bpawn to b5 doesn't look like Black's fundamental achievement. But here 12...b5 is simply dubious in view of 13.dxe5! (13.Nxb5 or 13.cxb6 are not that clear) 13...Nxe5 14.Nxb5! (Notkin) 14...cxb5 15.Bxb5+ Ke7 16.f4!?, etc. After 12...Be7 the idea 13.e4!? (Deviatkin) looks risky as White hasn't completed development. Even 13...Rh4!? comes into consideration. I believe Topalov would have played something more solid. 13.f4! Weakening the e4 square in addition. But perhaps everything is right  White should play for domination. 13...exf4. 13...Nh5!? (Pein) may lead to a tense position after 14.fxe5!? Ng3 15.Rg1 (15.hxg3!? Rxh1 16.e4 Notkin) 15...Rxh2 and here for example 16.Qf3 Be7, though I don't have too much trust in Black's prospects. 14.exf4. 14...Be7. 14...Nh5!? is critical again and here: a) 15.Qg4 Be7! (15...Ndf6 16.Qf3 Qe7+ 17.Be2 Ne4 18.00 f5 19.b5 Nhf6 20.bxc6 bxc6 21.Rb6+/ Shipov) 16.Qh3 Rh6! 17.f5 g5 looks quite viable for Black; b) probably stronger is 15.Qf3 f5 (15...Be7 16.g4!?) 16.g4!? Nhf6 17.g5 Ne4 18.Nxe4  and the white bishop gets a job whatever pawn would recapture on e4. 15.Be2. 15.Bd3!? was an alternative. Here is the quote from GM Vladimir Belov's comments on the official site: "After the game the Bulgarian grandmaster timidly suggested that the d4 might turn weak: 15.Bd3 Nf8 16.Qf3 Ne6, but after 17.Ne2 White's position should be preferred". 15...Nf8!? Black immediately begins to fight against the d4 pawn. 15...Nh5 is hardly sufficient because of 16.00 f5 17.g4!? (Shipov). The same concerns 15...Ne4?! (one move earlier this would be even less favourable) 16.Nxe4 dxe4 17.d5! Bh4+ 18.Kf1 cxd5 19.Qxd5 00 20.g3+/ (Yemelin) or 15...Ng8?! 16.00 f5 17.b5!? Ngf6 18.Qc2 (Deviatkin), and the e4 square by itself might not be enough for Black to be happy. More likely the alternatives are 15...b5 and 15...00. There is also Notkin's idea 15...Kf8 but Maxim himself was first to disapprove of it. 16.00 Ne6 17.g3! It seems that Black is fine after 17.Qd3 g5! (Marin) or 17.b5 g5! (after the eccentric 17...Ng5?! White can take the knight: 18.fxg5!? with idea 18...Qxh2+ 19.Kf2 Qh4+ 20.g3 Qh2+ 21.Kf3 Qh5+ 22.Kf4+). 17...Qd7!? The h2 pawn is already out of reach therefore the queen goes to the adjacent diagonal. After 17...Rd8, 18.Be3! suggests itself (18.b5 Ne4 19.Nxe4?! dxe4, and White can't hold the d4 square). In case of 17...00 White could start the preparation f4f5, or to make this move right away. 18.Qd3!? In the event of 18.Bf3 Ne4 19.Nxe4 dxe4 20.Bxe4 interesting is 20...Bf6!? (similar to the game, White is a bit better after 20...Qxd4+ 21.Qxd4 Nxd4 22.Bb2) 21.d5 Bd4+ 22.Kh1 Ra2 (Belov). The point is that 23.Bg2? isn't good due to 23...Rxg2! 