System and method for performance optimal partial rank/bank interleaving for non-symmetrically populated DIMMs across DDR channels
View Patent ↗An information handling system includes a processor having a plurality of memory channels. The information handling system also includes a plurality of dual inline memory modules non-symmetrically populated on the memory channels. The dual inline memory modules are divided by bank to create a plurality of interleave groups, and each of the interleave groups spans across all of the memory channels of the processor.
1. A method comprising:
determining, by a processor, that a plurality of dual inline memory modules are populated non-symmetrically across a plurality of memory channels;
dividing, by the processor, the dual inline memory modules by bank to create a plurality of interleave groups, wherein each of the interleave groups spans across all of the memory channels, and consecutive accesses to a single interleave group are to banks of different dual inline memory modules on different memory channels;
dividing all of the dual inline memory modules of a first type by bank to create a plurality of interleave groups, wherein each of the interleave groups that include the dual inline memory modules of the first type spans across all of the memory channels; and
creating a separate interleave group for a first dual inline memory module of the dual inline memory modules in response to the first dual inline memory module being a second type of dual inline memory module.
2. The method of claim 1 , wherein each of the interleave groups has the same performance level as the other interleave groups.
3. The method of claim 1 , wherein the dual inline memory modules are populated non-symmetrically across the memory modules in response to a first memory channel having more dual inline memory modules populated than the other memory channels have populated.
4. The method of claim 1 , wherein the dual inline memory modules are populated non-symmetrically across the memory modules in response to a first memory channel having a different type of dual inline memory module than the other memory channels have populated.
5. The method of claim 1 , further comprising:
receiving a user request to have consistent performance across all interleave groups prior to allocating first and second interleave groups of the interleave groups.
6. An information handling system comprising:
a processor including a plurality of memory channels; and
a plurality of dual inline memory modules non-symmetrically populated on the memory channels, wherein the dual inline memory modules are divided by bank to create a plurality of interleave groups, wherein each of the interleave groups spans across all of the memory channels of the processor, and consecutive accesses to a single interleave group are to banks of different dual inline memory modules on different memory channels, wherein all of the dual inline memory modules of a first type are divided by bank to create a plurality of interleave groups, wherein each of the interleave groups that include the dual inline memory modules of the first type spans across all of the memory channels, and a separate interleave group for a first dual inline memory module of the dual inline memory modules is created in response to the first dual inline memory module being a second type of dual inline memory module.
7. The information handling system of claim 6 , wherein each of the interleave groups has the same performance level as the other interleave groups.
8. The information handling system of claim 6 , wherein the dual inline memory modules are populated non-symmetrically across the memory modules in response to a first memory channel having more dual inline memory modules populated than the other memory channels have populated.
9. The information handling system of claim 6 , wherein the dual inline memory modules are populated non-symmetrically across the memory modules in response to a first memory channel having a different type of dual inline memory module than the other memory channels have populated.
10. The information handling system of claim 6 , wherein the first type of dual inline memory module is a volatile dynamic random access memory based dual inline memory module and the second type of dual inline memory module is a non-volatile memory based dual inline memory module.
11. The information handling system of claim 6 , wherein the allocation of first and second interleave groups of the interleave groups is in response to a user request to have consistent performance across all interleave groups being received.
12. A method comprising:
determining, by a processor, that a plurality of dual inline memory modules are populated non-symmetrically across a plurality of memory channels;
dividing, by the processor, the dual inline memory modules by rank and bank to create a plurality of interleave groups, wherein each of the interleave groups spans across all of the memory channels, wherein banks of a first rank of one of the dual inline memory modules are included in each of the interleave groups, and consecutive accesses to a single interleave group are to banks of different dual inline memory modules on different memory channels;
dividing all of the dual inline memory modules of a first type by rank to create a plurality of interleave groups, wherein each of the interleave groups that include the dual inline memory modules of the first type spans across all of the memory channels; and
creating a separate interleave group for a first dual inline memory module of the dual inline memory modules in response to the first dual inline memory module being a second type of dual inline memory module.
13. The method of claim 12 , wherein each of the interleave groups has the same performance level as the other interleave groups.
14. The method of claim 12 , wherein the dual inline memory modules are populated non-symmetrically across the memory modules in response to a first memory channel having more dual inline memory modules populated than the other memory channels have populated.
15. The method of claim 12 , wherein the dual inline memory modules are populated non-symmetrically across the memory modules in response to a first memory channel having a different type of dual inline memory module than the other memory channels have populated.
16. The method of claim 12 , further comprising:
receiving a user request to have consistent performance across all interleave groups prior to allocating first and second interleave groups of the interleave groups.
17. The method of claim 12 , wherein a first dual inline memory module and the dual inline memory modules includes two ranks, and a second dual inline memory module of the dual inline memory modules includes only one rank.