Adding initial text

docs/System_GPU_Specification_v0.1.md

@@ -6,8 +6,8 @@ supersedes: none
 status: draft
 released: true
 class: info
-date: 2025-09-12
-copyright: 2025
+date: 2026-01-12
+copyright: 2026
 paragraph_numbering: no
 bibliography: bibliography.yaml
 header-includes: |
@@ -30,49 +30,116 @@ License](https://creativecommons.org/licenses/by-sa/4.0/).
 
 # Scope
 
-This document specifies the requirements for managing a GPU in a system.
+This document specifies the requirements for managing GPUs within a system.
 
-# Requirements
+## Scoping a GPU System
 
-## Manageability Interface
+The scope of a GPU system may span multiple-racks. This definition is taken from the system managment section of the [Open Systems of AI whitepaper](https://www.opencompute.org/documents/ocp-open-systems-for-ai-whitepaper-v1-0-0-final-pdf) [1].
 
-The system shall expose a Redfish Service (v??).
+The [OCP Open Cluster Designs for AI](hhttps://www.opencompute.org/documents/white-paper-open-cluster-designs-for-ai-final-14-january-2026-pdf) [2] refers to the functional and physical aspects of an AI system as 'cluster' and 'pod', respectively.  Representing the physical aspects, the pod contains the power and cooling systems.
 
-## Redfish data model
+Both documents describe the possible management hierarchy (from node, to rack, to cluster, to DC manager), and specify Redfish (tm) [6] as the out-of-band management interface for the system [6]
+
+At the top level of the Redfish model [7], there exista the the 'System' resource and the 'Chassis' resource to represent the functional and physical aspects of a managed entity, respectively.    The model also contains  'Manager' resource to represent the management hierarchy of a managed entity. *TODO - add high level diagram*.
+
+
+** TBD - Are disaggregated GPU Systems included? - with GPU shelves, JBODs, etc**
+
+# Redfish Requirements
+
+The Redfish interface shall conform to the [Redfish Specification](https://www.dmtf.org/dsp/DSP0266) [6]
+
+## GPU node manageability
 
 The Redfish Service shall conform to the following OCP Profiles
-* [UBB_BaselineManagement profile v1.x](https://github.com/opencomputeproject/HWMgmt-OCP-Profiles/blob/master/gpu/OCP_UBB_BaselineManagement.v1.0.0.json)
-* [OCP Server Profile v1.1](https://github.com/opencomputeproject/HWMgmt-OCP-Profiles/blob/master/Server/OCPServerHardwareManagement.v1_1_0.json)
 
-### Firmware Management
-### Redfish Component Integrity
+| **Profile**         | **Usage Guide**             
+| :---                  | :-----------       |
+| [UBB_BaselineManagement](https://github.com/opencomputeproject/HWMgmt-OCP-Profiles/blob/master/gpu/OCP_UBB_BaselineManagement.v1.0.0.json) | [OCP GPU & Accelerator Management Interfaces](https://www.opencompute.org/documents/ocp-gpu-accelerator-management-interfaces-v1-1-pdf)  [3] |
+| [OCP Server](https://github.com/opencomputeproject/HWMgmt-OCP-Profiles/blob/master/Server/OCPServerHardwareManagement.v1_1_0.json) | [Usage Guide for OCP Server Management](https://www.opencompute.org/documents/usage-guide-server-hw-mgmt-profile-v1-0-0final-pdf-1) |
+| [OCP GPU Management]() | Usage Guide
+
+## GPU cluster/pod manageability
+
+From the Usage Guide for Rack Management (OpenRMC) v1.2
+
+| **Use Case**        | **Management Task**                                       | **Requirement** |
+| :---                | :-----------                                              | :---	|
+| Composability       | [Construct a system with GPUs](#construct-a-system-with-gpus)     | Recommended |
+|                     | [Get Health of GPUs from composed compute system](#get-health-of-gpus-from-composed-compute-system) | Recommended |
+|                     | [Set policy when a GPU in a composed system fails](#set-policy-when-a-gpu-in-a-composed-system-fails) | Recommended |
+
+## Testing Conformance to Profiles
+
+The [Redfish Interop Validator](https://github.com/DMTF/Redfish-Interop-Validator) reads a profile. The validator autogenerates tests, executes the tests against an implementation, and generated a report.  The validator is invoked the the following command.
+
+  \$\> python3 RedfishInteropValidator.py profileName --ip host:port
 
