This commit (b1f8356) includes two optimizations:
(1) When resetLogTail() is called, do not allocate a new backing array for the log arrays. Instead reuse the existing array by zeroing out the pointers
(2) When the transaction unLock() function is called, do not allocate a new array to hold the mutex pointers. Instead reuse the same array by zeroing out the pointers.
Memtier benchmark throughput observed while running against go-redis-pmem:
This commit (b1f8356) includes two optimizations: (1) When resetLogTail() is called, do not allocate a new backing array for the log arrays. Instead reuse the existing array by zeroing out the pointers (2) When the transaction unLock() function is called, do not allocate a new array to hold the mutex pointers. Instead reuse the same array by zeroing out the pointers.
Memtier benchmark throughput observed while running against go-redis-pmem:
Memtier benchmark configurations: threads=1 ratio=1:1