24.Kxg2 Qxd5+ 25.Rf3 (or 25.Qf3 Qa2+!) 25...Qf5! 26.Be3 Qh3+ 27.Kg1 Qxh2+ with an attack. 18.b5? is refuted by 18...Nxc5! with idea 19.dxc5? Bxc5++ 20.Kg2 Qh3+ 21.Kf3 Qh5+! 22.Kg2 Qxh2+ 23.Kf3 Nh5! 24.Qe1 Nxg3. To 18.Be3 Black may reply 18...Ne4 19.Nxe4 dxe4, when after 20.Bc4 he has the key move 20...Nc7! transferring the knight to d5. The last lines reveal the ideas behind 17...Qd7. 18...Ne4!? White has prepared to the exchange in the centre therefore it made sense to think about alternatives. For instance 18...Nc7 having in mind 19.b5 Qh3 20.Rf2 Ng4 21.Bxg4 Qxg4. 19.Nxe4 dxe4 20.Qxe4 Qxd4+. Here 20...Bf6 21.d5! (Shipov) promises White better chances thanks to the bishop pair in a "loose" pawn structure. 21.Qxd4 Nxd4 22.Bc4 00. White's advantage is small but doubtless. The position isn't closed thus the bishops should mean something! True had the potential weakness on b7 not existed they probably wouldn't. 23.Kg2. After 23.Kf2?! Black obtains counterplay by 23...Nc2! (угрожая 24...Nxb4!)  Notkin. 23...Ra4. It's always tempting to activate pieces and to create a threat (on this occasion ...Bxc5). But in Belov's opinion 23...Rfd8 was more precise. Black had other possibilities as well. 24.Rd1! Rd8! It transpires that 24...Bxc5?! is refuted by 25.Be3!+/ Ne6 (as concerns 25...Rd8 see 24...Rd8 25.Be3 Bxc5?!) 26.Bb3! Rxb4 27.Bxc5 Nxc5 28.Bxf7+! winning the exchange (Zargebelny). To 24...Bf6 Deviatkin suggests the unpleasant 25.g4. 25.Be3. 25.Rd3!? was also possible not scared by 25...Bxc5 26.Bxf7+ Kxf7 27.bxc5 Ra2+, to which 28.Kh3! follows and surely the king won't be checkmated on g4. 25...Bf6! Here 25...Bxc5? is bad too: 26.Bxf7+ Kxf7 27.bxc5 Ra2+ 28.Kh3 (or 28.Kf1). Dubious is 25...Nc2?! 26.Rxd8+ Bxd8 27.Bf2 (27.Bd2!? with idea 27...Na3 28.Ra1 Bf6 29.Ra2!? Nxc4 30.Rxa4 Nxd2 31.Ra8+ Kh7 32.Rb8 Notkin) 27...Nxb4 (27...Rxb4? 28.Bb3!!+ and Black sustains material losses; 27...Na3 28.Bb3 Nxb1 29.Bxa4+/). Here Black holds somehow or at least he doesn't lose by force: 28.Rd1 (28.Bb3!? Ra7 29.Bxf7+ Kxf7 30.Rxb4 g5! Yemelin) 28...Nd5!? (or 28...Bf6 29.Rd7 Nd5 30.Bxd5 cxd5 31.Rxb7!? Rc4 32.Rd7 d4 33.g4 Kf8!) 29.Bxd5 cxd5 30.Rxd5 Bf6 31.Rd7 Rb4, intending ...Kg8f8e8. However for Kramnik there was no point to take such risk. 25...Nf5 26.Rxd8+ Bxd8 27.Bf2 Bf6 looks unpleasant for Black  Yemelin. I tend to agree  Black shouldn't be in a hurry with this exchange. 26.g4. Topalov has already touched so many pawns in this game but only now these advances begin to raise doubt. White could avoid 26.g4 Ra3!? with the help of 26.Bf2 Kf8 (26...Ra3 27.Rb2! Rc3 28.Bf1 Shipov) and here 27.g4 (again, not essential) transposes into the game. Other possibilities are 26.h4!? (Marin), 26.Rd2 or 26.Rd3 with some advantage in all cases. 