"""
Copyright (c) 2026 Huawei Technologies Co., Ltd.
This program is free software, you can redistribute it and/or modify it under the terms and conditions of
CANN Open Software License Agreement Version 2.0 (the "License").
Please refer to the License for details. You may not use this file except in compliance with the License.
THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
See LICENSE in the root of the software repository for the full text of the License.
"""
import logging
import os
import sys
from dataclasses import dataclass
import torch
import torch_npu
import torch_sip
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(levelname)s - %(message)s',
handlers=[logging.StreamHandler(sys.stdout)]
)
logger = logging.getLogger("torch_sip_ssyr_test")
@dataclass(frozen=True)
class SyConfig:
"""参数对象:封装 Ssyr 测试的维度、标量及三角模式配置"""
n: int
alpha: float
uplo: str
name: str = "Case"
class SsyrTester:
"""Ssyr (对称矩阵秩-1更新) 算子测试类"""
def __init__(self, device: str = "npu:0"):
self.device = device
blocking_env = os.getenv("ASCEND_LAUNCH_BLOCKING", "0")
mode_str = "异步 (Async)" if blocking_env == "0" else "同步 (Blocking)"
logger.info("当前 NPU 模式: %s (BLOCKING=%s)", mode_str, blocking_env)
def run_case(self, cfg: SyConfig) -> bool:
"""
执行单次测试用例。
G.ERR.01: 最小化 try 块。
"""
mat_a_init_cpu = torch.randn(cfg.n, cfg.n, dtype=torch.float32)
vec_x_cpu = torch.randn(cfg.n, dtype=torch.float32)
mat_a_npu = mat_a_init_cpu.to(self.device)
vec_x_npu = vec_x_cpu.to(self.device).contiguous()
try:
torch_sip.asd_blas_ssyr(mat_a_npu, vec_x_npu, cfg.alpha, cfg.uplo)
except Exception as exc:
logger.error("[%s] 算子执行崩溃: %s", cfg.name, exc)
return False
update_mat = cfg.alpha * torch.outer(vec_x_cpu, vec_x_cpu)
expected_full = mat_a_init_cpu + update_mat
a_res_cpu = mat_a_npu.cpu()
is_lower = cfg.uplo.upper() == "L"
if is_lower:
expected = torch.tril(expected_full)
result = torch.tril(a_res_cpu)
else:
expected = torch.triu(expected_full)
result = torch.triu(a_res_cpu)
is_close = torch.allclose(result, expected, rtol=1e-4, atol=1e-4)
status = "PASS" if is_close else "FAIL"
logger.info("[%s] %-20s | N=%d | Uplo=%s | Alpha=%.2f",
status, cfg.name, cfg.n, cfg.uplo, cfg.alpha)
if not is_close:
logger.error(" Max Diff: %.6e", torch.abs(result - expected).max())
return is_close
def main():
"""测试主套件入口"""
tester = SsyrTester()
logger.info("开始 Ssyr 专项测试 (参数对象重构版本) ...\n")
test_suites = [
SyConfig(16, 1.0, "L", name="Small_Lower_Update"),
SyConfig(32, -0.5, "U", name="Upper_Neg_Alpha"),
SyConfig(128, 2.0, "L", name="Large_Lower_Update"),
SyConfig(512, 0.1, "U", name="Large_Upper_Update"),
]
all_passed = True
for config in test_suites:
if not tester.run_case(config):
all_passed = False
logger.info("-" * 60)
if all_passed:
logger.info("测试结论: ✅ 全部通过")
else:
logger.error("测试结论: ❌ 存在失败项")
return 0 if all_passed else 1
if __name__ == "__main__":
sys.exit(main())