There are two option setting values of "Splitting Option" in the fragment mapping.
One is "Split Length", another is "Overlap Length".
Why and how do I set them.

Reference genome sequences have better to be divided into fragments of same size because such split reference sequences require less memory and fast loading time than its whole sequence. Each neighboring pair of fragments in the reference sequences have better to overlap each other because reads to be mapped on the edges of fragments might be ignored unless the overlap regions exist.

The split length should be determined by the memory size of the GT installed computer, and the average read count that are mapped on one split fragment. 100,000 reads per one split fragment is a recommended parameter as the split length in case of the 8GB memory.
The overlap length should be determined by the read length. In case of 50bp for each read, 50 is recommended parameter as the overlap length even though there is a possibility of multiple mapping on the overlapped regions of the both sides of neighboring fragment.