26...Kf8. Consistently played. However there was an alternative  26...Ra3!? 27.Rd3 (27.Bxd4 Bxd4!, and White hardly has something special; 27.Bf2 Rc3! with counterplay, Shipov) 27...Nc2!? (the ending in the line 27...Rxd3 28.Bxd3 Nf5!? 29.Bxf5 gxf5 is preferable for White) 28.Rxd8+ Bxd8 29.Bd2  it looks like a roughly equal position. 26...Nc2?! 27.Rxd8+ Bxd8 28.Bd2 Ra3 29.Rb2 (or 29.Rc1 Ne3+ 30.Bxe3 Rxe3 31.Rd1) 29...Ne3+ 30.Bxe3 Rxe3 31.Rd2 leads to the ending with opposite squared bishops which is difficult for Black. 27.Bf2. After 27.g5 Nf5! 28.Rxd8+ Bxd8 the black knight gets a strong post. 27...Ne6. Inferior is 27...Nb5?! 28.Rxd8+ Bxd8 29.Rd1 Be7 30.Be1! (30.Rd7 Rxb4 31.Rxb7 Rxc4 32.Rb8+ Bd8 33.Rxd8+ Ke7, and White loses the f4 pawn  MG) 30...Ke8 31.g5!? (Shipov) with the white pressure. 28.Rxd8+ Bxd8. 29.f5. Perhaps this is another hasty move. After 29.Bxe6 fxe6 30.h4!? Bc7 31.Kf3 White scarcely has a tangible edge. It was worth trying to benefit from the delay in exchanging the bishops  29.Kf3, and if 29...Bf6 (Monokroussos), then 30.g5! Bc3 31.Bxe6 fxe6 32.b5 with idea 32...Bd2 33.bxc6 bxc6 34.Rb6. After 29.Be3, the continuation 29...Bf6!? is seemingly more effective (less good is 29...Bc7?! 30.Kf3 Ra3 31.Rb3 Rxb3 32.Bxb3, and 32...Bxf4? fails to 33.Bxf4 Nd4+ 34.Ke4 Nxb3 35.Kd3, Pein). 29...gxf5 30.gxf5 Nf4+. 30...Ng5!? was possible as well (Pein). 31.Kf3 Nh5! 31...Nd5?! 32.Bxd5 cxd5 33.b5!? is rather dangerous for Black. 32.Rb3?! 32.Rd1! Be7! 33.Be1! was worth trying with idea 33...Bxc5 34.Rd7. 32...Bc7! Black is reluctant to cede the h2b8 diagonal (32...Nf6 33.Bg3!?). Kramnik's position looks unassailable now. White has nothing to clutch at. With the pawn on f5 the white bishop has no access to c8 even in theory. 33.h4. Forward! But where? 33.h3!? is more solid. 33...Nf6=. It seems that White's advantage has disappeared completely. 34.Bd3. Or 34.Bd4 Nd7 35.Ke4 Bd8!? (Shipov). 34...Nd7!? Also possible was 34...Nd5 followed by for example 35.b5 Rf4+ 36.Ke2 Ra4= (Deviatkin). 35.Be4?! Kramnik: "After the extremely poor 35.Be4 Black plays for a win without any risk and it's inconvenient for Topalov to offer a draw although to my mind the position never really ceased to be drawn". White could maintain the balance by 35.Bb1!? (Marin) or 35.b5!?. 35...Ne5+ 36.Kg2. 