Various performance improvements to parsing
cover Various performance improvements to parsing
This patch series fixes up profiling of the parsing processes and then uses that to find some performance improvements in parsing speed. What follows is a quick overview of what each patch is worth using "time". The data was confirmed in the actual profiles through a reduction in function calls, this is really just an interesting side by side comparison of the significance of each change:
real 0m31.700s
user 1m49.140s
sys 0m2.272s
#### Baseline ####
real 0m31.498s
user 1m48.448s
sys 0m2.380s
#### data: Be explicit in data_db check
real 0m31.079s
user 1m47.420s
sys 0m2.400s
#### siggen: Use lookup cache exclusively
real 0m30.890s
user 1m46.684s
sys 0m2.492s
#### data_smart: Improve variable expansion regexp
real 0m30.547s
user 1m44.124s
sys 0m2.292s
#### data_smart: use the expand_cache in VariableParse
real 0m29.750s
user 1m41.240s
sys 0m2.536s
#### data: Use direct iteration, not keys()
real 0m28.610s
user 1m38.096s
sys 0m2.304s
#### data: Cache an list of export variables
real 0m28.718s
user 1m38.080s
sys 0m2.344s
### data_smart: Add explict None checks
real 0m28.726s
user 1m37.924s
sys 0m2.568s
### data_smart: Allow expansion of flags in getVarFlags
real 0m28.947s
user 1m38.564s
sys 0m2.644s
### data_smart: Allow flags to use the expand cache
real 0m29.135s
user 1m38.816s
sys 0m2.280s
### data_smart: Cache the fact a variable accesses another even if its unset
real 0m28.250s
user 1m36.132s
sys 0m2.340s
### data: Optimise flag exclusion list handling
real 0m28.584s
user 1m36.116s
sys 0m2.376s
### data: Optimise flag lookup in build_dependencies
real 0m26.662s
user 1m28.952s
sys 0m2.296s
### data: Optimise build_dependencies a little
Cheers,
Richard