Look into caching read/write indices in SPSC?

[sharpobject]

Hello,

I found an interesting post in which the author claims that he gets more throughput from storing a separate cached value of the read/write index of his SPSC queue and using that value until it no longer implies the operation is possible, and only then performing the atomic read of the uncached value. If you use the SPSC a bunch, maybe you could see if this helps.