In the game lichess.org/y7aV0BwR the evaluation graph in the analysis shows a huge advantage for black from move 25 on, but stockfish correctly evaluates it as clearly better for white when I use the local evaluation.
I don't think that's a bug, as evalutation graph shows eval at small depth so Black has a N for 2pawns. At higher depth SF see that White has huge compensation for the N.
It *is* a bug. I loaded the game in a standard chess app and in that app Stockfish correctly evaluated the position as getting better and better for white after move 25. Do you see that in move 36 the position is evaluated as -51.5? That is outright absurd - white is winning.
SF analysis for my games has been acting very erratic of late. For some reason along with the usual analysis which starts at the end of games, at the same time the graph shows the 1st half dozen moves with the bar doing funny things. It even skipped over the seventh move, with no evaluation... just blank.
Also today on the mobile usually it boops when it has finished evaluating but now after asking for evaluation it boops every 2 seconds and disrupts from the position I am looking at locally.
True, the SF analysis has been acting extremely strange the past few days. This even applies to variants.
The past couple of days I have seen this on both the iPad and iPhone app. Perhaps it is an iOS problem?
Nope, must be more general. I'm using chromium on a linux machine.
The bug seems to be taken care of. At least, the analysis of the above game is correct now. Thanks to whoever did this.
This topic has been archived and can no longer be replied to.