[CI/Build] Fix unused fake implementation testing (#55797)

Signed-off-by: Jee Jee Li <jeejeelee@inferact.ai>
This commit is contained in:
Jee Jee Li
2026-09-08 03:36:18 +00:00
committed by GitHub
parent a69402aaa8
commit 607a6e48c5
6 changed files with 0 additions and 66 deletions
@@ -152,14 +152,3 @@ class TestDynamicPerTokenScaledFp8QuantIntegration:
assert kernel_wrapper.op_name == "dynamic_per_token_scaled_fp8_quant"
assert kernel_wrapper._config_picker is not None
assert kernel_wrapper._mutates_args == ["result", "scale"]
def test_fake_impl_functionality(self):
skip_if_platform_unsupported("dynamic_per_token_scaled_fp8_quant")
from vllm.kernels.helion.register import get_registered_kernels
registered_kernels = get_registered_kernels()
kernel_wrapper = registered_kernels["dynamic_per_token_scaled_fp8_quant"]
fake_impl = kernel_wrapper._fake_impl
args = _generate_fake_input(16, 4096)
assert fake_impl(*args) is None
@@ -248,14 +248,3 @@ class TestFusedQkNormRopeIntegration:
assert kernel_wrapper.op_name == "fused_qk_norm_rope"
assert kernel_wrapper._config_picker is not None
assert kernel_wrapper._mutates_args == ["qkv"]
def test_fake_impl_functionality(self):
skip_if_platform_unsupported("fused_qk_norm_rope")
from vllm.kernels.helion.register import get_registered_kernels
registered_kernels = get_registered_kernels()
kernel_wrapper = registered_kernels["fused_qk_norm_rope"]
fake_impl = kernel_wrapper._fake_impl
args = _generate_fake_input(16, 4096, 128)
assert fake_impl(*args) is None
@@ -230,14 +230,3 @@ class TestPerTokenGroupFp8QuantIntegration:
assert kernel_wrapper.op_name == "per_token_group_fp8_quant"
assert kernel_wrapper._config_picker is not None
assert kernel_wrapper._mutates_args == ["output_q", "output_s"]
def test_fake_impl_functionality(self):
skip_if_platform_unsupported("per_token_group_fp8_quant")
from vllm.kernels.helion.register import get_registered_kernels
registered_kernels = get_registered_kernels()
kernel_wrapper = registered_kernels["per_token_group_fp8_quant"]
fake_impl = kernel_wrapper._fake_impl
args = _generate_fake_input(16, 4096, 128)
assert fake_impl(*args) is None
@@ -194,14 +194,3 @@ class TestRmsNormDynamicPerTokenQuantIntegration:
assert kernel_wrapper.op_name == "rms_norm_dynamic_per_token_quant"
assert kernel_wrapper._config_picker is not None
assert kernel_wrapper._mutates_args == ["result", "scale", "residual"]
def test_fake_impl_functionality(self):
skip_if_platform_unsupported("rms_norm_dynamic_per_token_quant")
from vllm.kernels.helion.register import get_registered_kernels
registered_kernels = get_registered_kernels()
kernel_wrapper = registered_kernels["rms_norm_dynamic_per_token_quant"]
fake_impl = kernel_wrapper._fake_impl
args = _generate_fake_input(16, 4096)
assert fake_impl(*args) is None
@@ -298,14 +298,3 @@ class TestRmsNormPerBlockQuantIntegration:
assert kernel_wrapper.op_name == "rms_norm_per_block_quant"
assert kernel_wrapper._config_picker is not None
assert kernel_wrapper._mutates_args == ["result", "scale", "residual"]
def test_fake_impl_functionality(self):
skip_if_platform_unsupported("rms_norm_per_block_quant")
from vllm.kernels.helion.register import get_registered_kernels
registered_kernels = get_registered_kernels()
kernel_wrapper = registered_kernels["rms_norm_per_block_quant"]
fake_impl = kernel_wrapper._fake_impl
args = _generate_fake_input(16, 4096, 128)
assert fake_impl(*args) is None
@@ -211,14 +211,3 @@ class TestSiluAndMulPerBlockQuantIntegration:
assert kernel_wrapper.op_name == "silu_and_mul_per_block_quant"
assert kernel_wrapper._config_picker is not None
assert kernel_wrapper._mutates_args == ["out", "scales"]
def test_fake_impl_functionality(self):
skip_if_platform_unsupported("silu_and_mul_per_block_quant")
from vllm.kernels.helion.register import get_registered_kernels
registered_kernels = get_registered_kernels()
kernel_wrapper = registered_kernels["silu_and_mul_per_block_quant"]
fake_impl = kernel_wrapper._fake_impl
args = _generate_fake_input(16, 4096, 128)
assert fake_impl(*args) is None