Only way to change that is by forcing 8 via the rclone.conf.
See the output from yesterday: 2019/07/26 01:27:36 INFO : gcache: Chunk Clean Interval: 1m0s 2019/07/26 01:27:36 INFO : gcache: Workers: 4 2019/07/26 01:27:36 INFO : gcache: File Age: 1w
My bad - I claim heatstroke as my defense! (and when it's in Norway you know it's real bad )
I think I must have confused it with the default for checkers.
Number of workers = number of simulations downloads of chunks to the cache - so it will to some extent help hide the "sawtooth" pattern that the TCP generates due to small chunks.
In terms of bandwith optimization there will be some balance-point between your chunk-size and the number of workers that is ideal for your setup. Increasing either/both of these will increase the time to open a new streamable file, but you want a balance between the two to get the optimal result. In terms of how the nitty-gritty of it works it seems like if you have 8 workers you will start downloading 8 chunks immediately - and then try to keep that amount of chunks in cache. That will obviously slow down the first chunk from finishing (slowing opening time) but also provide a bigger buffer for high-bitrate media.