36.Ke3!? Ng4+ 37.Kf3, and White has nothing to be afraid of. 36...Ra2. It's already Black who has the initiative but Topalov starts defending well. 37.Bb1!? An alternative could be 37.Kf1 if there was one at all. 37...Rd2 38.Kf1! To 38.Ra3?! there is 38...Ng4!? (38...Rb2 39.f6! gxf6 40.Bf5 Rxb4 41.h5 Shipov) 39.Ra2 Ne3+ 40.Kf3 Rxa2 41.Bxa2 Nxf5 and Black has a sound (it's important!) extra pawn. 38...Ng4! 39.Bg1. It's probably stronger than 39.Be1?! Nh2+!? 40.Kg1 Re2, and 41.Bg3 (41.Bf2 Ng4 42.Bd4 Be5!) 41...Bxg3 42.Rxg3 Re1+ 43.Kxh2 Rxb1 44.f6 (44.Rg4 Rf1! Deviatkin) cannot solve all White's problems due to 44...Rxb4! 45.fxg7+ Kg8, for example 46.h5? (46.Kh3 Rc4!? with a basic idea 47.Rg5 f6 48.Rg6 Rxc5 49.Kg4 b5 50.h5? Rg5+!+) leads to a lost pawn ending after 46...Rh4+ 47.Rh3 Rxh3+ 48.Kxh3 Kxg7 49.Kh4 Kh6 50.Kg4 f6 51.Kf5 Kxh5 52.Kxf6 Kg4 53.Ke5 Kf3 54.Kd6 Ke4 55.Kc7 Kd5+. 39...Bh2 40.Ke1! Stronger than 40.Bxh2?! Rxh2 41.b5 Rxh4 and even more so 40.Rd3? Rb2. 40...Rd5. Perhaps 40...Rg2!? is more acute and dangerous  41.Bd4 Be5 42.Be4!? (after 42.Bxe5!? Nxe5 43.f6, Pein, 43...Rg4! is rather strong when an easy draw for White is nowhere to be seen) 42...Rh2 43.Bg1 Rxh4 44.b5 (Shipov, Yemelin) and White shouldn't lose anyway. 41.Bf2. Many experts opined that 41.Bxh2!? Nxh2 42.Rd3! Rxd3 (42...Rxf5?! 43.Rd8+ Ke7 44.Bxf5 Kxd8 45.Kf2!; 42...Re5+?! 43.Kf2) 43.Bxd3 Ng4 (43...Nf3+ 44.Kf2 Nxh4 45.Kg3 g5 46.Kg4 f6 47.Be4= and the fortress is built up) 44.Ke2 led to equality. Though it's not obvious from afar White may avoid the punishment for his overadvanced pawns. 41...Ke7. 42.h5! If 42.Rd3?! then 42...Re5+! 43.Kf1 Nxf2 44.Kxf2 Rxf5+, and it's not clear whether White could escape after 45.Kg2 Bc7!? (or 45...Be5!? 46.Re3 Rh5) 46.Rd7+ Kxd7 47.Bxf5+ Kd8 and the bishop goes for the b4 pawn. 42...Nxf2. Or else Topalov might have transferred the bishop to h4 with the check. Now Black retains the initiative in the opposite squared bishop ending. 43.Kxf2 Kf6 44.Kf3 Rd4 45.b5!? White could try also 45.Be4 Rc4 (45...Ke5 46.Re3!) 46.Bd3. Topalov's method is more reasonable. Exchanging the queenside pawns he transposes into a more simple and objectively drawn position. 45...Rc4 46.bxc6 bxc6 47.Rb6. 47.Be4 Rxc5 48.Rb6 is the same. 47...Rxc5 48.Be4 Kg5. The ending after 48...Bc7 49.Rxc6+ Rxc6 50.Bxc6 Kxf5 51.Be8= is dead drawn. Or 48...Rc3+ 49.Kg4!? Ke5 50.Bxc6 Rg3+ 51.Kh4 Kxf5 52.Rb2 (the simplest) 52...Rg4+ 53.Kh3 Be5 54.Bd7+ Kg5 55.Bxg4 Bxb2 56.Bd1, followed by Kg2 with a draw. 49.Rxc6 Ra5 50.Rb6. Rejecting 50.f6 gxf6 51.h6 Kxh6 (51...f5 52.Bd3 Be5 53.h7 Ra8 54.Rc5=) 52.Rxf6+ Kg7 53.Rf5 Be5, followed by ...f6, Kf8e7d6. This is a draw as well but Black can still torture his opponent. 50...Ra3+ 51.Kg2 Bc7 52.Rb7. 52.f6 would lead to the situations similar to 50.f6, after 52...gxf6 53.Rb7 (or 53.Rb5+ Be5 54.Bd5 Kxh5) 53...Rc3 54.Bd5 Kxh5 55.Bxf7+. 