Describe the current behavior / 问题描述 (Mandatory / 必填)

在启智平台上910a训练SDXL网络时，recompute_Default/BatchMatMul-op6207算子启动失败

Environment / 环境信息 (Mandatory / 必填)

• Hardware Environment(Ascend/GPU/CPU) / 硬件环境: Ascend

Please delete the backend not involved / 请删除不涉及的后端:
/device ascend/GPU/CPU/kirin/等其他芯片

• Software Environment / 软件环境 (Mandatory / 必填):
  -- MindSpore version (e.g., 1.7.0.Bxxx) :mindspore2.2.0
  -- Python version (e.g., Python 3.7.5) :python3.9
  -- OS platform and distribution (e.g., Linux Ubuntu 16.04):eulerOS2.8
  -- GCC/Compiler version (if compiled from source):

• Excute Mode / 执行模式 (Mandatory / 必填)(PyNative/Graph): PyNative

Please delete the mode not involved / 请删除不涉及的模式:
/mode pynative
/mode graph

Related testcase / 关联用例 (Mandatory / 必填)

mindone/example/stable_diffusion_xl/train.py

Steps to reproduce the issue / 重现步骤 (Mandatory / 必填)

1. 设置训练为pynative模式
2. peft lokr包裹unet
3. 使用原程序开启训练

Describe the expected behavior / 预期结果 (Mandatory / 必填)

算子启动成功

Related log / screenshot / 日志 / 截图 (Mandatory / 必填)

[WARNING] KERNEL(16170,ffffa3d51440,python):2024-04-27-14:32:35.815.416 [mindspore/ccsrc/transform/acl_ir/op_api_convert.h:75] GetOpApiLibHandler] Dlopen /home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/run_check/../lib/plugin/ascend/custom_aicpu_ops/op_api/lib/libcust_opapi.so failed!/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/run_check/../lib/plugin/ascend/custom_aicpu_ops/op_api/lib/libcust_opapi.so: cannot open shared object file: No such file or directory
/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/ops/operations/array_ops.py:376: RuntimeWarning: overflow encountered in cast
value = Tensor(np.array(x).astype(np_dst_type), dtype=dst_type)
[ERROR] KERNEL(16170,ffffa3d51440,python):2024-04-27-14:33:26.077.708 [mindspore/ccsrc/plugin/device/ascend/kernel/acl/acl_kernel_mod.cc:253] Launch] Kernel launch failed, msg: The pointer[block] is null.

• Ascend Warning Message:

W11001: Op [Range] does not hit the high-priority operator information library, which might result in compromised performance.
W11001: Op [Cast] does not hit the high-priority operator information library, which might result in compromised performance.

• Framework Unexpected Exception Raised:

This exception is caused by framework's unexpected error. Please create an issue at https://gitee.com/mindspore/mindspore/issues to get help.

• C++ Call Stack: (For framework developers)

