Re: [go-nuts] Re: The right way for golang binary to reduce iTLB cache miss

2021-09-17 Thread rmfr
Thanks a lot :-D > This is the icache or code - not the data the code is accessing. Yes, it is the cache miss of iTLB, i.e. *Instruction* Translation Lookaside Buffer. I have tried the https://github.com/facebookincubator/BOLT, but the linker Golang used seems does not allow `--emit-relocs`

Re: [go-nuts] Re: The right way for golang binary to reduce iTLB cache miss

2021-09-17 Thread Robert Engels
This is the icache or code - not the data the code is accessing. > On Sep 17, 2021, at 8:12 AM, peterGo wrote: > >  > rmfr, > > What is your data, what is its structure, and what are the access paths and > frequencies? > > Your data is unlikely to be a randomly accessed blob of bits. Is

[go-nuts] Re: The right way for golang binary to reduce iTLB cache miss

2021-09-17 Thread peterGo
rmfr, What is your data, what is its structure, and what are the access paths and frequencies? Your data is unlikely to be a randomly accessed blob of bits. Is there a set of Pareto access paths that you can use to order your data by access frequency? You want the most frequently accessed