AMD Instinct™ MI355X for Training & Inference

White-glove onboarding and AMD Instinct expertise included.

• 288GB HBM3E per GPU
  
  
• 8 TB/s peak memory bandwidth
  
  
• Efficient inference datatypes (MXFP6 / MXFP4)
  
  
• Full support for PyTorch and TensorFlow