mindspore/ccsrc/transform/acl_ir/acl_allocator.cc:40 AllocFunc

[ERROR] DEVICE(16170,ffffa3d51440,python):2024-04-27-14:33:26.077.779 [mindspore/ccsrc/plugin/device/ascend/hal/hardware/ge_kernel_executor.cc:583] LaunchKernel] Launch kernel failed, kernel full name: recompute_Default/BatchMatMul-op6207
Traceback (most recent call last):
File "/tmp/code/mindone/examples/stable_diffusion_xl/train_lokr.py", line 717, in
train(args)
File "/tmp/code/mindone/examples/stable_diffusion_xl/train_lokr.py", line 352, in train
train_fn(
File "/tmp/code/mindone/examples/stable_diffusion_xl/train_lokr.py", line 391, in train_txt2img
loss, overflow = train_step_fn(image, *tokens)
File "/tmp/code/mindone/examples/stable_diffusion_xl/gm/models/diffusion.py", line 181, in train_step_pynative
loss, , overflow = grad_func(x, noised_input, sigmas, w, **cond)
File "/tmp/code/mindone/examples/stable_diffusion_xl/gm/models/diffusion.py", line 161, in jit_warpper
return grad_and_update_func(*args, **kwargs)
File "/tmp/code/mindone/examples/stable_diffusion_xl/gm/models/diffusion.py", line 147, in grad_and_update_func
loss, grads = grad_fn(x, noised_input, sigmas, w, concat, context, y)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/ops/composite/base.py", line 625, in after_grad
return grad(fn_, weights)(*args, **kwargs)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/common/api.py", line 121, in wrapper
results = fn(*arg, **kwargs)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/ops/composite/base.py", line 600, in after_grad
res = self.pynative_forward_run(fn, grad, weights, args, kwargs)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/ops/composite/base.py", line 650, in _pynative_forward_run
outputs = fn(*args, **new_kwargs)
File "/tmp/code/mindone/examples/stable_diffusion_xl/gm/models/diffusion.py", line 132, in _forward_func
model_output = model(
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/nn/cell.py", line 705, in call
raise err
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/nn/cell.py", line 701, in call
output = self._run_construct(args, kwargs)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/nn/cell.py", line 482, in _run_construct
output = self.construct(*cast_inputs, **kwargs)
File "/tmp/code/mindnlp/mindnlp/peft/tuners/tuners_utils.py", line 113, in construct
return self.model.construct(*args, **kwargs)
File "/tmp/code/mindone/examples/stable_diffusion_xl/gm/modules/diffusionmodules/wrappers.py", line 23, in construct
return self.diffusion_model(x, timesteps=t, context=context, y=y, **kwargs)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/nn/cell.py", line 705, in call
raise err
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/nn/cell.py", line 701, in call
output = self._run_construct(args, kwargs)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/nn/cell.py", line 482, in _run_construct
output = self.construct(*cast_inputs, **kwargs)
File "/tmp/code/mindone/examples/stable_diffusion_xl/gm/modules/diffusionmodules/openaimodel.py", line 801, in construct
h = self.middle_block(h, emb, context)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/nn/cell.py", line 705, in call
raise err
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/nn/cell.py", line 701, in call
output = self._run_construct(args, kwargs)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/nn/cell.py", line 482, in _run_construct
output = self.construct(*cast_inputs, **kwargs)
File "/tmp/code/mindone/examples/stable_diffusion_xl/gm/modules/diffusionmodules/openaimodel.py", line 42, in construct
x = cell(x, context)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/nn/cell.py", line 705, in call
raise err
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/nn/cell.py", line 701, in call
output = self._run_construct(args, kwargs)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/nn/cell.py", line 482, in _run_construct
output = self.construct(*cast_inputs, **kwargs)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/ops/_tracefunc.py", line 235, in _trace_wrap
return pack_func(*args, **kwargs)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/ops/_tracefunc.py", line 84, in call
output = super().call(*args)
File "/tmp/code/mindnlp/mindnlp/injection.py", line 176, in _op_call
return old_op_call(self, *args)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/ops/primitive.py", line 314, in call
return _run_op(self, self.name, args)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/ops/primitive.py", line 913, in _run_op
stub = _pynative_executor.run_op_async(obj, op_name, args)
File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/mindspore/common/api.py", line 1186, in run_op_async
return self._executor.run_op_async(*args)
RuntimeError: Launch kernel failed, name:recompute_Default/BatchMatMul-op6207

• Ascend Warning Message:

W11001: Op [Range] does not hit the high-priority operator information library, which might result in compromised performance.
W11001: Op [Cast] does not hit the high-priority operator information library, which might result in compromised performance.

• C++ Call Stack: (For framework developers)

mindspore/ccsrc/runtime/pynative/run_op_helper.cc:1052 LaunchKernels

• C++ Call Stack: (For framework developers)

mindspore/ccsrc/runtime/pynative/run_op_helper.cc:1052 LaunchKernels

Special notes for this issue/备注 (Optional / 选填)