I've been monitoring the PCIe Read packets on my Linux box. Much to my surprise, when requesting a block read in the application software, this request is broken up into many length one memory read requests. Is there a way to improve this?
After searching the internet, I stumbled across this document: http://edc.intel.com/Link.aspx?id=3949 http://edc.intel.com/Link.aspx?id=3949
On page 53, in the pdf, it states:
As opposed to the write-case, there is no such thing like "read combining". However, there is a
My 2nd question is: How do you enable IO caching in Linux?
For more complete information about compiler optimizations, see our Optimization Notice.