I've got a function that reads `map[string]interface{}` off a channel,
processing each one until that channel is closed. Processing each `map`
entails allocating some slices and other data structures, grouping by like
keys, etc.
I've got a benchmark that reads 10000 rows of JSON from a file, unmarshals
each to a `map[string]interface`, sets up channels, runs the aforementioned
function in a Goroutine, and sends each unmarshaled JSON over the input
channel of the function, etc., etc.
When I profile CPU usage, I get insightful output, and I've managed to
improve performance by a factor of 4.
However, when I run the benchmark with `memprofile` flag and point pprof at
it, I get something like:
0 of 0 total ( 0%)
flat flat% sum% cum cum%
0 0% 0% 0 0% pkgname.SomeFunc
0 0% 0% 0 0% pkgname.SomeOtherFunc
I'm running Go 1.6 on Ubuntu 12.04.
I've searched and searched but not found any answers as to why I am unable
to obtain a good heap profile. `runtime.mallocgc` features prominently in
my cpu profiles, so I know objects are getting allocated and collected.
Any help?
Thanks
--
You received this message because you are subscribed to the Google Groups
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to golang-nuts+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
