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

[vLLM IR] 1/N Implement IR skeleton and rms_norm op (#33825)

Browse Source 
Signed-off-by: Luka Govedič <lgovedic@redhat.com>
Signed-off-by: Xinyu Chen <xinyu1.chen@intel.com>
Signed-off-by: chzhang <chaojun.zhang@intel.com>
Signed-off-by: Luka Govedic <luka.govedic@gmail.com>
Co-authored-by: Xinyu Chen <xinyu1.chen@intel.com>
Co-authored-by: Chaojun Zhang <chaojun.zhang@intel.com>
Co-authored-by: Luka Govedič <ProExpertProg@h100-01.nemg-001.lab.rdu2.dc.redhat.com>
This commit is contained in:
Luka Govedič
2026-03-31 22:15:05 -04:00
committed by GitHub
parent 0fab52f0aa
commit 40bb175027
49 changed files with 2177 additions and 265 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
tests/engine/test_arg_utils.py
View File

@@ -523,3 +523,20 @@ def test_human_readable_model_len():
for invalid in ["1a", "pwd", "10.24", "1.23M", "1.22T"]:
with pytest.raises(ArgumentError):
parser.parse_args(["--max-model-len", invalid])
def test_ir_op_priority():
from vllm.config.kernel import IrOpPriorityConfig, KernelConfig
ir_op_priority = IrOpPriorityConfig(rms_norm=["vllm_c"])
cfg1 = EngineArgs(ir_op_priority=ir_op_priority).create_engine_config()
cfg2 = EngineArgs(
kernel_config=KernelConfig(ir_op_priority=ir_op_priority)
).create_engine_config()
assert cfg1.kernel_config.ir_op_priority == cfg2.kernel_config.ir_op_priority
with pytest.raises(ValueError, match="rms_norm"):
_ = EngineArgs(
ir_op_priority=ir_op_priority,
kernel_config=KernelConfig(ir_op_priority=ir_op_priority),
).create_engine_config()