To use the solver press the Random button and after a scrambled cube appears press Solve. After some time the solve will complete and a green bar appears under the cube which is the media player control panel, press the Play button (right arrow) to see the solution. Subsequent solves will run faster, more about that in the "Search Time" section below.

To solve an actual scrambled cube press the Clear button and enter the colors by first clicking on a color in the palette (six squares above layout) then clicking on a square in the layout which changes it to the selected color. Press the Show button to update the cube on the right side of the screen to see that it matches.

Press the Solve button after you have colored the entire layout. It will first verify that the cube is valid and if not it will give a message indicating what the problem is. If you get an error, compare your cube to the one on the screen and find the square(s) that do not match and fix them, then press Solve again. Repeat until there are no more errors.

Once all the errors are fixed it will solve the cube and the green bar will appear, press the second button from the right to show the solution one move at a time.

If a solution is more than 20 moves then a shorter one can be found by increasing Search Time which is normally 5 seconds (by default). Search Time is just one part of the total processing time. Before the Solver can search it must first make several tables for moves & symmetry operations in addition to making the large search arrays (dist). That takes a significant amount of time which can be seen in the log where it shows "Init + Dist Time", only the first solve within a session incurs init & dist time. The total amount of time to solve a random cube varies depending on the computing capability of the device you are running it on.

After the solve completes the amount of search time used is shown below the Solve Log button. Also shown is the number of moves in each solve phase (phase 1 is the intermediate 3-color solution, phase 2 is the final solution). The display is in the form N+N which normally sums to the number of moves shown in the cube area (for solved cube). In some cases the sum will be larger which means that the first move in phase 2 was to the same face as the last move in phase 1 so the moves were either combined or if they cancel each other (R + R' for example) then both are removed from the sequence. Following the N+N display is the worker number that found the solution.

Search performance can be improved by generating the dist arrays to Depth 9 which takes longer initially, anywhere from 6 seconds (on 3.7 Ghz Quad-Core) to a minute or more depending on device performance. Generate time can be eliminated by using dist files as described in the next section.

By saving the dist arrays to files they only have to be generated once, then the files can be loaded into the Solver instead of generating. The dist arrays can be saved to files here or they can be downloaded at these links: Dist1 Dist2 DistP2. The content of these files is described in more detail here. To use, click on the Dist Files button (on main page), then press the Load Dist Files button and select the files (that you generated or downloaded) then click Open. After a few seconds a message will appear under the cube indicating that the files have been loaded. Prior to loading, the list of files needed by the solver can be seen by pressing Solve.

Better performance can be achieved by adding search array Dist3 (418 MB or 835 MB) or by adding both the larger Dist3 and Dist4 (1.8 GB) for the Optimal Solver. Your device or computer needs to have enough RAM so that the dist arrays fit into available memory. To use, download the file(s) of your choice and then click the appropriate button on the main page and load as described above.

To stop the search based on solution length rather than search time use the "s" option (append to Search Time as in 5s which is now the default). With a sufficiently long search time the solver will stop at the first solution of 20 moves or less but if it is too short then the search will be stopped before reaching the goal, in that case increase it.

The "s" option will remain in effect for subsequent solves, to switch back to the normal search (stop based on time only) append "t" to Search Time (as in 5t). With the "s" option, there will be a large variance in the amount of time it takes to find a solution (with 20 moves or less) so the wait may be long in some cases. It is possible to stop the search for solution lengths other than 20 moves by adding URL parameter stoplen=N where N is the desired number of moves.

The Quad-Core Solver will normally run as many concurrent processes as your device has processors but the number can be set to: 1, 2, 3, 4, 6, 8, 9, 10, 12, 14, 16 or 18 by adding parameter "conc=N" to the solver URL where N is the desired number of processes. The number of concurrent processes being used by the solver is shown in the log.

