| CPC G06F 3/0616 (2013.01) [G06F 3/0635 (2013.01); G06F 3/0659 (2013.01); G06F 3/0673 (2013.01); G06F 12/1009 (2013.01); G06F 2212/1036 (2013.01)] | 20 Claims |

|
1. A method, comprising:
determining, via a long short-term memory deep-learning recurrent neural network executed on a memory controller, a first assignment that causes a first logical address to correspond to a first physical address for a first memory portion;
determining, via the long short-term memory deep-learning recurrent neural network executed on the memory controller, a first memory access latency associated with the first assignment based on a first set of memory access patterns;
generating, via the long short-term memory deep-learning recurrent neural network executed on the memory controller, a second assignment that causes the first logical address to correspond to a second physical address for a second memory portion and a third assignment that causes the first logical address to correspond to a third physical address for a third memory portion; and
selecting, via the long short-term memory deep-learning recurrent neural network executed on the memory controller, the second assignment or the third assignment based on the first memory access latency, a second memory access latency associated with the second assignment based on a second set of memory access patterns, and a third memory access latency associated with the third assignment based on a third set of memory access patterns.
|