[en] May 1983 marked the beginning of an intensive effort to both improve the operating reliability, and improve the performance of the TMX-U Diagnostic Computer System. At that time, the system was handling (acquiring, storing, processing, plotting, displaying, and archiving) about 3 million bytes (Mb) of data per shot, with a 15-minute cycle time between shots. In addition, the system was fairly fragile, with frequent (about 5 times/day) crashes, requiring re-booting. At the present time, the system reliably handles about 5 Mb of data per shot, with a 7-minute cycle time between shots. This improvement was accomplished by a combination of new hardware, rearranging existing hardware, and new or revised software. Hardware changes were made in two areas. First, the shared disks were rearranged into different domains to make more efficient use of locking features. Second, we purchased and installed a solid-state RAM disk emulator (8 megabytes) to provide extremely fast access to lists and files that must be accessed frequently. In the software area, we made improvements in several areas. Initial effort went into finding bugs and optimizing existing code. We developed a template so that we could produce efficient code from applications that had first been developed on a very powerful, general-purpose scientific spread sheet. Also, we adopted a ticket book algorithm to allocate disk resources more efficiently