Merge pull request #285017 from bryan-strassner/bstrassner/SIunits

Clarify references to GB to indicate GiB where applicable

Author: Stacyrch140

articles/operator-nexus/concepts-compute.md

@@ -41,11 +41,11 @@ Azure Operator Nexus reserves a small set of CPUs for the host operating system
 
 ### Huge page support
 
-Huge page usage in workloads refers to the utilization of large memory pages, typically 2 MB or 1 GB in size, instead of the standard 4-KB pages. This approach helps reduce memory overhead and improves the overall system performance. It reduces the translation look-aside buffer (TLB) miss rate and improves memory access efficiency.
+Huge page usage in workloads refers to the utilization of large memory pages, typically 2 MiB or 1 GiB in size, instead of the standard 4 KiB pages. This approach helps reduce memory overhead and improves the overall system performance. It reduces the translation look-aside buffer (TLB) miss rate and improves memory access efficiency.
 
-Workloads that involve large data sets or intensive memory operations, such as network packet processing, can benefit from huge page usage because it enhances memory performance and reduces memory-related bottlenecks. As a result, users see improved throughput and reduced latency.
+Workloads that involve large data sets or intensive memory operations such as network packet processing, can benefit from huge page usage because it enhances memory performance and reduces memory-related bottlenecks. As a result, users see improved throughput and reduced latency.
 
-All virtual machines created on Azure Operator Nexus can make use of either 2-MB or 1-GB huge pages, depending on the type of virtual machine.
+All virtual machines created on Azure Operator Nexus are backed by 1GiB(1G) hugepages for the requested memory.  The kernel running inside the VM can manage these available memory anyway it likes, including the allocation of memory to support hugepages (2M or 1G).
 
 ### Dual-stack support
 

articles/operator-nexus/howto-kubernetes-cluster-huge-pages.md

@@ -33,7 +33,7 @@ When configuring huge-pages for a Nexus Kubernetes cluster, you need to provide
 ## Limitations
 Nexus Kubernetes cluster enforces the following constraints to ensure proper configuration:
    * The total size of huge-pages (HugepageSize multiplied by HugepageCount) must not exceed 80% of the VM's memory.
-   * At least 2 GB of memory must be left for the host kernel after allocating huge-pages.
+   * At least 2 GiB of memory must be left for the host kernel after allocating huge-pages.
    * If the huge-page size is ```2M```, the huge-page count must be a power of 2 (for example, 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048, 4096, 8192, etc.).
    * Huge-page configurations must be done during agent pool creation, it can't be added or modified after.
 

articles/operator-nexus/includes/virtual-machine/virtual-machine-arm-template.json

@@ -62,7 +62,7 @@
             "type": "int",
             "defaultValue": 4,
             "metadata": {
-                "description": "The memory size of the virtual machine in GB (max 224 GB)"
+                "description": "The memory size of the virtual machine in GiB (max 224 GiB)"
             }
         },
         "networkAttachments": {

articles/operator-nexus/includes/virtual-machine/virtual-machine-bicep-template.bicep

@@ -28,7 +28,7 @@ param cloudServicesNetworkId string
 @description('Number of CPU cores for the virtual machine. Choose a value between 2 and 46.')
 param cpuCores int = 2
 
-@description('The memory size of the virtual machine in GB (max 224 GB)')
+@description('The memory size of the virtual machine in GiB (max 224 GiB)')
 param memorySizeGB int = 4
 
 @description('The list of network attachments to the virtual machine.')