52...Rc3 53.Kf2 Kxh5. Black's extra pawn defies realization. 54.Bd5!? f6 55.Ke2. 55.Bf7+ Kg4 56.Bg6 looks risky. Therefore White leaves the bishop close to his king albeit its position is somewhat shaky. 55...Kg4 56.Be4 Kf4 57.Bd3 Rc5 58.Rb4+ Kg3 59.Rc4 Re5+ 60.Re4. 60...Ra5! Any version of the rook exchange makes no sense neither does 60...Rxf5?! 61.Re3+ Kg4 62.Bxf5+ Kxf5 on account of 63.Re7! when White wins the g7 pawn. And by the way  what if he wouldn't? 61.Re3+. After 61.Re7 Ra2+ White is forced to play 62.Kd1 (62.Ke3? Bb6+ 63.Ke4 Ra4+!+ etc.) 62...Bf4! and if 63.Re2!? (not 63.Rxg7+? Kf2! 64.Rg4 Kf3! 65.Be2+ Ke3 66.Bb5 Rd2+! 67.Kc1 Rd5 68.Be2 Bg5 69.Bd1 Kf2+ 70.Kc2 Rd6+ and White is in Zugzwang) 63...Ra1+ 64.Kc2 Rc1+ when the white king's remoteness begins to cause alarm. 61...Kg2?! 61...Kg4! would have posed more problems. The draw isn't that "dead". Nevertheless with a precise defence White achieves this result. 62.Re7 (62.Re4+ Bf4!) 62...Ra2+ 63.Kd1 Bf4, and here an unnecessary capture 64.Rxg7+!? Kf3 transposes into a position examined by Shipov where Black cannot win: 65.Bb5 Rd2+ 66.Ke1 Rc2 67.Rd7. 62.Be4+ Kh2 63.Rb3 Ra2+ 64.Kd3 Bf4. The black king has lost its way and it's late to come back: 64...Kg3!? 65.Rb7! Kf4 66.Rxc7 Ra3+ 67.Ke2 Kxe4 68.Rxg7 Ra2+ 69.Ke1 Kxf5 70.Rg3 with a draw according to Philidor. 65.Kc4 Re2 66.Kd5=. Now White isn't worse. Draw agreed. Slav Defence D12

8.g3. Deviating from the new paths discovered by Topalov  8.a3!? Nbd7 9.g3 and 8.Rb1!?.
8...Nbd7. In my notes to Game 9 I guessed that by playing 8.a3 Topalov was trying to avoid 8...Bb4, employed in 2005 in Dos Hermanas by Rustemov against one of the future members of the infamous Appeal Committee. And we will soon see that playing Black Topalov wants his bishop on b4!
9.Bd2. 9.Bg2 offers a pawn sacrifice that Black may well accept: 9...dxc4!? 10.Qe2 Nb6 11.00 Bb4 12.a3 Ba5N 13.Rd1 Qe7 14.e4 e5 15.Be3 draw agreed, Ponomariov  Shirov, Foros 2006. In the final position White is by no means better.
9...Bb4!?N. 9...Be7 has been tested in a dozen of games. Each of our protagonists made his mark there: 10.b3 (10.Rc1 Topalov  Vallejo, Monaco (rapid) 2005; 10.Qe2 Karpov  Gelfand, Russia vs Rest of the World 2002; 10.Qc2 Alterman  Avrukh, Israel 2002) 10...00 11.Bg2 dxc4 (11...Ne8 12.00 f5 13.f4 g5? 14.cxd5 exd5 15.Nxd5+ Malakhov  Nikolic, KhantyMansyisk 2005) 12.bxc4 e5 13.Qb3 Rb8 14.Rd1 Qc7 15.00 Rfd8 Kramnik  Gelfand, Monaco (blindfold) 2005, and in this position the continuation a la Topalov 16.f4!? promises White slightly better chances.
10.Qb3 Bxc3 11.Bxc3 Ne4!
