Skip to main content

Watsonx ai installation

Setup GPU node on OpenShift​

Install "Node Feature Discovery"​

Go to OperatorHub

Search for "Node Feature Discovery"

Install

Create a NodeFeatureDisocvery CR​

Go to Installed Operators

Select "Node Feature Discovery"

Select the box "Provided APIs"

Select "Create Instance"

Review the values

Select "Create"

Install "Nvidia GPU Operator"​

Go to OperatorHub

Search for "Nvidia GPU Operator"

Install

Create "Cluster Policy"​

Go to "Installed Operators"

Click on "Nvidia GPU Operator"

Select "ClusterPolicy" tab

Click "Create ClusterPolicy"

Click "Create"

note

For additional Information:

OpenShift Docs on Node Feature Discovery

Nvidia docs on Node Feature Discovery

Nvidia docs on Nvidia GPU Operator

Install WatsonX.AI on Cloud Pak for Data​

Apply the WatsonX.AI OLM:​

To install the WatsonX.AI OLMs:

cpd-cli manage apply-olm \
--release=${VERSION} \
--cpd_operator_ns=${PROJECT_CPD_INST_OPERATORS} \
--components=watsonx_ai \
--case_download=true \
--from_oci=true

Apply the WatsonX.AI CR:​

To install WatsonX.AI run the following command:

cpd-cli manage apply-cr \
--components=watsonx_ai \
--release=${VERSION} \
--cpd_instance_ns=${PROJECT_CPD_INST_OPERANDS} \
--block_storage_class=${STG_CLASS_BLOCK} \
--file_storage_class=${STG_CLASS_FILE} \
--license_acceptance=true \
--case_download=true \
--from_oci=true

Install Additional Models:​

This installs the following models:

granite-13b-instruct-v2 llama-2-70b-chat mixtral-8x7b-instruct-v01-q

Run the following command:

oc patch watsonxaiifm watsonxaiifm-cr \
--namespace=${PROJECT_CPD_INST_OPERANDS} \
--type=merge \
--patch='{"spec":{"install_model_list": ["ibm-mistralai-mixtral-8x7b-instruct-v01-q","ibm-granite-13b-chat-v2","meta-llama-llama-2-70b-chat"]}}'