True 64-bit Release
This release is the first truely 64-bit version of UniData, built to utilize the power of the 64-bit architecture. This means you can address more memory while running programs, this could be useful if you are working with large string manipulations or U2 Dynamic Objects. UniData 8.1 also now uses memory more efficiently, using heap instead of shared memory for Basic variables.
NOTE: Starting at 8.1, NO more 32-bit builds on any platform, only 64-bit builds.
Tired of bumping into that pesky 2G file size limit with 32-bit files?
Worry no more since 8.1 introduces 64-bit files. This allows you to simplify administration by having static files well over 2GB, as well as improving dynamic files to be able to have 2GB+ part files as well.
Speaking of dynamic files, we’ve improved the split/merge algorithm of dynamic files to more effectively manage the split and merge process, again improving performance. This new dynamic file split type known as WHOLEFILE, helps better manage your dynamic files by splitting based on the entire file, rather than individual groups.
New Hashing Algorithm
Achieving an even distribution of keys is the goal of any good hashing algorithm, which would improve performance when doing reads, writes, and selects. So we’ve added a new hash type for files, which is very good at hashing keys of any type. This uses the industry standard NIST algorithm and should help avoid overflow and improve performance. Use hash type 3 for this new hashing algorithm.
Recoverable File System Improvements
There were several improvements made to improve RFS efficiency and stability, including:
- A new tool to identify potential problem files
- Increased page size to 64k for UniData processes improves efficiency
- More control with UDT daemon process priority management
- Improved replication and RFS handling regarding dbpause