MAF-19131: feat(preset): add new preset for DeepSeek-R1 max throughput

[bongwoobak]

https://docs.moreh.io/benchmarking/deepseek_r1_671b_on_amd_mi300x_gpus_maximum_throughput/ example is

apiVersion: odin.moreh.io/v1alpha1
kind: InferenceService
metadata:
  name: deepseek-r1-prefill
  namespace: mif
spec:
  replicas: 2
  inferencePoolRefs:
    - name: heimdall
  templateRefs:
    - name: vllm-prefill-dp
    - name: vllm-deepseek-r1-prefill-mi300x-dp8ep-maxthroughput
  workerTemplate:
    spec:
      containers:
        - name: main
          env:
            - name: HF_TOKEN
              value: "<huggingfaceToken>"
          resources:
            limits:
              mellanox/hca: "1"
            requests:
              mellanox/hca: "1"
---
apiVersion: odin.moreh.io/v1alpha1
kind: InferenceService
metadata:
  name: deepseek-r1-decode
  namespace: mif
spec:
  replicas: 3
  inferencePoolRefs:
    - name: heimdall
  templateRefs:
    - name: vllm-decode-dp
    - name: vllm-deepseek-r1-decode-mi300x-dp8ep-maxthroughput
  workerTemplate:
    spec:
      containers:
        - name: main
          env:
            - name: HF_TOKEN
              value: "<huggingfaceToken>"
          resources:
            limits:
              mellanox/hca: "1"
            requests:
              mellanox/hca: "1"

[hhk7734]

quickstart 로 이동시켜주세요

[hhk7734]

          resources:
            requests:
              amd.com/gpu: "8"
            limits:
              amd.com/gpu: "8"
      nodeSelector:
        moai.moreh.io/accelerator.vendor: amd
        moai.moreh.io/accelerator.model: mi300x
      tolerations:
        - key: amd.com/gpu
          operator: Exists
          effect: NoSchedule

추가해주세요