KTransformers

CPU/GPU Requirements

KTransformers serving depends on both GPU memory and CPU expert throughput. A usable setup depends on the GPU, CPU ISA, memory capacity, NUMA layout, model checkpoint, and KT method.

Baseline Requirements

ComponentGuidance
OSLinux x86-64 for current public packages.
PythonPython 3.10, 3.11, or 3.12 for kt-kernel wheels.
GPUNVIDIA Ampere or newer is the current main path for serving.
CPUAVX2 minimum for compatibility paths; AVX512 or AMX for higher-throughput paths.
MemoryLarge MoE models need high system RAM; size depends on method and CPU weight format.
NUMAMulti-socket systems should tune --kt-threadpool-count and CPU placement.

Current Hardware Scope

PlatformCurrent status
NVIDIA GPU + x86 CPUMain documented path for NVIDIA/AMX workflows.
AMD CPU pathSupported direction; no public model validation combination yet.
Ascend NPUNot current public support.
Intel xPUNot current public support.
ROCmHistorical documentation only until a current package path is validated.

Planning Approach

Start from the model support combination:

model + method + CPU backend + GPU count + system RAM + package versions

Then tune:

  • --kt-cpuinfer
  • --kt-threadpool-count
  • --kt-num-gpu-experts
  • prefill threshold and deferred experts if applicable

Hardware Scope

Hardware support is based on concrete model and method smoke results. When you see a hardware class listed, still check that the model, method, CPU backend, GPU count, system memory, and package versions match your environment.