|
|
Post by Michael Shen on Dec 24, 2021 6:40:22 GMT
Hi,
I am trying to find a schema for reporting the DIMM topology such as rows, columns, banks_group, bank_per_group, IO_width etc.(Those info are defined in the `DDR5 SPD contents specification`)
However I didn’t find any feasible schema for it. Does anyone have any suggestions?
If there is no feasible schema, then what is the procedure for proposing a new Redfish schema?
Michael
|
|
|
|
Post by Michael Shen on Dec 24, 2021 7:20:25 GMT
More further, can we have a SPD schema that reports all the information within SPD?
|
|
|
|
Post by mraineri on Dec 28, 2021 14:07:19 GMT
The Memory schema contains information regarding connectivity on the system; MemoryLocation contains the properties Socket, MemoryController, Channel, and Slot. But, I suspect given the examples you want more details of the internals of the DIMM.
For reporting the SPD contents, you can use the Assembly resource subordinate to Memory for reporting its contents (in the binary format). Within Assembly, there's a "BinaryDataUri" property that would allow a client to get the full, unparsed contents. Typically though we want to be able to translate information that clients would find useful in the appropriate resource though (in this case the Memory resource).
|
|
|
|
Post by Michael Shen on Dec 30, 2021 3:03:41 GMT
Thanks for your reply. We also hope that we can put the translated SPD information into the appropriate "Memory" resource, instead of directly putting the entire SPD raw data (binary format) into "BinaryDataUri" For this reason, I would like to ask if it is possible to add several new resources to the "Memory" schema? I listed them below and all of them are translated from the SPD raw. - rows (SPD raw byte 5 (0x005): First SDRAM Addressing)
- columns (SPD raw byte 5 (0x005): First SDRAM Addressing)
- bank_groups (SPD raw byte 7 (0x007): First SDRAM Bank Groups & Banks Per Bank Group)
- banks_per_bank_group (SPD raw byte 7 (0x007): First SDRAM Bank Groups & Banks Per Bank Group)
|
|
|
|
Post by mraineri on Jan 4, 2022 19:22:50 GMT
Is there a specific use case for understanding the rows, columns, groups, and banks for a particular DIMM? We'd like to avoid extraneous data if possible, and it's not clear how this could be used by a Redfish client.
|
|
|
|
Post by Michael Shen on Jan 11, 2022 3:24:08 GMT
Good catch. From my understanding those data should be consumed by other health-monitoring related service.
I will collect more information about who consumes the data and how they use the data.
Thanks!
|
|
