[Quant][Kernel] Remove GPTQ Group/Dynamic Activation Ordering (#54809)

Signed-off-by: Roderick-Wu <roderickwu2003@gmail.com>
Signed-off-by: Roderick Wu <roderickwu2003@gmail.com>
Signed-off-by: mgoin <mgoin64@gmail.com>
Co-authored-by: mgoin <mgoin64@gmail.com>
This commit is contained in:
Roderick Wu
2026-09-08 12:45:12 -07:00
committed by GitHub
co-authored by mgoin
parent 1a522b6949
commit 2c9d68f80b
99 changed files with 1182 additions and 3917 deletions
@@ -196,7 +196,6 @@ def test_dense_kernel_rejects_on_non_gfx1100():
act_type=torch.float16,
group_size=128,
zero_points=False,
has_g_idx=False,
)
ok, reason = RDNA3W4A16LinearKernel.can_implement(config)
assert ok is False, f"RDNA3 dense kernel accepted on non-gfx1100: {reason}"
@@ -431,7 +430,6 @@ class TestDenseKernelSelectionMocked:
act_type=torch.float16,
group_size=128,
zero_points=False,
has_g_idx=False,
)
ok, _ = RDNA3W4A16LinearKernel.can_implement(config)
assert ok is True
@@ -461,7 +459,6 @@ class TestDenseKernelSelectionMocked:
act_type=torch.float16,
group_size=128,
zero_points=False,
has_g_idx=False,
)
with (
patch("vllm.platforms.rocm.on_gfx1100", return_value=False),