|
|
|
The hash function that c-tree uses to assign memory file records to memory file hash bins did not evenly distribute memory records among the hash bins for memory addresses having values over 4 GB. We have implemented an alternative hash function that uses more of the higher-order bits to compute the hash bin.
This new hash function is used on 64-bit systems. We also moved the memory file hash function computation before the call to request the I/O semaphore for the memory file. This might slightly improve performance when more than one client reads/writes a memory file at the same time.
