biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Bugfix] Fix some narrowing conversion warnings (#20141)

Browse Source 
Signed-off-by: Tyler Michael Smith <tyler@neuralmagic.com>
This commit is contained in:
Tyler Michael Smith
2025-06-27 12:01:28 -04:00
committed by GitHub
parent c6c983053d
commit e8c3bd2cd1
6 changed files with 8 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
csrc/mamba/causal_conv1d/causal_conv1d.cu
View File

@@ -185,9 +185,7 @@ void causal_conv1d_fwd(const at::Tensor &x, const at::Tensor &weight,
params.conv_states_ptr = nullptr;
}
// Otherwise the kernel will be launched from cuda:0 device
// Cast to char to avoid compiler warning about narrowing
at::cuda::CUDAGuard device_guard{(char)x.get_device()};
const at::cuda::OptionalCUDAGuard device_guard(device_of(x));
auto stream = at::cuda::getCurrentCUDAStream().stream();
DISPATCH_WTYPE_ITYPE_FLOAT_AND_HALF_AND_BF16(x.scalar_type(), "causal_conv1d_fwd", [&] {
causal_conv1d_fwd_cuda<input_t, weight_t>(params, stream);
@@ -278,9 +276,7 @@ void causal_conv1d_update(const at::Tensor &x,
params.conv_state_indices_ptr = nullptr;
}
// Otherwise the kernel will be launched from cuda:0 device
// Cast to char to avoid compiler warning about narrowing
at::cuda::CUDAGuard device_guard{(char)x.get_device()};
const at::cuda::OptionalCUDAGuard device_guard(device_of(x));
auto stream = at::cuda::getCurrentCUDAStream().stream();
DISPATCH_WTYPE_ITYPE_FLOAT_AND_HALF_AND_BF16(x.scalar_type(), "causal_conv1d_update", [&] {
causal_conv1d_update_cuda<input_t, weight_t>(params, stream);