 ## Redfish Aggregation Behavior
 
-The system should follow the guidance of the *(insert document name*) (GPU specific version)
+In order to improve interoperablity, the behavior of the Redfish data model for the system should be consistent . The guidance of the ]Redfish Aggregation Guidance for Compute Expansion Modules](https://www.dmtf.org/sites/default/files/standards/documents/DSP2090_1.0.0.pdf) [20] should be followed.
+
+## Firmware Management
+
+The Redfish Service should support the MultipartHttpPushUri. If MultipartHttpPushUri property is not supported, then the HttpPushUri property shall be supported.
+
+*Note*: The HttpPushUri property is 'deprecated' in favor of the MultiPartHttpPushUri property.
+
 
 ## Redfish Message Registry
 
 # PLDM Requirements
 
+## Firmware Management
+
+The GPU shall support the requirements specified in the ["GPU Firmware Update Requirements v1.1"](https://www.opencompute.org/documents/ocp-gpu-fw-update-specification-v1-1-pdf)
+
+The CPU shall support the requirements specified in the "CPU Hyperscale Impactless Firmware Update v1.0". The current draft is [v0.7](https://www.opencompute.org/documents/hyperscale-cpu-impactless-firmware-updates-requirements-specification-v0-7-9-29-2025-pdf) (2025).
+
+## Sensors and Effecters
+
 # Security Requirements (SPDM)
 
+## Redfish Component Integrity
+
 # Physical Interfaces and connections
+
 ## USB Requirements
 
 # Performance requirements (e.g. KPIs)
 
-# Testing Conformance to Profiles
 
-The [Redfish Interop Validator](https://github.com/DMTF/Redfish-Interop-Validator) reads a profile. The validator autogenerates tests, executes the tests against an implementation, and generated a report.  The validator is invoked the the following command.
+# References
 
-  \$\> python3 RedfishInteropValidator.py profileName --ip host:port
+\[1\] "[OCP Open Systems of AI Whitepaper](https://www.opencompute.org/documents/ocp-open-systems-for-ai-whitepaper-v1-0-0-final-pdf)"
+
+\[2\] "[OCP Open Cluster Designs for AI](hhttps://www.opencompute.org/documents/white-paper-open-cluster-designs-for-ai-final-14-january-2026-pdf)"
+
+\[3\] "[OCP Usage Guide for Managing a Server](https://www.opencompute.org/documents/usage-guide-server-hw-mgmt-profile-v1-0-0final-pdf-1)"
+
+\[4\] "[OCP Usage Guide for Managing a Rack (OpenRMC)](https://www.opencompute.org/documents/openrmc-usageguide-r1-1-draft0-6-1-1-pdf)"
+
+\[5\] "[OCP GPU & Accelerator Management Interfaces v1.1](https://www.opencompute.org/documents/ocp-gpu-accelerator-management-interfaces-v1-1-pdf)"
+
+\[6\] "[Redfish Specification](https://www.dmtf.org/dsp/DSP0266)"
+
+\[7\] "[Redfish Data Model Specification](https://www.dmtf.org/dsp/DSP0268)"
+
+\[8\] "[Redfish Interop Validator](https://github.com/DMTF/Redfish-Interop-Validator)"
+
+\[9\] "[Redfish Interoperability Profile Specification](https://www.dmtf.org/dsp/DSP0272)"
+
+\[10\] "[UBB_BaselineManagement profile v1.x](https://github.com/opencomputeproject/HWMgmt-OCP-Profiles/blob/master/gpu/OCP_UBB_BaselineManagement.v1.0.0.json)
+
+\[11\] "[OCP Server Profile v1.1](https://github.com/opencomputeproject/HWMgmt-OCP-Profiles/blob/master/Server/OCPServerHardwareManagement.v1_1_0.json)
 
+\[12\] "[OCP GPU Management Profle vx.x]()
 
+\[20\] "[Redfish Aggregation Guidance for Compute Expansion Modules](https://www.dmtf.org/sites/default/files/standards/documents/DSP2090_1.0.0.pdf)" 
 
 # Revision 
 
 | Revision/Version | Date | Description |
 |----|----|----|
-| 0.1 | 9/12/2025 |initial draft |
+| 0.1 | 1/12/2026 |initial draft |