# 通用GPU Pod模板配置文件 # 包含多种GPU工作负载的Pod模板 --- # 基础GPU测试Pod apiVersion: v1 kind: Pod metadata: name: gpu-test-basic labels: app: gpu-test type: basic spec: containers: - name: gpu-test image: nvidia/cuda:12.4-runtime-ubuntu22.04 command: ["nvidia-smi"] resources: limits: nvidia.com/gpu: "1" memory: "2Gi" cpu: "1" requests: memory: "1Gi" cpu: "500m" tolerations: - key: nvidia.com/gpu operator: Exists effect: NoSchedule restartPolicy: Never --- # TensorFlow GPU训练Pod apiVersion: v1 kind: Pod metadata: name: tensorflow-gpu-training labels: app: tensorflow type: training spec: containers: - name: tensorflow-gpu image: tensorflow/tensorflow:2.13.0-gpu command: ["python", "-c"] args: - | import tensorflow as tf import numpy as np print("TensorFlow version:", tf.__version__) print("GPU devices:", tf.config.list_physical_devices('GPU')) # 简单的神经网络训练示例 with tf.device('/GPU:0'): # 生成训练数据 X = tf.random.normal([1000, 20]) y = tf.random.normal([1000, 1]) # 创建模型 model = tf.keras.Sequential([ tf.keras.layers.Dense(64, activation='relu', input_shape=(20,)), tf.keras.layers.Dense(32, activation='relu'), tf.keras.layers.Dense(1) ]) model.compile(optimizer='adam', loss='mse') # 训练模型 print("开始训练...") history = model.fit(X, y, epochs=10, batch_size=32, verbose=1) print("训练完成！") resources: limits: nvidia.com/gpu: "1" memory: "4Gi" cpu: "2" requests: memory: "2Gi" cpu: "1" env: - name: CUDA_VISIBLE_DEVICES value: "0" tolerations: - key: nvidia.com/gpu operator: Exists effect: NoSchedule restartPolicy: Never --- # PyTorch GPU训练Pod apiVersion: v1 kind: Pod metadata: name: pytorch-gpu-training labels: app: pytorch type: training spec: containers: - name: pytorch-gpu image: pytorch/pytorch:2.1.0-cuda12.1-cudnn8-runtime command: ["python", "-c"] args: - | import torch import torch.nn as nn import torch.optim as optim import numpy as np print(f"PyTorch version: {torch.__version__}") print(f"CUDA available: {torch.cuda.is_available()}") print(f"CUDA device count: {torch.cuda.device_count()}") if torch.cuda.is_available(): device = torch.device('cuda:0') print(f"Using device: {device}") print(f"Device name: {torch.cuda.get_device_name(0)}") # 简单的神经网络 class SimpleNet(nn.Module): def __init__(self): super(SimpleNet, self).__init__() self.fc1 = nn.Linear(20, 64) self.fc2 = nn.Linear(64, 32) self.fc3 = nn.Linear(32, 1) self.relu = nn.ReLU() def forward(self, x): x = self.relu(self.fc1(x)) x = self.relu(self.fc2(x)) x = self.fc3(x) return x # 创建模型和数据 model = SimpleNet().to(device) X = torch.randn(1000, 20).to(device) y = torch.randn(1000, 1).to(device) criterion = nn.MSELoss() optimizer = optim.Adam(model.parameters(), lr=0.001) # 训练循环 print("开始训练...") for epoch in range(10): optimizer.zero_grad() outputs = model(X) loss = criterion(outputs, y) loss.backward() optimizer.step() print(f"Epoch {epoch+1}/10, Loss: {loss.item():.4f}") print("训练完成！") else: print("CUDA不可用") resources: limits: nvidia.com/gpu: "1" memory: "4Gi" cpu: "2" requests: memory: "2Gi" cpu: "1" env: - name: CUDA_VISIBLE_DEVICES value: "0" tolerations: - key: nvidia.com/gpu operator: Exists effect: NoSchedule restartPolicy: Never --- # 多GPU并行训练Pod apiVersion: v1 kind: Pod metadata: name: multi-gpu-training labels: app: multi-gpu type: distributed-training spec: containers: - name: multi-gpu-trainer image: pytorch/pytorch:2.1.0-cuda12.1-cudnn8-runtime command: ["python", "-c"] args: - | import torch import torch.nn as nn import torch.distributed as dist import torch.multiprocessing as mp from torch.nn.parallel import DistributedDataParallel as DDP print(f"PyTorch version: {torch.__version__}") print(f"CUDA available: {torch.cuda.is_available()}") print(f"CUDA device count: {torch.cuda.device_count()}") if torch.cuda.device_count() > 1: print(f"使用 {torch.cuda.device_count()} 个GPU进行训练") # 简单的数据并行示例 class SimpleNet(nn.Module): def __init__(self): super(SimpleNet, self).__init__() self.fc1 = nn.Linear(20, 64) self.fc2 = nn.Linear(64, 32) self.fc3 = nn.Linear(32, 1) self.relu = nn.ReLU() def forward(self, x): x = self.relu(self.fc1(x)) x = self.relu(self.fc2(x)) x = self.fc3(x) return x model = SimpleNet() if torch.cuda.device_count() > 1: model = nn.DataParallel(model) model = model.cuda() # 创建数据 X = torch.randn(1000, 20).cuda() y = torch.randn(1000, 1).cuda() criterion = nn.MSELoss() optimizer = torch.optim.Adam(model.parameters(), lr=0.001) # 训练 print("开始多GPU训练...") for epoch in range(5): optimizer.zero_grad() outputs = model(X) loss = criterion(outputs, y) loss.backward() optimizer.step() print(f"Epoch {epoch+1}/5, Loss: {loss.item():.4f}") print("多GPU训练完成！") else: print("只有一个GPU，使用单GPU训练") resources: limits: nvidia.com/gpu: "2" memory: "8Gi" cpu: "4" requests: memory: "4Gi" cpu: "2" env: - name: CUDA_VISIBLE_DEVICES value: "0,1" tolerations: - key: nvidia.com/gpu operator: Exists effect: NoSchedule restartPolicy: Never --- # GPU推理服务Pod apiVersion: v1 kind: Pod metadata: name: gpu-inference-service labels: app: inference type: service spec: containers: - name: inference-server image: nvidia/tritonserver:23.10-py3 command: ["tritonserver"] args: - "--model-repository=/models" - "--strict-model-config=false" - "--log-verbose=1" ports: - containerPort: 8000 name: http - containerPort: 8001 name: grpc - containerPort: 8002 name: metrics resources: limits: nvidia.com/gpu: "1" memory: "4Gi" cpu: "2" requests: memory: "2Gi" cpu: "1" env: - name: CUDA_VISIBLE_DEVICES value: "0" volumeMounts: - name: model-repository mountPath: /models readinessProbe: httpGet: path: /v2/health/ready port: 8000 initialDelaySeconds: 30 periodSeconds: 10 livenessProbe: httpGet: path: /v2/health/live port: 8000 initialDelaySeconds: 60 periodSeconds: 30 volumes: - name: model-repository emptyDir: {} tolerations: - key: nvidia.com/gpu operator: Exists effect: NoSchedule restartPolicy: Always