LogoTensorFusion Docs
LogoTensorFusion Docs
HomepageDocumentation

Getting Started

OverviewKubernetes InstallVM/Server Install(K3S)Helm On-premises InstallHost/GuestVM InstallTensorFusion Architecture

Application Operations

Create WorkloadConfigure AutoScalingMigrate Existing WorkloadBest Practices

Customize AI Infra

Production-Grade DeploymentConfig QoS and BillingBring Your Own CloudManaging License

Maintenance & Optimization

Upgrade ComponentsSetup AlertsGPU Live MigrationPreload ModelOptimize GPU Efficiency

Troubleshooting

HandbookTracing/ProfilingQuery Metrics & Logs

Reference

Comparison

Compare with NVIDIA vGPUCompare with MIG/MPSCompare with Run.AICompare with HAMi

GPU/Driver/OS Support Matrix

TensorFusion supports all series of NVIDIA GPUs from Volta architecture and above, with NVIDIA driver versions starting from 530.x, and CUDA versions range from 12.1 to the latest.

GPU VendorGPU ArchitectureGPU SeriesDriver Version
NVIDIAVoltaV100, Tesla V100530.x - Latest
TuringRTX 20 Series, Quadro RTX Series530.x - Latest
AmpereA100, A40, A10, A10G, A800, RTX 30 Series, RTX A Series530.x - Latest
HopperH100, H200, H800, H20530.x - Latest
Ada LovelaceL40, L40S, L4, RTX 40 Series, RTX Ada Series530.x - Latest
BlackwellB100, B40, RTX 50 Series530.x - Latest
AMDRDNARX 6000 Series, RX 7000 SeriesPlanned
CDNAMI200 Series, MI300 SeriesPlanned

Operation System Support

TensorFusion ControlPlane and Worker supports all major Linux distributions, including:

  • Ubuntu
  • Red Hat
  • SUSE
  • Amazon Linux
  • CentOS
  • ...

For GPU Client side, you can run in following OS:

  • Linux Kernel 4.x and above
  • Windows 10 and above (64-bit)
  • MacOS (WIP)

CPU Architecture Support

  • TensorFusion Worker & ControlPlane supports x86_64 and ARM64, x86_64 is battle-tested and ARM64 is currently under testing
  • TensorFusion GPU Client side supports x86_64 and ARM64

Table of Contents

Operation System Support
CPU Architecture Support