vanytchouck: Was 32...a6 forced?
This move looks like a simple blunder, giving the quality for nothing.39...Bxf4 looks like a far too optimistic move leaving Kramnik with a Rook for two pawns without any compensation.
Although Kramnik was obviously better at the 32th move, the position is not desperate.
p.s.
I made Fritz 10 analyse the position from the 31th black move
* At the 31th move (40 minute of analysis)
1. = ( - 0,14) : 31...a5 32. Rac1 Ne6 33. Kg2 Qb8 34. a3 Bd6 35. Rc3 Rdd8 36.d5 Nf8 37. Nxd6
2. 
(0,38): 31...Be7 32. Nxa7 Qa8 33. Nb5 Rc8 34. Qe2 Qa5 35. Nf5 Ng6 36. Nxe7 Nxe7 37. Rdc1
3. (0,58) 31...Bd6 32. Nxa7 Ne6 33. Qb5 Rde7 34. Rac1 Nxf4 35. gxf4 Bxf4 36. Re1
* At the 32th black move (1h10 of analysis)
1. = (- 0,04):32...Qb8 33. Nc3 a6 34. Kg2 b5 35. Qe2 Bd8 36. Ba3 Ba5 37. Qh5 Qd8 38. Bc5
2. (0,27) : 32...Ba6 33. Qa4 Bxb5 34. Qxb5 Ne6 35. Qh5 g6 36. Qb5 Nc7 37. Qc6 Nd5 38.Qe1 Nxc3
3. (0,55) : 32...Qc8 33. Qa4 Qb8 34. Nxa7 Qa8 35. Re1 Bd6 36. f5 Kg8 37. Kg2 Red8 38. Qc4 + Kh8
4. 
(0,82) :32...Bd6 33. Nxa7 Ng6 34. Nb5 Bb8 35. Nc3 Bd6 36. f5 Ne7 37. Qe2 g6 38. Rf1 Nc6
5. 
(1,61) : 32...Ne6 33. Qxe6 Ba3 34. Qxe8+ Qxe8 35. Bxa3 Qh5 36. Nd6
6. (1,70) : 32...a6 33.Nc7 b5 34. Qc2 Bd6 35. Nxe8 Qxe8 36. Qe2 Ne6 37. d5 Nc5 38. Nf5
This lines seems to back my impression about 32...a6.
But the line 39...Bxf4 is evaluated as the best one (after 50 minutes of analysis):
1. (2,50) :39...Bxf4 40. gxf4 Nxf4 41. Qf1 Qe5 42. Rc2 Bd7 43. Kh1 Nd3 44. b4
The best line but in an already lost position...
Kibitzing
sac: 47.Rxd3
