Are current PC chess programs powerful enough to beat the Deep Blue program that beat Kasparov in 97

[Learjeff]

@dan4112

This means (with the little data we have) that Deep Blue wasn’t significantly better than the top grandmaster.

Not only that, but it wasn’t even close to a fair match.

Deep Blue studied Kasparov’s recorded matches.
Kasparov asked to see DB matches but his request was denied.

Result: an unfair match, which showed that computers are at least on a similar level to the GM, but didn’t show who was really better.

IMHO, Kasparov should have said “No, thanks” when his request was denied.

[Chronos]

The rematch, which the computer won, was also on a pretty grueling schedule, designed to tire a human out (an effect the computer, of course, need not worry about).

[yurikvelo]

@Darryl_Lict

I’d hazard a guess that a powerful desktop computer would outclass 1997’s Deep Blue.

Pocket Phone (ARM) or Pentium-90 from 1993 will easily beat Deep Blue-II with modern chess software like Stockfish/Droidfish or Komodo (actually there is no modern software ported to P-1, but if clockdown modern CPU to MIPS level of P-1 it will be enough to beat DB-II.

DroidFish with 20 kN/sec is better than DeepBlue with 200 MN/sec

[yurikvelo]

@Isilder

Pretty much they are saying they don’t bother making Deep Blue software since its pointless having a software that is impossible to beat.

it is easily beaten with pocket device.
Reason for Hydra project (40-50 MN/sec) to abandon - it was easily beaten by Rybka at hardware with 2-3 MN/sec

A modern intel cpu with 4 cores is only 100,000 MIPS.

ARM is even much less. That is not problem, 200 MN/sec (million nodes search per second) of DeepBlue-2 are worse of 20 kn/sec (10 000x slower) DroidFish or Komodo

Nowadays, it’s the GPUs on the video card that have the most power in better PCs. You can have thousands of cores working in parallel. Not very powerful each, but good enough for simple alpha-beta searching.

They are both theoretically and practically slower than CPU of the same generation.
Chess algoritms rely on uniform memory array (hash table) which GPUs lack.

[yurikvelo]

For all we know, an engine with a 3000 rating might only be rated 2800 against humans. Then again, a 2800 human player might have a 2600 rating against computers. We simply have no way to know.

do you think any 2700+ GM (including Magnus Carlsen) will stand against weak engine (e.g. TCEC outsiders with 2400 elo)?