Closed LdDl closed 1 year ago
https://github.com/project-echo/lddi-contraction-hierarchies/ please take look at our fork , we already made so much improvement at there including this ,
I profiled this library under some serious request load in huge graphs , and optimized for performance , and it is running faster.
That's awesome job!
Current PR is composition of following commits in yours fork: https://github.com/project-echo/lddi-contraction-hierarchies/commit/a9fe55f14ec03869e09723af65b60d79cc90a5b5 https://github.com/project-echo/lddi-contraction-hierarchies/commit/b15c4243758e6e0a242cf9fa818b6fd217a2fc6e https://github.com/project-echo/lddi-contraction-hierarchies/commit/1851d23abd844c789d917d2c9cd5ac1cd224ae56 https://github.com/project-echo/lddi-contraction-hierarchies/commit/3205652c27aa7d03dd125c03c0c2d26703114556 https://github.com/project-echo/lddi-contraction-hierarchies/commit/f9fd927614f9af30f695d9eb961bcf232b084c9b (last two exclude each other)
am I right?
Also please check https://github.com/project-echo/lddi-contraction-hierarchies/pull/9/files . Reduced the response times for many to many twice . Benchmarks in repo are bit micro benchmarks , but we are trying with few hundred sources and destinations .
I'll take a look into for sure, but in another PR/commits set. Merging this one
Since early exit from
[][]bool
we can use[]map[k]v
Master branch:
Current branch:
benchcmp (yeah, it is deprecated)
:benchstat
: