[Hardware] Replace torch.cuda.device_count/current_device/set_device API (#36145)

Signed-off-by: Kunshang Ji <jikunshang95@gmail.com>
Signed-off-by: Kunshang Ji <kunshang.ji@intel.com>
This commit is contained in:
Kunshang Ji
2026-03-12 07:57:47 -07:00
committed by GitHub
parent 2e693f48e7
commit 53ec16a705
89 changed files with 254 additions and 219 deletions
+1 -1
View File
@@ -15,7 +15,7 @@ llm = LLM(model="ibm-granite/granite-3.1-8b-instruct", tensor_parallel_size=2)
```
!!! warning
To ensure that vLLM initializes CUDA correctly, you should avoid calling related functions (e.g. [torch.cuda.set_device][])
To ensure that vLLM initializes CUDA correctly, you should avoid calling related functions (e.g. [torch.accelerator.set_device_index][])
before initializing vLLM. Otherwise, you may run into an error like `RuntimeError: Cannot re-initialize CUDA in forked subprocess`.
To control which devices are used, please instead set the `CUDA_VISIBLE_DEVICES` environment variable.
+3 -3
View File
@@ -91,8 +91,8 @@ If GPU/CPU communication cannot be established, you can use the following Python
import torch
import torch.distributed as dist
dist.init_process_group(backend="nccl")
local_rank = dist.get_rank() % torch.cuda.device_count()
torch.cuda.set_device(local_rank)
local_rank = dist.get_rank() % torch.accelerator.device_count()
torch.accelerator.set_device_index(local_rank)
data = torch.FloatTensor([1,] * 128).to("cuda")
dist.all_reduce(data, op=dist.ReduceOp.SUM)
torch.accelerator.synchronize()
@@ -337,7 +337,7 @@ import vllm
import torch
print(f"CUDA available: {torch.cuda.is_available()}")
print(f"CUDA device count: {torch.cuda.device_count()}")
print(f"CUDA device count: {torch.accelerator.device_count()}")
EOF
```