1. There are two parts for Accelerated Networking at different place

One is https://docs.openshift.com/container-platform/4.16/machine_management/creating_machinesets/creating-machineset-azure.html#machineset-azure-accelerated-networking_creating-machineset-azure 
Accelerated Networking for Microsoft Azure VMs
which is in the middle part

the other one is https://docs.openshift.com/container-platform/4.16/machine_management/creating_machinesets/creating-machineset-azure.html#machineset-azure-enabling-accelerated-networking-existing_creating-machineset-azure
Enabling Accelerated Networking on an existing Microsoft Azure cluster
which is in the end. There are many other contents between them, seems the Accelerated Networking contents should be combined together.

2.In doc https://docs.openshift.com/container-platform/4.16/machine_management/creating_machinesets/creating-machineset-azure.html#nvidia-gpu-aws-adding-a-gpu-node_creating-machineset-azure 

There are 13 steps for Procedure, in "Example machineset-azure.yaml file" of step3, shows
  resourceID: /resourceGroups/myclustername-rg/providers/Microsoft.Compute/galleries/gallery_myclustername_n6n4r/images/myclustername-gen2/versions/latest

_n6n4r should be removed to keep consistent with other place.

Because I checked on a 4.16 azure cluster, it looks like 

  resourceID: /resourceGroups/huliu-az89a-j8r4z-rg/providers/Microsoft.Compute/galleries/gallery_huliu_az89a_j8r4z/images/huliu-az89a-j8r4z-gen2/versions/latest

Same in "Example machineset-azure-gpu.yaml file" of step5,

  resourceID: /resourceGroups/myclustername-rg/providers/Microsoft.Compute/galleries/gallery_myclustername_n6n4r/images/myclustername-gen2/versions/latest

_n6n4r should be removed

And same in "Example output" of step8,

clustername-n6n4r-nc4ast4-gpu-worker-centralus1    1         1         1       1           122m
clustername-n6n4r-worker-centralus1                1         1         1       1           8h
clustername-n6n4r-worker-centralus2                1         1         1       1           8h
clustername-n6n4r-worker-centralus3                1         1         1       1           8h

_n6n4r should be removed

3. In "Example output" of step9,

NAME                                                PHASE     TYPE                   REGION      ZONE   AGE
myclustername-master-0                              Running   Standard_D8s_v3        centralus   2      6h40m
myclustername-master-1                              Running   Standard_D8s_v3        centralus   1      6h40m
myclustername-master-2                              Running   Standard_D8s_v3        centralus   3      6h40m
myclustername-nc4ast4-gpu-worker-centralus1-w9bqn   Running      centralus   1      21m
myclustername-worker-centralus1-rbh6b               Running   Standard_D4s_v3        centralus   1      6h38m
myclustername-worker-centralus2-dbz7w               Running   Standard_D4s_v3        centralus   2      6h38m
myclustername-worker-centralus3-p9b8c               Running   Standard_D4s_v3        centralus   3      6h38m

Missing Standard_NC4as_T4_v3 for the gpu machine, it looks like
huliu-az89a-j8r4z-gpu-worker-eastus1-c4ph2   Running   Standard_NC4as_T4_v3   eastus   1      4m45s

4. Step11 and step1 are repeated, Step12 and step7 are repeated, Step13 and step8 are repeated.

5.There is an additional resources of accelerated networking under deploying NFD Operator part. Seems not right.
https://docs.openshift.com/container-platform/4.16/machine_management/creating_machinesets/creating-machineset-azure.html#nvidia-gpu-aws-deploying-the-node-feature-discovery-operator_creating-machineset-azure
Deploying the Node Feature Discovery Operator

 https://drive.google.com/file/d/1IsBpRGrduGHx7U961L_orDCOvkT0Af9y/view?usp=sharing 

    4.16

    1.
    2.
    3.