WCCC June 27th in Leiden

[Harvey Williamson]

What hardware is it this time ?

Quad core i7 16gb ram

[lkaufman]

Perhaps Komodo should play with many more cores too when hardware is important.

I'm sure that if they had had access to a machine with more cores they would have done so.

Actually we couldn't even use the 60 cores we had available, using more than 48 was apparently worse. To use huge numbers of cores effectively is a challenge in itself, one that we could only tackle if we actually had regular access to them, not just for a tournament. We really can't expect to remain competitive with a University that has such hardware, but it hardly matters if we finish behind such hardware. How many have access to it?

[Henk]

Perhaps Komodo should play with many more cores too when hardware is important.

I'm sure that if they had had access to a machine with more cores they would have done so.

Actually we couldn't even use the 60 cores we had available, using more than 48 was apparently worse. To use huge numbers of cores effectively is a challenge in itself, one that we could only tackle if we actually had regular access to them, not just for a tournament. We really can't expect to remain competitive with a University that has such hardware, but it hardly matters if we finish behind such hardware. How many have access to it?

Cloud has many virtual machines. Perhaps only a matter of money.

[IanO]

What hardware is it this time ?

Quad core i7 16gb ram

Still 2.7 GHz like the last two events?

(I find it really strange that we are now at a computer performance plateau. Computer chess events used to be a barometer for the ever increasing speeds of the microprocessor industry. Now it seems all you can do is add more cores or make use of another low-level bitmap instruction.)

[mjlef]

What hardware is it this time ?

Quad core i7 16gb ram

Still 2.7 GHz like the last two events?

(I find it really strange that we are now at a computer performance plateau. Computer chess events used to be a barometer for the ever increasing speeds of the microprocessor industry. Now it seems all you can do is add more cores or make use of another low-level bitmap instruction.)

They appear to be the same machines as last year.

[MikeB]

What hardware is it this time ?

Quad core i7 16gb ram

Still 2.7 GHz like the last two events?

(I find it really strange that we are now at a computer performance plateau. Computer chess events used to be a barometer for the ever increasing speeds of the microprocessor industry. Now it seems all you can do is add more cores or make use of another low-level bitmap instruction.)

The demand is changing, hence the R&D budgets being reallocated. More and more R&D is being devoted to low voltage cpus for the tablets and phones. Technology is migrating to distributive processing via data farms and web technology and what not. The big universities are holding on to the big machines. My main machine is a 2010 Mac Pro, 12 Cores with 3.46Ghz X5690 - still a very competitive machine purchased used for about for a fraction of new Mac Pro that would only offer marginally better performance. It's more sad than strange - we're at a crossroads - almost the end of an era really.

[mjlef]

Perhaps Komodo should play with many more cores too when hardware is important.

I'm sure that if they had had access to a machine with more cores they would have done so.

I have gotten access in the past for short times to pretty big machines. Like a 120 core machine we could have used last year. But having a machine available and being able to use all of those cores effectively is not an easy problem. There is a diminishing return as cores increase. You can see this, for example, on the fastgm.de core doubling experiments. So I would need access for a longer time to program using it effectively (like Jonny). It is on my to do list, and it is possible we will continue to have access to the current 60 core machine in the near future.

Mark

[AdminX]

In the final game was this the critical position or was the game already bust form a strategical point of view?


Black to move 36. ...
Fen:5r1k/1pp3np/2nq1rpN/p7/3pRPPP/3P4/PPN1Q3/1K3R2 b - - 8 36

[pgn]
[Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "New game"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "5r1k/1pp3np/2nq1rpN/p7/3pRPPP/3P4/PPN1Q3/1K3R2 b - - 0 36"]
[PlyCount "6"]