12.Bg2. Of course not 12.Bb4?? (a praiseworthy yearning to keep the bishop pair isn't welltimed) 12...Qf6 13.f4 (13.Qc2 Qf3+) 13...Nxg3+. Or 12.Qxb7? (inappropriate avidity) 12...Rb8 13.Qxa7 (13.Qxc6 Rxb2!! 14.Bxb2 Qa5+ 15.Ke2 Qd2+ 16.Kf3 Qxf2+ mating on the next move) 13...Qf6! 14.f4 Nxg3.
12...Nxc3 13.Qxc3. I don't think Black is inferior after 13.bxc3 Rb8!? with ideas ...b5 or ...Nb6. To 14.cxd5 cxd5 15.c4 he has 15...Qa5+.
13...f5!?
This interesting move looks somewhat wild, but I wouldn't dare weighing its strategic risk against the psychological benefits it brings. I would only say that after 13...Nf6 White would enjoy a small advantage in a very quiet situation. Clearly this is something Topalov wanted to avoid in the critical game.
14.00. Each of the opponents is true to his style. If Topalov was White we would probably see something like 14.000 if not 14.h4.
14...Qe7!? Again, a less obvious decision than 14...Nf6.
15.cxd5! This seems to be right. Otherwise it's not clear whether White should advance the bpawn.
15...exd5!? Certainly Topalov undertook all this not for the sake of defending the symmetrical position after 15...cxd5 where White has an evident initiative: 16.Rfc1!? (or 16.Qc7 Nf6 17.Rfc1 when Black should probably play 17...Rd8 with idea ...Kf7) 16...Nf6 17.Qa5.
16.b4! As a rule in such positions "the minority attack" is the only plan worthy of consideration. The appearance of a weakness on Black's queenside is unavoidable here.
16...Nf6.
17.Rfc1. The straight 17.b5 cxb5 was a significant alternative followed by for example 18.Qb3!? Qd7 19.Rfb1!? (19.Rab1 a6 20.a4 Ne4) 19...Rc8 (19...a6 20.a4! and the creation of a passed pawn by 20...b6? 21.axb5 a5 only doubles Black's trouble as the weakening of the c6 square is of a greater importance and after 22.Rc1 00 23.Rc6+/ the passer may become unprotected very soon) 20.a4!? (20.Qxb5 Rc1+ 21.Rxc1 Qxb5 22.Rc8+ Kd7!? 23.Rxh8 Qb2! looks messy) 20...Ne4 21.Qxb5 (there's no point in 21.Bxe4 fxe4 since White all the same cannot play 22.axb5?? due to 22...Qh3; the immediate 21.axb5 Nd2 22.Qxd5 Nxb1 23.Rxb1 is doubleedged) 21...Rc6 22.Rb2!? and White is slightly better (but not 22.Qxb7? Rc1+ 23.Bf1 Qxb7 24.Rxb7 Rxa1 25.Rb8+ Ke7 26.Rxh8 Nd2).
17...Ne4 18.Qb2 00. With the uncastled king Black wouldn't have time to exploit the hfile for an attack. Nevertheless in the given pawn structure it's favourable for him to have the double pawns on the king's flank. For instance he may push the g6 pawn towards g4 because after its exchange (by means of the f2f3 undermining) he will not get a weakness on f5. The advance of the b4pawn could be detained for one move  most likely not by 18...a6?! 19.a4 Nd6?! 20.b5 Nc4? 21.Rxc4! dxc4 22.bxc6+ or 18...Nd6?! 19.b5!? Nxb5 20.Bxd5 (here White is better), but with the help of 18...Qd7. Nevertheless after 19.a4 the benefit of inserting ...Qd7, a4 aren't clear.
19.b5 Rac8. Maybe Black could manage without this particularly defensive move. 19...Rf6!? deserved attention e.g. 20.Rab1 g5 (20...Re8 21.bxc6 bxc6 22.Qb7 Qe6 23.Qc7!, Shipov and 23...Nxf2? fails to 24.Rb7! Nh3+ 25.Kf1!+) 21.bxc6 bxc6 22.Qb7 Qe8 (Belov), where Black has counterchances for example 23.Qc7 f4 24.Rb7 Qg6! 25.Rxa7 (or 25.Bxe4 dxe4 26.Rxa7 Raf8) 25...Rxa7 26.Qxa7 Nxf2. On the contrary, 19...cxb5 20.Qxb5 gives White evident pressure "for free".
