SPO600 Project Post 4

So this is the post where I'm testing my supposed optimizations. A quick recap, I made changes to Redis's version of the reallocate function. It will now only reallocate memory if the size of the new pointer is larger than the current size pointer, otherwise it will just 0 the block of memory.

These are the results of running the following command
-redis-benchmark -t set -n 1000000

which means we're running 1 million set commands with 50 parallel clients of 3 bytes payload.

These are the results without any optimizations



Average run time is about 79.3775 seconds

Here are the results after some preliminary modifications:

Average run time is about 81.8175 seconds. Wait... that isn't an optimization. It has become slower! After some slight checking of the "optimizations" I made, I realized I missed updating an area. Here are the new results after:

These are the more notable results. As I was testing it randomly throughout the day, I got some results that were 0.14s-0.2s faster to times that were sometimes up to 1.5s -2 s slower. As explained in class by Professor Tyler, I don't think the upstream will accept this code change, as it is not reliably faster AND also potential improvements are only roughly 2.5% gains.