[npu] add npu support for gemini and zero (#5067)

* [npu] setup device utils (#5047)

* [npu] add npu device support

* [npu] support low level zero

* [test] update npu zero plugin test

* [hotfix] fix import

* [test] recover tests

* [npu] gemini support npu (#5052)

* [npu] refactor device utils

* [gemini] support npu

* [example] llama2+gemini support npu

* [kernel] add arm cpu adam kernel (#5065)

* [kernel] add arm cpu adam

* [optim] update adam optimizer

* [kernel] arm cpu adam remove bf16 support
This commit is contained in:
Hongxin Liu
2023-11-20 16:12:41 +08:00
committed by GitHub
parent 8d56c9c389
commit e5ce4c8ea6
46 changed files with 994 additions and 233 deletions

View File

@@ -206,7 +206,10 @@ class ChunkManager:
tensor (torch.Tensor): An extern static tensor. E.g. optimizer state.
"""
assert tensor not in self.tensor_chunk_map
self.total_mem[tensor.device.type] += tensor.numel() * tensor.element_size()
device_type = tensor.device.type
if device_type == "npu":
device_type = "cuda"
self.total_mem[device_type] += tensor.numel() * tensor.element_size()
def __repr__(self) -> str:
msg = [