Each solver process searches for a solution starting with one or more of the 18 possible moves (R, R2, R', etc). With a conc setting of 2, 3, 6, 9 or 18, the 18 moves can be equally divided among the processes (to see the move assignments for all settings see function init_conc in solve.html). The number of processes does not have to match the number of processors, the operating system will balance the load if needed. While the conc parameter is available for experimenting, the default setting is likely fine for most cases.

The log normally shows search results for each concurrent process but some workers may not have found a solution before the search ends so to suppress display of those workers add param log_brief=1 to the solver URL (which is now the default).

The number of nodes generated during the search can be shown in the log by adding URL parameter show_node_counts=N (or snc=N) where N=1 for phase 1, N=2 for phase 2 or N=3 for both, the default is snc=1. For phase 1, three counts are shown at each depth which are the total number of nodes generated, the number of nodes that actually reached that depth (did not get pruned) and the number of 3-color solutions. Lower depths at which all nodes are immediately pruned (on first move) are excluded from the log. For phase 2, one line is shown after each solution which shows the totals for all depths searched to complete the solution from phase 1.

The phase 2 search can be eliminated by adding URL param use_p2seq=1 (which is now the default) which will store the move sequences for configs that can be solved in 8 moves or less (these are 6-color configs that are already solved as 3-color). The log will show 0 nodes for phase 2 if both params are used (use_p2seq=1 & show_node_counts=2 or 3).

The following AnimCubeJS parameters can be passed to the solver to customize the cube display (click on any link to see documentation).

The facelets parameter can be used with color codes: RGYBWO (Red, Green, Yellow, Blue, White, Orange), the Text Format is used (different from AnimCubeJS format) which can be entered using Text Input Cube, see Help on that page for the layout.

A file containing a list of cubes to solve can be loaded by adding URL parameter batch=1. The following is an example of three cubes as would be entered in a file (cube layout is text format as described above):

YBBWRGGOWRGYRGRBYYOOBRGRWYYRBBYWGOYGYBRGWOBRGOWWOOOWBW
WRWYROOGBGBBYORYYRBGOBGWGYRWBYRWRGBGWBOBORYYORWWOOWYGG
WBWOROBYYOWRYOOBBRBYBBGRBYOGBYRWWOWGWGWGYYGGYRWOROGGRR

When the batch=1 parameter is used, press the Load Cube File button to select and load the desired file. The first cube in the file will be displayed. The Next and Prev buttons can be used to navigate through the cubes in the file. Press Solve to solve all cubes in the file starting from the current position, the cube display does not change while the solver is running. The solve results can be seen immediately in the console during the run (press F12 and select the Console tab). When the run completes the results can be viewed in the Solve Log which looks like this for optimal solutions:

Init Time: 1.68
1 B D R' L' D' L' U L2 U F U' L' D L D2 L' U' D2 [17+1] (18f*) 12.11
2 F U2 F' B' R D' L2 F B' U2 R' B' U2 L2 U' D2 R2 F2 [15+3] (18f*) 9.98
3 B2 U2 B D' F R U' L U2 F' B2 L' U' D B2 U R' D2 [17+1] (18f*) 11.26
Search Time: 33.35

Logging of solve results to the console as shown above can also be done without using a cube file by adding URL parameter log_result=1.

The batch parameter can also be used to load the built-in files that are used for the benchmark tests shown in the performance summaries (quad-core, solver2), the batches are:

2   100 random cubes
3   100 cubes with 19 move solutions
4   1000 random cubes
5   77 presolved cubes with 16 move solutions
6   1000 presolved cubes

A batch of random cubes of any size can be generated using this, copy/paste the results to a file to use as input to the solver (with batch=1 param). A batch file can also be made from a file containing a list of solutions like the following example using this.

D' R' F' U' B' L2 D R F' U' F' R2 L' D2 R F L2 D2
U2 F' D' L2 U R2 D2 F2 L' U F2 L2 B' L' U2 B L
R2 B2 D R' B U L' B2 L' B D F2 L2 B' L B2 R