36... b5 (36... Qd5 37. a3 (37. Rf3 Ne6 38. f5 g5 $11 {[%eval 0,40]}) (37. Qg2
Re6 38. a3 a4 $11 {[%eval 7,37]}) (37. Qd2 Qd6 38. Qe2 Qd5 $11 {[%eval 21,32]})
(37. Rf2 Qd7 (37... Nh5 38. gxh5 gxh5 $11 {[%eval 18,36]}) 38. Qf1 Re6 (38...
Nf5 $11) 39. Rxe6 $11 {[%eval 0,42]}) (37. Qh2 Re6 38. Rxe6 Nxe6 39. f5 $11 {
[%eval 18,38]}) 37... b5 38. h5 gxh5 39. g5 $14 {[%eval 41,38]}) (36... Nf5 37.
gxf5 (37. Qe1 Nxh6 38. g5 Rxf4 $11 {[%eval -17,40]}) (37. Nxf5 gxf5 38. gxf5
Rxf5 39. Qf2 $14 {[%eval 32,36]}) 37... gxf5 38. Nxf5 Rxf5 39. Qf2 $14 {
[%eval 34,35]}) (36... Qd7 37. Qd2 (37. Qe1 Ne6 38. Qd2 Qd6 39. f5 $11 {
[%eval 0,42]}) (37. Qd1 Qd6 38. Qe2 Qd5 $11) (37. h5 g5 38. Nf5 Nxf5 $11) 37...
Qd6 (37... Qd5 38. Qe2 b5 $14 {[%eval 38,29]}) 38. Na3 Re6 39. Nc4 $11 {
[%eval 11,37]}) (36... Re6 37. Ne1 (37. Rxe6 Qxe6 38. Qf3 Qd7 39. f5 $11 {
[%eval 0,40]}) (37. Qe1 Rxe4 38. Qxe4 Qe6 (38... Qd8 $11 {[%eval 0,42]}) 39.
Qg2 $11 {[%eval 0,42]}) 37... Nb4 38. a3 Nd5 39. f5 $11 {[%eval 6,39]}) 37. Qd1
(37. Rf2 b4 38. Ne1 Nd8 39. Ng2 $11 {[%eval 0,36]}) (37. g5 Re6 38. Ng4) 37...
b4 38. g5 Re6 39. Ne1 $11 {[%eval 4,40]} *

[/pgn]

[tpoppins]

In the final game was this the critical position or was the game already bust form a strategical point of view?
Black to move 36. ...
[d]5r1k/1pp3np/2nq1rpN/p7/3pRPPP/3P4/PPN1Q3/1K3R2 b - - 8 36

fully bust.

In addition to Ted's analysis (he forgot to specify the engine but going by what I can see on Let's Check, Komodo 10 is his current weapon of choice) that shows that Black has a wide choice of continuations that keep the game on the even keel there's also this:

Code: Select all

Analysis by Stockfish 020716 64 POPCNT:

36...Re6 37.Ne1 Nb4 38.a3 Nd5 39.f5 Rxe4 40.Qxe4 Ne3 41.Rg1 c6 42.Nf3 gxf5 43.Nxf5 Nexf5 44.gxf5 Nxf5 45.h5 a4 46.Ne5 Nh6 47.Nc4 Qf6 48.Qe2 Qf3 49.Qe5+ Qf6 50.Qc5 Rg8 51.Rc1 Nf5 52.Rf1 Rg5 53.h6 Qe7 54.Qxe7 Nxe7 55.Rf8+ Rg8 56.Rf7 Rg1+ 57.Ka2 Ng8 58.Rxb7 Nxh6 59.Rb4 Ng4 60.Rxa4 h5 61.Ra5 Rh1 62.Kb3 Kh7 63.Kb4 h4 64.Kc5 Kh6 65.b4 h3 
  =  (0.17)   Depth: 46/76   01:58:10  102182MN, tb=5840010

It's possible that Lyudmil is right and the engines are wrong; in such cases I usually root for the human - the silicone monstars have no concept of strategical plans or chess per se for that matter - but here methinks Lyudmil's "fully bust" eval is "a bit out there" as Prof. Hyatt would put it.

This also reminds me of Nunn's comment on Alekhine's annotations from the foreword to the Batsford algebraic edition of Alekhine's Best Games - that they sometimes suffer from the "I won so I must have been winning all the way" syndrome.

If there are some fine strategic nuances in that position that doom Black to defeat, a 1900-rated patzer like me would like to see them exposed, whenever Lyudmil finds the time.

[JJJ]

Let's Lyudmil Tsvetkov play this position against Stockfish or Komodo then