![]() |
LINE Solver (C++)
Templated C++ port of the LINE queueing solver
|
What the cache analyzer returns beyond the usual metric table. More...
#include <line/solvers/nc/solver_nc_cache.h>
Public Attributes | |
| NcSolution< T > | sol |
| Matrix< T > | pij |
| (n x h+1) per-item occupancy, column 0 = miss | |
| Matrix< T > | itemprob |
| (n x h+1) the same from the EXACT recursion, NaN above 10 items | |
| Matrix< T > | hitproblist |
| (u x h) access-weighted per-list hit probability, NaN if not exact | |
| std::vector< T > | missrate |
| (u) per-class miss rate | |
| std::vector< T > | missprob |
| (u) per-class miss and hit PROBABILITIES, missrate divided by the read class's arrival rate and its complement. | |
| std::vector< T > | hitprob |
| std::vector< T > | listcost |
| (h) mean storage cost held by each list, EMPTY without item sizes | |
| std::vector< cache::CacheBlockedPair > | costcap_blocked |
| Storage cost caps that block a promotion path, so that the exact recursion normalizes over MORE states than the cache can reach. | |
| bool | costcap_method_switched = false |
| True when the requested method had no cost-capped counterpart and was switched. | |
What the cache analyzer returns beyond the usual metric table.
Definition at line 76 of file solver_nc_cache.h.
| std::vector<cache::CacheBlockedPair> line::nc::NcCacheSolution< T >::costcap_blocked |
Storage cost caps that block a promotion path, so that the exact recursion normalizes over MORE states than the cache can reach.
The port has no warning channel, so this is a FLAG, like SjnResult::capped: a non-empty vector means cross-check the answer with SolverLDES.
Definition at line 99 of file solver_nc_cache.h.
Referenced by line::nc::solver_nc_cache_analyzer().
| bool line::nc::NcCacheSolution< T >::costcap_method_switched = false |
True when the requested method had no cost-capped counterpart and was switched.
Definition at line 101 of file solver_nc_cache.h.
Referenced by line::nc::solver_nc_cache_analyzer().
| std::vector<T> line::nc::NcCacheSolution< T >::hitprob |
Definition at line 91 of file solver_nc_cache.h.
Referenced by line::nc::solver_nc_cache_analyzer(), and line::nc::solver_nc_solve().
| Matrix<T> line::nc::NcCacheSolution< T >::hitproblist |
(u x h) access-weighted per-list hit probability, NaN if not exact
Definition at line 80 of file solver_nc_cache.h.
Referenced by line::nc::solver_nc_cache_analyzer(), and line::nc::solver_nc_solve().
| Matrix<T> line::nc::NcCacheSolution< T >::itemprob |
(n x h+1) the same from the EXACT recursion, NaN above 10 items
Definition at line 79 of file solver_nc_cache.h.
Referenced by line::nc::solver_nc_cache_analyzer(), and line::nc::solver_nc_solve().
| std::vector<T> line::nc::NcCacheSolution< T >::listcost |
(h) mean storage cost held by each list, EMPTY without item sizes
Definition at line 92 of file solver_nc_cache.h.
Referenced by line::nc::solver_nc_cache_analyzer(), and line::nc::solver_nc_solve().
| std::vector<T> line::nc::NcCacheSolution< T >::missprob |
(u) per-class miss and hit PROBABILITIES, missrate divided by the read class's arrival rate and its complement.
Carried beside the rate because getAvgCacheTable reports a probability and the two differ by a factor no consumer downstream can recover: the arrival rate is the Source's and is gone by the time the table is built. NaN for a class with no read stream, which is not a hit ratio of zero.
Definition at line 91 of file solver_nc_cache.h.
Referenced by line::nc::solver_nc_cache_analyzer(), and line::nc::solver_nc_solve().
| std::vector<T> line::nc::NcCacheSolution< T >::missrate |
(u) per-class miss rate
Definition at line 81 of file solver_nc_cache.h.
Referenced by line::nc::solver_nc_cache_analyzer().
| Matrix<T> line::nc::NcCacheSolution< T >::pij |
(n x h+1) per-item occupancy, column 0 = miss
Definition at line 78 of file solver_nc_cache.h.
Referenced by line::nc::solver_nc_cache_analyzer().
| NcSolution<T> line::nc::NcCacheSolution< T >::sol |
Definition at line 77 of file solver_nc_cache.h.
Referenced by line::nc::solver_nc_cache_analyzer(), and line::nc::solver_nc_solve().