20.bxc6. It was critical to keep the tension by 20.a4!? displaying willingness to go for complications after 20...g5 (20...c5?! 21.dxc5 Rxc5 22.Rxc5 Qxc5 23.Rc1 leads to White's clear edge; 20...Qd7!?) 21.a5 f4 22.a6!
20...bxc6.
Position has stabilized. White is a bit better, but he cannot cause Black real troubles on the queenside without allowing him counterplay on the opposite flank.
21.Qe2. Judging by White's next move he should have preferred 21.Rab1, assigning his queen to a new place only after 21...g5 was played. Then Black would have to reckon with both 22.Qb7 and 22.Qb4!?
21...g5! 22.Rab1. The answer to the previous question probably lies in the line pointed out by Kramnik at the pressconference  22.Rc2 (preparing Rac1) 22...f4! with idea 23.exf4 gxf4 24.f3? Qf6 25.fxe4? Qxd4+ which he was little late in noticing.
It's too early for weakening the kingside 22.Qa6?! while 22.Qh5?! doesn't pose Black problems but only helps him to gain tempi by 22...Rf6!? with the idea of Rh6. To grab the pawn by 23.Bxe4 fxe4 24.Qxg5 is dangerous for White on account of 24…Qf7 followed by ...Rf8.
22...Qd7. Here 22...f4? is already bad because of 23.exf4 with idea 23...gxf4 24.f3!+.
23.Rc2 Rf6 24.Rbc1!? 24.Qa6 f4!? 25.Rb7 Qf5! is risky for White whereas after 24.a4 (intending a4a5a6) Black may play 24...Rcf8!? 25.f3 (25.a5 f4!? 26.exf4 gxf4 27.Bxe4 dxe4 28.Qxe4 f3 29.Qd3 Rh6 30.Qf1 Qxd4=) 25...Nd6. Perhaps this is what Kramnik is trying to prevent adding pressure upon the c6 pawn.
24...g4!? With the rook still placed at c8 24...f4?! 25.Bxe4 dxe4 cannot give equality while 24...Rcf8? is bad due to 25.f3! Nd6 26.Rxc6. Topalov could try to keep the idea ...f5f4 in reserve by 24...Rh6 but perhaps he was afraid of 25.f3!?. 24...Nd6?! looks suspect as after 25.Qa6 Black is forced to play 25...Nc4 (25...Ne4? 26.f3).
25.Rb2. In Marin's opinion 25.Rb1 is more precise preserving one of the rooks on the second rank.
25...Rh6 26.Qa6 Rc7! Feeble is 26...c5?!+/ on account of any of the following moves 27.Qb5, 27.Qb7 or 27.Rb7. Instead 26...Nd6 was possible, but then Black would only dream of attacking the kingside.
27.Rb8+ Kh7 28.Qa3! Rb7.
Here Shipov mentioned 28...Rf6 29.Rcb1 c5!? 30.dxc5 Nxc5, which is risky for Black  instead of the weakness on c6 he now has a more vulnerable target on d5. Then Black's role would be reduced to mere defence after 28...Qf7 29.Qf8! Qxf8 30.Rxf8.
29.Qf8?! One way or another White hardly had a tangible advantage. But with this move he allows an attack along the hfile sufficient to level the sides' chances. He lets Black enter his territory! White should have chosen between 29.Rxb7 Qxb7, or 29.Ra8!? Rf6 or 29.Rf8!?
29...Rxb8 30.Qxb8 Qf7!=.
The following course of the game as well as the analysis show total equality.
31.Qc8!? The fastest way to a draw (assuming that both sides are eager to reach it) was discovered by Deviatkin  31.Bxe4!? fxe4 32.Qe5 Rxh2 33.Kxh2 Qxf2+=.
