Congratulations to Komodo, Protector, Rybka, Junior and Chiron for qualifying for Stage 2 after Stage 1c ended yesterday. They will be joined by the top 5 from Stage 1a and Stage 1b, as well as Equinox as the overall best 6th placed engine making the total engines 16.
Stage 2:
16 cores, ponder off. Up to 16 GB hash. Time control is 120' + 30".
Single round robin.
Top 8 qualifies for Stage 3.
In Gull vs Spike, the reported nodes for Gull are wrong.
It shows, for example at move 26:
Speed: 26778 KN/s
Move time: 00:03:03
Nodes:307.7 M
But if you take a calculator, and you multiply the Speed by Move time (in seconds, so 183 seconds here), you get 4904.4 M nodes, more than 10 times more than the displayed "307.7 M" nodes.
It does not happen to Spike.
What's going on here?
Isaac wrote:In Gull vs Spike, the reported nodes for Gull are wrong.
It shows, for example at move 26:
Speed: 26778 KN/s
Move time: 00:03:03
Nodes:307.7 M
But if you take a calculator, and you multiply the Speed by Move time (in seconds, so 183 seconds here), you get 4904.4 M nodes, more than 10 times more than the displayed "307.7 M" nodes.
It does not happen to Spike.
What's going on here?
P.S.:Both are UCI engines.
It's just a matter of Gull not sending a final node count update to the GUI before making the move. So the GUI uses the "last" data it sent which was probably much earlier (15 seconds?), considering it was only 307.7 million nodes.
Martin Thoresen wrote:
It's just a matter of Gull not sending a final node count update to the GUI before making the move. So the GUI uses the "last" data it sent which was probably much earlier (15 seconds?), considering it was only 307.7 million nodes.