whn09 ed49792a0c Add AWS Trainium (Neuron) platform, basic mode — part 1
Run DeepInfer on the native torch-neuronx device (PrivateUse1 `neuron`,
torch-neuronx 2.14, no torch-xla), following the Ascend port's structure.

- deep_cluster: Neuron discovery (one DeepInfer device = one logical
  NeuronCore, NEURON_RT_VISIBLE_CORES, NUMA from neuron-ls).
- PlatformNeuron: takes the Ascend (is_npu) kernel layouts, adds is_neuron;
  DEEPINFER_NEURON_DEVICE=cpu runs the same code on the host for
  correctness work.
- neuron_ops/transfer_to_neuron: map "cuda" devices, torch.device and
  torch.cuda.* onto neuron:0 (counterpart of torch_npu transfer_to_npu).
- C++ extension: -DWITH_NEURON host-only build; torch-allocated buffers and
  tensor-view slot copies instead of raw device-pointer copies.
- Pure-torch drop-in vendor packages (Ascend API semantics): deep_gemm,
  flash_mla, deep_select, tile_kernels, tilelang (constants only).
- tilelang_ops/neuron backend: KV cache / metadata / norm / compress kernels;
  CP, ZOS and VL entry points raise NotImplementedError.
- Basic mode defaults on Neuron: eager, no DSpark/MTP drafting, no
  zero-overhead scheduling; MoE avoids the GPU-only mega kernel; mHC
  prenorm GEMM uses the Ascend single-split reduction.

Not yet included: deep_ep, the sampling kernels, end-to-end harness.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-20 18:34:51 +08:00

25 lines
596 B
Python

"""Mirrors ``tile_kernels.rand.seed`` (DeepSeek TileKernels 2.1.0), verbatim."""
# Copyright (c) 2026 DeepSeek, MIT License (see LICENSE).
import torch
_seed = torch.initial_seed()
def set_seed(seed: int) -> None:
global _seed
torch.manual_seed(seed)
_seed = seed
def reserve_seed(num_blocks: int) -> int:
global _seed
seed = _seed
_seed += num_blocks
seed = (seed + 0x9E3779B9) & 0xFFFFFFFF
seed = ((seed ^ (seed >> 16)) * 0x21F0AAAD) & 0xFFFFFFFF
seed = ((seed ^ (seed >> 15)) * 0x735A2D97) & 0xFFFFFFFF
return (seed ^ (seed >> 15)) & 0xFFFFFFFF