In this video, I optimized the C++ bindings of my souffle-haskell library.
Before this change, I allocated one big contiguous bytearray each time data was sent from Haskell to C++. After the change, a previously allocated bytearray is reused. Even with GHC Haskell's pinned memory bytearrays (not as expensive to allocate as normal
new in C or C++), this ended up saving 5-10% time in terms of performance.
The code can be found in this pull request. This also includes the optimizations I did when serializing from C++ to Haskell, where I applied the same trick to get a 30-40% performance improvement.