articles/operator-nexus/quickstarts-virtual-machine-deployment-cli.md

@@ -38,8 +38,8 @@ Before you run the commands, you need to set several variables to define the con
 | ADMIN_USERNAME             | The username for the virtual machine administrator.                                                   |
 | SSH_PUBLIC_KEY             | The SSH public key that is used for secure communication with the virtual machine.                    |
 | CPU_CORES                  | The number of CPU cores for the virtual machine (even number, max 46 vCPUs)                           |
-| MEMORY_SIZE                | The amount of memory (in GB, max 224 GB) for the virtual machine.                                     |
-| VM_DISK_SIZE               | The size (in GB) of the virtual machine disk.                                                         |
+| MEMORY_SIZE                | The amount of memory (in GiB, max 224 GiB) for the virtual machine.                                     |
+| VM_DISK_SIZE               | The size (in GiB) of the virtual machine disk.                                                         |
 | VM_IMAGE                   | The URL of the virtual machine image.                                                                 |
 | ACR_URL                    | The URL of the Azure Container Registry.                                                              |
 | ACR_USERNAME               | The username for the Azure Container Registry.                                                        |

articles/operator-nexus/quickstarts-virtual-machine-deployment-ps.md

@@ -38,8 +38,8 @@ Before you run the commands, you need to set several variables to define the con
 | ADMIN_USERNAME             | The username for the virtual machine administrator.                                                   |
 | SSH_PUBLIC_KEY             | The SSH public key that is used for secure communication with the virtual machine.                    |
 | CPU_CORES                  | The number of CPU cores for the virtual machine (even number, max 46 vCPUs)                                         |
-| MEMORY_SIZE                | The amount of memory (in GB, max 224 GB) for the virtual machine.                                                 |
-| VM_DISK_SIZE               | The size (in GB) of the virtual machine disk.                                                         |
+| MEMORY_SIZE                | The amount of memory (in GiB, max 224 GiB) for the virtual machine.                                                 |
+| VM_DISK_SIZE               | The size (in GiB) of the virtual machine disk.                                                         |
 | VM_IMAGE                   | The URL of the virtual machine image.                                                                 |
 | ACR_URL                    | The URL of the Azure Container Registry.                                                              |
 | ACR_USERNAME               | The username for the Azure Container Registry.                                                        |

articles/operator-nexus/reference-near-edge-compute.md

@@ -39,9 +39,9 @@ Azure Operator Nexus supports a range of geometries and configurations. This tab
 | Number of virtual CPUs (vCPUs) for tenant usage       | 96 vCPUs, with hyperthreading enabled per compute server. |
 | Number of vCPUs available for workloads | 2 to 48 vCPUs, with an even number of vCPUs only. No cross-NUMA (nonuniform memory access) virtual machines (VMs). |
 | CPU pinning                            | Default. |
-| RAM for running tenant workloads        | 448 GB (224 GB per NUMA).  |
-| Huge pages for tenant workloads        | All VMs are backed by 1-GB huge pages. |
-| Disk (ephemeral) per compute           | Up to 3.5 TB per compute host. |
+| RAM for running tenant workloads        | 448 GiB (224 GiB per NUMA).  |
+| Huge pages for tenant workloads        | All VMs are backed by 1GiB (1G) huge pages. |
+| Disk (ephemeral) per compute           | Up to 3.5 TB (3.18 TiB) per compute host. |
 | Data plane traffic path for workloads  | Single-root I/O virtualization (SR-IOV). |
 | Number of SR-IOV virtual functions (VFs)                   | Maximum of 32 virtual NICs (vNICs), with 30 VFs available for tenant workloads per NUMA. |
 | SR-IOV NIC support                     | Enabled on all 100G NIC ports on VMs, with VFs assigned out of Mellanox-supported VF link aggregation (VF LAG). The allocated VFs are from the same physical NIC and within the same NUMA boundary. NIC ports that provide VF LAG are connected to two different top-of-rack (ToR) switches for redundancy. <br><br>Support includes trunked VF receive-side scaling (RSS) with hardware queuing. Support also includes multiple queues on VMs. |

articles/operator-nexus/troubleshoot-hardware-validation-failure.md

@@ -47,7 +47,7 @@ Expanding `result_detail` for a given category shows detailed results.
 
 ### System info category
 
-* Memory/RAM Related Failure (memory_capacity_GB)
+* Memory/RAM Related Failure (memory_capacity_GB) (measured in GiB)
     * Memory specs are defined in the SKU. Memory below threshold value indicates missing or failed Dual In-Line Memory Module (DIMM). A failed DIMM would also be reflected in the `health_info` category. The following example shows a failed memory check.
 
     ```json