31...Qh5. 31...Nd2? is completely wrong  32.Rxc6 Rxh2? 33.Kxh2 Nf3+ 34.Bxf3 Qh5+ 35.Kg1 gxf3 36.Qa6 (36.Qe6+) 36...Qh3 37.Qf1+.
32.Kf1. Also sufficient for equality is 32.Bxe4 fxe4 33.Kf1, with the idea of 33...Qxh2 34.Qf5+= Rg6 35.Ke2! Qg2 36.Qh5+.
32...Nd2+. Better than 32...Qg6?! 33.Bxe4! fxe4 34.Rb1 and White surely doesn't run any risk.
33.Ke1! After 33.Ke2?! Nc4! the exchange sacrifice is forced  34.Rxc4 (34.Rh1?! Qg6! 35.h4? f4+) 34...dxc4 35.Bxc6 where the draw pointed by Belov  35...Qf7 36.Bb5 (36.Bd7 Rb6! 37.Bxf5+ g6 MG) 36...Qd5 37.Bxc4 Qf3+ 38.Ke1 Rxh2 39.Qg8+ Kg6 40.Qe8+ looks like an achievement for White.
33...Nc4. Less accurate is 33...Nf3+ 34.Ke2 Qg6 35.h3!?
34.Bf1! Rf6! 35.Bxc4!? 35.Rb1 is more complex but not stronger  35…Nd6 (35...Nb6 comes into consideration as well) 36.Qa8 (36.Qxc6 Qxh2!? 37.Qxd5 Ne4 38.Rb2 Nxg3 39.Qg2! Nxf1 40.Kxf1 Qc7) 36...Qxh2 37.Rb8 Qg1 38.Rh8+ Kg6 39.Qg8 Rf7! 40.Qh7+ Kf6 41.Qh4+ Kg6. There's no point in 35.Rxc4?! dxc4 36.Bxc4 Qxh2 37.Qg8+ Kg6 38.Qe8+ Kg5! or in 35.Bd3 Qxh2 and bad is 36.Bxf5+? g6 as the f2 pawn is hanging.
35...dxc4 36.Rxc4. Or 36.Qa6!? Qxh2 (36...c5!? 37.Qxc4 cxd4 38.Qxd4 Qxh2 39.Rc7 Qh1+ 40.Kd2 Rg6 41.Rxa7 Qf3=) 37.Qxc4=.
36...Qxh2. 36...f4? is hardly good because of 37.Rc5! Then White can decide which pawn should take on f4  37...Qg6 38.gxf4 or 37...Qxh2 38.exf4 Qh1+ 39.Kd2.
37.Ke2 Qh1. Here after 37...f4?! White easily forces a draw if he wants to: 38.exf4 (38.Rxc6?! f3+ 39.Kd3 Qxf2 40.Rxf6 Qf1+; 38.gxf4? g3; 38.Qxg4!?) 38...Qh1 39.Qxg4 Qb1 40.Rc5 Qe4+ 41.Kf1. Therefore this makes little sense for Black to stake his all.
38.Rc5! Qb1!? 38...Qf3+= would have led to immediate draw. Black doesn't risk anything by continuing to play, but objectively he can't win here.
39.Qa6!? Qb2+. Soundly refraining from 39...Rh6?! 40.Qd3! Qxa2+ 41.Qc2 Qxc2+ 42.Rxc2 when Black should fight for a draw despite being a pawn up.
40.Kf1 Qb1+ 41.Ke2 Qb2+ 42.Kf1 Rh6 43.Qd3! g6. Or 43...Qa1+ 44.Ke2 Qxa2+ 45.Kf1 g6 46.Qc4!=.
44.Qb3 Rh1+. Topalov deprives his opponent of a pleasure to check endlessly  44...Qa1+ 45.Ke2 Rh1 46.Qf7+ etc. leaving it for himself.
45.Kg2 Rh2+ 46.Kxh2 Qxf2+ 47.Kh1 Qf1+. Draw agreed.