@@ -43,10 +43,10 @@ This implementation derives from or is based on the ideas presented in:
43
43
- "Worst case fragmentation of first fit and best fit storage allocation strategies" -- J. M. Robson, 1975.
44
44
45
45
This implementation does not make any non-trivial assumptions about the behavioral properties of the application.
46
- Per Herter [ 2014] , it can be described as a * predictably bad allocator* :
46
+ Per Herter \ [ 2014] , it can be described as a * predictably bad allocator* :
47
47
48
48
> An allocator that provably performs close to its worst-case memory behavior which, in turn,
49
- > is better than the worst-case behavior of the allocators discussed [ in Herter 2014] ,
49
+ > is better than the worst-case behavior of the allocators discussed \ [ in Herter 2014] ,
50
50
> but much worse than the memory consumption of these for normal programs without (mostly theoretical)
51
51
> bad (de-)allocation sequences.
52
52
@@ -102,7 +102,7 @@ catastrophic fragmentation cannot occur.
102
102
103
103
The above-defined theoretical worst-case upper bound H may be prohibitively high for some
104
104
memory-constrained applications.
105
- It has been shown [ Robson 1975] that under a typical workload,
105
+ It has been shown \ [ Robson 1975] that under a typical workload,
106
106
for a sufficiently high amount of memory available to the allocator which is less than $H$,
107
107
the probability of a (de-)allocation sequence that results in catastrophic fragmentation is low.
108
108
When combined with an acceptable failure probability and a set of adequate assumptions about the behaviors of
@@ -111,7 +111,7 @@ the heap while ensuring a sufficient degree of predictability and reliability.
111
111
The methods of such optimization are outside the scope of this document;
112
112
interested readers are advised to consult with the referred publications.
113
113
114
- Following some of the ideas from [ Herter 2014] , this implementation takes caching-related issues into consideration
114
+ Following some of the ideas from \ [ Herter 2014] , this implementation takes caching-related issues into consideration
115
115
by choosing the most recently used memory fragments to minimize cache misses in the application.
116
116
117
117
### Implementation
0 commit comments