### What this PR does / why we need it?
**Scope of Changes**:
| File Path |
| :--- |
| `vllm_ascend/ops/triton/activation/swiglu_quant.py` |
| `vllm_ascend/ops/triton/batch_invariant/matmul.py` |
| `vllm_ascend/ops/triton/batch_invariant/mean.py` |
| `vllm_ascend/ops/triton/batch_invariant/rmsnorm.py` |
| `vllm_ascend/ops/triton/fla/chunk.py` |
| `vllm_ascend/ops/triton/fla/chunk_delta_h.py` |
| `vllm_ascend/ops/triton/fla/chunk_o.py` |
| `vllm_ascend/ops/triton/fla/chunk_scaled_dot_kkt.py` |
| `vllm_ascend/ops/triton/fla/cumsum.py` |
| `vllm_ascend/ops/triton/fla/fused_qkvzba_split_reshape.py` |
| `vllm_ascend/ops/triton/fla/l2norm.py` |
| `vllm_ascend/ops/triton/fla/layernorm_guard.py` |
| `vllm_ascend/ops/triton/fla/sigmoid_gating.py` |
| `vllm_ascend/ops/triton/fla/solve_tril.py` |
| `vllm_ascend/ops/triton/fla/utils.py` |
| `vllm_ascend/ops/triton/fla/wy_fast.py` |
| `vllm_ascend/ops/triton/fused_gdn_gating.py` |
| `vllm_ascend/ops/triton/layernorm_gated.py` |
| `vllm_ascend/ops/triton/linearnorm/split_qkv_rmsnorm_rope.py` |
| `vllm_ascend/ops/triton/mamba/causal_conv1d.py` |
| `vllm_ascend/ops/triton/reject_sample.py` |
| `vllm_ascend/ops/triton/rope.py` |
| `vllm_ascend/ops/triton/spec_decode/utils.py` |
| `vllm_ascend/ops/triton/triton_utils.py` |
### Does this PR introduce _any_ user-facing change?
### How was this patch tested?
- vLLM version: v0.14.0
- vLLM main:
d68209402d
Signed-off-by: MrZ20 <2609716663@qq.com>
This commit is contained in:
@@ -100,8 +100,8 @@ def rms_norm(
|
||||
"""
|
||||
assert weight.dim() == 1, "Weight must be 1-dimensional"
|
||||
assert input_.shape[-1] == weight.shape[0], (
|
||||
f"Input last dimension ({input_.shape[-1]}) must match "
|
||||
f"weight dimension ({weight.shape[0]})")
|
||||
f"Input last dimension ({input_.shape[-1]}) must match weight dimension ({weight.shape[0]})"
|
||||
)
|
||||
|
||||
# Flatten all dimensions except the last one
|
||||
original_shape = input_.shape
|
||||
@@ -113,10 +113,9 @@ def rms_norm(
|
||||
|
||||
output = torch.empty_like(input_2d, dtype=input_.dtype)
|
||||
BLOCK_SIZE = 1024
|
||||
max_grid_size = driver.active.utils.get_device_properties(
|
||||
torch.npu.current_device())["num_vectorcore"]
|
||||
max_grid_size = driver.active.utils.get_device_properties(torch.npu.current_device())["num_vectorcore"]
|
||||
|
||||
grid = (min(n_rows, max_grid_size), )
|
||||
grid = (min(n_rows, max_grid_size),)
|
||||
|
||||
_rms_norm_kernel[grid](
|
||||
input_2d,
|
||||
|
||||
Reference in New Issue
Block a user