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

[Bugfix] Fix ConchLinearKernel channelwise quantization (group_size=-1) (#37329)

Browse Source 
Signed-off-by: Matthias Gehre <matthias.gehre@amd.com>
This commit is contained in:
Matthias Gehre
2026-03-20 16:32:21 +01:00
committed by GitHub
parent 56a62c310c
commit 5e806bcf54
1 changed files with 9 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
vllm/model_executor/kernels/linear/mixed_precision/conch.py
View File

@@ -124,6 +124,14 @@ class ConchLinearKernel(MPLinearKernel):
w_q, w_s, w_zp, _ = self._get_weight_params(layer) w_q, w_s, w_zp, _ = self._get_weight_params(layer)
# Map channelwise group_size=-1 to the actual input dimension K.
# The conch kernel computes stride_mul = block_k / group_size;
# passing -1 produces a negative stride that reads out-of-bounds
# scale values for all K-blocks after the first.
group_size = self.config.group_size
if group_size == -1:
group_size = x.shape[-1]
output = mixed_precision_gemm( output = mixed_precision_gemm(
x=x, x=x,
w_q_packed=w_q.data, w_q_packed=w_q.data,
@@ -131,7 +139,7 @@ class ConchLinearKernel(MPLinearKernel):
w_zp=w_zp.data if w_zp is not None else None, w_zp=w_zp.data if w_zp is not None else None,
weight_size_bits=self.config.weight_type.size_bits, weight_size_bits=self.config.weight_type.size_bits,
weight_bias=self.config.weight_type.bias, weight_bias=self.config.weight_type.bias,
group_size=self.config.group_size, group_size=group_size,
) )
if bias is not None: if bias is not None: