Files
xc-llm-ascend/.github/workflows/scripts/run_suite.py
Li Wang e35f304419 [CI] Auto partition for test cases (#6379)
### What this PR does / why we need it?
This patch add auto-partition feat for tests, for example, before this
pr, we are running e2e single card test for 2h40min, after the auto
partition, test case is automatically allocated into the required n
parts based on its test duration (greedy strategy) and run in parallel.
The advantage of doing this is that our overall test duration will
become 1/n of the original.

### Does this PR introduce _any_ user-facing change?
Before:
e2e single card test spend 2h40min
After:
e2e single card test spend 1h13min

### How was this patch tested?

```shell
python .github/workflows/scripts/run_suite.py --auto-partition-size 2 --auto-partition-id 0 
args=Namespace(timeout_per_file=2000, suite='e2e-singlecard', auto_partition_id=0, auto_partition_size=2, continue_on_error=False, enable_retry=False, max_attempts=2, retry_wait_seconds=60, retry_timeout_increase=600)
+----------------+--------------------+
| Suite          | Partition          |
|----------------+--------------------|
| e2e-singlecard | 1/2 (0-based id=0) |
+----------------+--------------------+
 Enabled 13 test(s) (est total 4020.0s):
  - tests/e2e/singlecard/spec_decode/test_v1_spec_decode.py (est_time=1800)
  - tests/e2e/singlecard/test_aclgraph_accuracy.py (est_time=480)
  - tests/e2e/singlecard/test_guided_decoding.py (est_time=354)
  - tests/e2e/singlecard/test_batch_invariant.py (est_time=320)
  - tests/e2e/singlecard/pooling/test_embedding.py (est_time=270)
  - tests/e2e/singlecard/test_quantization.py (est_time=200)
  - tests/e2e/singlecard/test_llama32_lora.py (est_time=162)
  - tests/e2e/singlecard/test_cpu_offloading.py (est_time=132)
  - tests/e2e/singlecard/pooling/test_classification.py (est_time=120)
  - tests/e2e/singlecard/test_camem.py (est_time=77)
  - tests/e2e/singlecard/compile/test_norm_quant_fusion.py (est_time=70)
  - tests/e2e/singlecard/test_auto_fit_max_mode_len.py (est_time=25)
  - tests/e2e/singlecard/test_profile_execute_duration.py (est_time=10)

(base) wangli@Mac-mini vllm-ascend % python .github/workflows/scripts/run_suite.py --auto-partition-size 2 --auto-partition-id 1 
args=Namespace(timeout_per_file=2000, suite='e2e-singlecard', auto_partition_id=1, auto_partition_size=2, continue_on_error=False, enable_retry=False, max_attempts=2, retry_wait_seconds=60, retry_timeout_increase=600)
+----------------+--------------------+
| Suite          | Partition          |
|----------------+--------------------|
| e2e-singlecard | 2/2 (0-based id=1) |
+----------------+--------------------+
 Enabled 13 test(s) (est total 4025.0s):
  - tests/e2e/singlecard/spec_decode/test_mtp_eagle_correctness.py (est_time=1500)
  - tests/e2e/singlecard/pooling/test_scoring.py (est_time=500)
  - tests/e2e/singlecard/test_aclgraph_batch_invariant.py (est_time=410)
  - tests/e2e/singlecard/test_vlm.py (est_time=354)
  - tests/e2e/singlecard/test_models.py (est_time=300)
  - tests/e2e/singlecard/test_multistream_overlap_shared_expert.py (est_time=200)
  - tests/e2e/singlecard/test_sampler.py (est_time=200)
  - tests/e2e/singlecard/test_async_scheduling.py (est_time=150)
  - tests/e2e/singlecard/test_aclgraph_mem.py (est_time=130)
  - tests/e2e/singlecard/test_ilama_lora.py (est_time=95)
  - tests/e2e/singlecard/test_completion_with_prompt_embeds.py (est_time=76)
  - tests/e2e/singlecard/test_qwen3_multi_loras.py (est_time=65)
  - tests/e2e/singlecard/test_xlite.py (est_time=45)
```
- vLLM version: v0.14.1
- vLLM main:
dc917cceb8

---------

Signed-off-by: wangli <wangli858794774@gmail.com>
2026-01-29 20:28:10 +08:00

245 lines
9.0 KiB
Python

import argparse
import os
from pathlib import Path
import tabulate
import yaml
from ci_utils import TestFile, run_e2e_files
def load_suites_from_config(config_path: str = "config.yaml") -> dict[str, list[TestFile]]:
# Get absolute path relative to this script
script_dir = Path(__file__).parent
abs_config_path = script_dir / config_path
with open(abs_config_path) as f:
suites_data = yaml.safe_load(f)
suites = {}
for suite_name, test_files in suites_data.items():
suites[suite_name] = []
for file_data in test_files:
name = file_data.get("name")
estimated_time = file_data.get("estimated_time", 60)
is_skipped = file_data.get("is_skipped", False)
suites[suite_name].append(TestFile(name, estimated_time, is_skipped))
return suites
suites = load_suites_from_config()
def auto_partition(files, rank, size):
"""
Partition files into size sublists with approximately equal sums of estimated times
using stable sorting, and return the partition for the specified rank.
Args:
files (list): List of file objects with estimated_time attribute
rank (int): Index of the partition to return (0 to size-1)
size (int): Number of partitions
Returns:
list: List of file objects in the specified rank's partition
"""
# Filter out skipped files
files = [f for f in files if not f.is_skipped]
weights = [f.estimated_time for f in files]
if not weights or size <= 0 or size > len(weights):
return []
# Create list of (weight, original_index) tuples
# Using negative index as secondary key to maintain original order for equal weights
indexed_weights = [(w, -i) for i, w in enumerate(weights)]
# Stable sort in descending order by weight
# If weights are equal, larger (negative) index comes first (i.e., earlier original position)
indexed_weights = sorted(indexed_weights, reverse=True)
# Extract original indices (negate back to positive)
indexed_weights = [(w, -i) for w, i in indexed_weights]
# Initialize partitions and their sums
partitions = [[] for _ in range(size)]
sums = [0.0] * size
# Greedy approach: assign each weight to partition with smallest current sum
for weight, idx in indexed_weights:
# Find partition with minimum sum
min_sum_idx = sums.index(min(sums))
partitions[min_sum_idx].append(idx)
sums[min_sum_idx] += weight
# Return the files corresponding to the indices in the specified rank's partition
indices = partitions[rank]
return [files[i] for i in indices]
def _get_disk_covered_dirs(all_suite_files: set[str], project_root: Path | str) -> list[str]:
covered_dirs = set()
for file_path in all_suite_files:
# e.g. tests/e2e/singlecard/test_foo.py -> tests/e2e/singlecard
parent_dir = (project_root / file_path).parent if os.path.isfile(file_path) else (project_root / file_path)
if parent_dir.exists():
# Store relative path to project root
try:
rel_dir = parent_dir.relative_to(project_root)
# Check if this directory is already covered by a parent directory
is_covered = False
for existing_dir in list(covered_dirs):
# If existing_dir is a parent of rel_dir, rel_dir is already covered
if existing_dir in rel_dir.parents or existing_dir == rel_dir:
is_covered = True
break
# If rel_dir is a parent of existing_dir, replace existing_dir with rel_dir
elif rel_dir in existing_dir.parents:
covered_dirs.remove(existing_dir)
# We continue checking other existing_dirs, but we know rel_dir should be added
# unless another parent covers it (which is handled by the first if block logic effectively
# but we need to be careful with modification during iteration, so we use list copy)
if not is_covered:
covered_dirs.add(rel_dir)
except ValueError:
pass
return covered_dirs
def _sanity_check_suites(suites: dict[str, list[TestFile]]):
"""
Check if all test files defined in the suites exist on disk.
"""
# 1. Collect all test files defined in all suites
all_suite_files = set()
for suite in suites.values():
for test_file in suite:
# Handle ::test_case syntax
file_path = test_file.name.split("::")[0]
all_suite_files.add(file_path)
# 2. Identify all directories covered by the suites
project_root = Path.cwd()
if not (project_root / "tests").exists():
script_dir = Path(__file__).parent
# .github/workflows/scripts -> ../../../ -> root
project_root = script_dir.parents[2]
# For now, we only check dirs under [tests/e2e/singlecard, tests/e2e/multicard]
covered_dirs = _get_disk_covered_dirs(all_suite_files, project_root)
# 3. Scan disk for all test_*.py files in these directories
all_disk_files = set()
for dir_path in covered_dirs:
full_dir_path = project_root / dir_path
# rglob is equivalent to glob('**/' + pattern)
for py_file in full_dir_path.rglob("test_*.py"):
try:
rel_path = py_file.relative_to(project_root)
all_disk_files.add(str(rel_path))
except ValueError:
pass
# 4. Find files on disk but missing from ANY suite
# We check if a disk file is present in 'all_suite_files' (union of all suites)
missing_files = sorted(list(all_disk_files - all_suite_files))
missing_text = "\n".join(f'TestFile("{x}"),' for x in missing_files)
if missing_files:
assert len(missing_files) == 0, (
f"Some test files found on disk in covered directories are not in ANY test suite.\n"
f"Scanned directories: {sorted([str(d) for d in covered_dirs])}\n"
f"Missing files:\n"
f"{missing_text}\n"
f"If this is intentional, please label them as 'is_skipped=True' and add them to the test suite."
)
# 5. check if all files in suites exist on disk
non_existent_files = sorted(list(all_suite_files - all_disk_files))
non_existent_text = "\n".join(f'TestFile("{x}"),' for x in non_existent_files)
assert len(non_existent_files) == 0, (
f"Some test files in test suite do not exist on disk:\n"
f"{non_existent_text}\n"
f"Please check if the test files are correctly specified in the local repository."
)
def main():
arg_parser = argparse.ArgumentParser()
arg_parser.add_argument(
"--suite",
type=str,
default=list(suites.keys())[0],
choices=list(suites.keys()) + ["all"],
help="The suite to run",
)
arg_parser.add_argument(
"--auto-partition-id",
type=int,
help="Use auto load balancing. The part id.",
)
arg_parser.add_argument(
"--auto-partition-size",
type=int,
help="Use auto load balancing. The number of parts.",
)
arg_parser.add_argument(
"--continue-on-error",
action="store_true",
default=False,
help="Continue running remaining tests even if one fails (useful for nightly tests)",
)
args = arg_parser.parse_args()
print(f"{args=}")
_sanity_check_suites(suites)
files = suites[args.suite]
files_disabled = [f for f in files if f.is_skipped]
if args.auto_partition_size:
files = auto_partition(files, args.auto_partition_id, args.auto_partition_size)
# Print test info at beginning (similar to test/run_suite.py pretty_print_tests)
if args.auto_partition_size:
partition_info = (
f"{args.auto_partition_id + 1}/{args.auto_partition_size} (0-based id={args.auto_partition_id})"
)
else:
partition_info = "full"
headers = ["Suite", "Partition"]
rows = [[args.suite, partition_info]]
msg = tabulate.tabulate(rows, headers=headers, tablefmt="psql") + "\n"
total_est_time = sum(f.estimated_time for f in files)
msg += f"✅ Enabled {len(files)} test(s) (est total {total_est_time:.1f}s):\n"
for f in files:
msg += f" - {f.name} (est_time={f.estimated_time})\n"
msg += f"\n❌ Disabled {len(files_disabled)} test(s)(Please consider to recover them):\n"
for f in files_disabled:
msg += f" - {f.name} (est_time={f.estimated_time})\n"
print(msg, flush=True)
exit_code = run_e2e_files(
files,
continue_on_error=args.continue_on_error,
)
# Print tests again at the end for visibility
msg = "\n" + tabulate.tabulate(rows, headers=headers, tablefmt="psql") + "\n"
msg += f"✅ Executed {len(files)} test(s) (est total {total_est_time:.1f}s):\n"
for f in files:
msg += f" - {f.name} (est_time={f.estimated_time})\n"
print(msg, flush=True)
exit(exit_code)
if __name__ == "__main__":
main()