Uploaded March 2021 | Updated September 2026, 1 week ago
I recently thought of an optimization for all clearing stages, which also saves a little time on the 4th center. This solve is a demonstration of it as well as the current WR. Final time 27:11.384, 4595 moves at 2.817 moves/sec.
The idea is to do an extra U (or U2 D2) move instead of a rotation, so that you can only move pieces into one area, and not have to group pieces together in one area and then move them all somewhere else. This saves 1 rotation, costs 1-2 extra moves, but should save many slice moves on average.
Here is the optimization written out in cube notation. Imagine you're clearing the m'th column, so mR is "an R turn of the m'th layer". I will also write "?U" for "some number of U turns on varying axes". The current center 1 clearing would look like this:
?U (mR') y' ?U' y' ?U2 y' ?U y' (mR)
The new method is this:
?U (mR' U) ?U y ?U2 y ?U' y (mR)
And for center 2 clearing, the current method is:
?U (mR U2 mR') y' ?U' y' ?U2 y' ?U y' (mR U2 mR')
The new method is:
?U (mR U2 mR' U) ?U y ?U2 y ?U' y (mR U2 mR')
For center 3 clearing, the current method (at a different angle from what I use) is:
?U2 (mR') y2 ?U2 y2 (mR)
The new method is:
?U2 (mR' U2 D2) ?U2 y2 (mR)
For center 4, you should only use this for the first half of the columns, and only for the first pass on each column. The old first pass would look like:
?U2 (mR U2 mR') y2 ?U2 y2 (mR U2 mR')
And the new first pass is:
?U2 (mR U2 mR' U2 D2) ?U2 y2 (mR U2 mR')
In all of these stages, feel free to do the additional U, U2, and D2 moves wherever they are convenient. You don't have to do them before the ?U2 moves.
I recently thought of an optimization for all clearing stages, which also saves a little time on the 4th center. This solve is a demonstration of it as well as the current WR. Final time 27:11.384, 4595 moves at 2.817 moves/sec.
The idea is to do an extra U (or U2 D2) move instead of a rotation, so that you can only move pieces into one area, and not have to group pieces together in one area and then move them all somewhere else. This saves 1 rotation, costs 1-2 extra moves, but should save many slice moves on average.
Here is the optimization written out in cube notation. Imagine you're clearing the m'th column, so mR is "an R turn of the m'th layer". I will also write "?U" for "some number of U turns on varying axes". The current center 1 clearing would look like this:
?U (mR') y' ?U' y' ?U2 y' ?U y' (mR)
The new method is this:
?U (mR' U) ?U y ?U2 y ?U' y (mR)
And for center 2 clearing, the current method is:
?U (mR U2 mR') y' ?U' y' ?U2 y' ?U y' (mR U2 mR')
The new method is:
?U (mR U2 mR' U) ?U y ?U2 y ?U' y (mR U2 mR')
For center 3 clearing, the current method (at a different angle from what I use) is:
?U2 (mR') y2 ?U2 y2 (mR)
The new method is:
?U2 (mR' U2 D2) ?U2 y2 (mR)
For center 4, you should only use this for the first half of the columns, and only for the first pass on each column. The old first pass would look like:
?U2 (mR U2 mR') y2 ?U2 y2 (mR U2 mR')
And the new first pass is:
?U2 (mR U2 mR' U2 D2) ?U2 y2 (mR U2 mR')
In all of these stages, feel free to do the additional U, U2, and D2 moves wherever they are convenient. You don't have to do them before the ?U2 moves.
