xxmd1132 opened a new issue, #16285:
URL: https://github.com/apache/tvm/issues/16285
### Expected behavior
can run target_model which had compplile throught python API function:
tvmc.run(target_model, device="cpu")
### Actual behavior
run fail when run line: `result = tvmc.run(target_model, device="cpu") # 第 3
步：运行` of python script 'python resnet_test.py'
here is error message:
```bash
get tvm delay model
One or more operators have not been tuned. Please tune your model for better
performance. Use DEBUG logging level to see more details.
get tvm target model, lib name:mod.so
Traceback (most recent call last):
File "tests/zxq/resnet_test.py", line 60, in <module>
local_compile(onnx_path)
File "tests/zxq/resnet_test.py", line 55, in local_compile
result = tvmc.run(target_model, device="cpu") # 第 3 步：运行
File "/Users/zxq/EzvizEngine/tvm/python/tvm/driver/tvmc/runner.py", line
589, in run_module
lib = session.load_module(tvmc_package.lib_name)
File "/Users/zxq/EzvizEngine/tvm/python/tvm/rpc/client.py", line 178, in
load_module
return _ffi_api.LoadRemoteModule(self._sess, path)
File "/Users/zxq/EzvizEngine/tvm/python/tvm/_ffi/_ctypes/packed_func.py",
line 239, in __call__
raise_last_ffi_error()
File "/Users/zxq/EzvizEngine/tvm/python/tvm/_ffi/base.py", line 481, in
raise_last_ffi_error
raise py_err
File "/Users/zxq/EzvizEngine/tvm/python/tvm/rpc/server.py", line 80, in
load_module
m = _load_module(path)
File "/Users/zxq/EzvizEngine/tvm/python/tvm/runtime/module.py", line 693,
in load_module
return _ffi_api.ModuleLoadFromFile(path, fmt)
File "/Users/zxq/EzvizEngine/tvm/python/tvm/_ffi/_ctypes/packed_func.py",
line 239, in __call__
raise_last_ffi_error()
File "/Users/zxq/EzvizEngine/tvm/python/tvm/_ffi/base.py", line 481, in
raise_last_ffi_error
raise py_err
tvm.error.InternalError: Traceback (most recent call last):
[bt] (8) 9 libffi.8.dylib 0x000000010512004c
ffi_call_SYSV + 76
[bt] (7) 8 libtvm.dylib 0x0000000179b0580c
TVMFuncCall + 64
[bt] (6) 7 libtvm.dylib 0x0000000179b2c720 void
tvm::runtime::TypedPackedFunc<tvm::runtime::Module (tvm::runtime::String
const&, tvm::runtime::String const&)>::AssignTypedLambda<tvm::runtime::Module
(*)(tvm::runtime::String const&, tvm::runtime::String
const&)>(tvm::runtime::Module (*)(tvm::runtime::String const&,
tvm::runtime::String const&), std::__1::basic_string<char,
std::__1::char_traits<char>,
std::__1::allocator<char>>)::'lambda'(tvm::runtime::TVMArgs const&,
tvm::runtime::TVMRetValue*)::operator()(tvm::runtime::TVMArgs const&,
tvm::runtime::TVMRetValue*) const + 148
[bt] (5) 6 libtvm.dylib 0x0000000179b27e48
tvm::runtime::Module::LoadFromFile(tvm::runtime::String const&,
tvm::runtime::String const&) + 952
[bt] (4) 5 libtvm.dylib 0x0000000179b16748
tvm::runtime::PackedFuncObj::Extractor<tvm::runtime::PackedFuncSubObj<tvm::runtime::$_0>>::Call(tvm::runtime::PackedFuncObj
const*, tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) + 148
[bt] (3) 4 libtvm.dylib 0x0000000179b16568
tvm::runtime::DSOLibrary::Load(std::__1::basic_string<char,
std::__1::char_traits<char>, std::__1::allocator<char>> const&) + 216
[bt] (2) 3 libtvm.dylib 0x00000001780033f4
__clang_call_terminate + 0
[bt] (1) 2 libtvm.dylib 0x0000000178006280
tvm::runtime::detail::LogFatal::Entry::Finalize() + 0
[bt] (0) 1 libtvm.dylib 0x00000001780062c4
tvm::runtime::detail::LogFatal::Entry::Finalize() + 68
File "/Users/zxq/EzvizEngine/tvm/src/runtime/dso_library.cc", line 125
InternalError: Check failed: (lib_handle_ != nullptr) is false: Failed to
load dynamic shared library
/private/var/folders/jf/nl4dq34n14xcm_xbslm9qyyr0000gq/T/tmpqs9vsxka/mod.so
dlopen(/private/var/folders/jf/nl4dq34n14xcm_xbslm9qyyr0000gq/T/tmpqs9vsxka/mod.so,
0x0005): tried:
'/private/var/folders/jf/nl4dq34n14xcm_xbslm9qyyr0000gq/T/tmpqs9vsxka/mod.so'
(mach-o file, but is an incompatible architecture (have 'x86_64', need
'arm64')),
'/System/Volumes/Preboot/Cryptexes/OS/private/var/folders/jf/nl4dq34n14xcm_xbslm9qyyr0000gq/T/tmpqs9vsxka/mod.so'
(no such file),
'/private/var/folders/jf/nl4dq34n14xcm_xbslm9qyyr0000gq/T/tmpqs9vsxka/mod.so'
(mach-o file, but is an incompatible architecture (have 'x86_64', need 'arm64'))
```
### Environment
Hardware:
Hardware Overview:
Model Name: Mac
Model Identifier: Mac14,14
Model Number: Z180000JDCH/A
Chip: Unknown
Total Number of Cores: 24 (16 performance and 8 efficiency)
Memory: 192 GB
System Firmware Version: 8422.121.3
OS Loader Version: 8422.121.3
Serial Number (system): JXP6CFK63K
Hardware UUID: 3DEB4A8F-FA91-5263-948A-8C73653657CF
Provisioning UDID: 3DEB4A8F-FA91-5263-948A-8C73653657CF
Activation Lock Status: Disabled
llvm-g++ version:
Apple clang version 14.0.3 (clang-1403.0.22.14.1)
Target: x86_64-apple-darwin22.5.0
Thread model: posix
InstalledDir: /Library/Developer/CommandLineTools/usr/bin
python version(conda virtual env):3.9.16
### Steps to reproduce
1. build 'libtvm.dylib', here is the command which of [refer
link](https://tvm.hyper.ai/docs/install/from_source) :
Here are some modification Settings for cmakelist.txt
```bash
set(USE_LLVM ON)
set(USE_LIBBACKTRACE OFF) # Use libbacktrace to supply linenumbers on stack
traces
```
then get `libtvm_runtime.dylib` and `libtvm.dylib`
2. export PYTHONPATH=$tvm_root/python
3. run python script:`python resnet_test.py`
here is python script(resnet_test.py) which refer by example of repo and
test model is a resnet-50 which download by python example of repo
```python
from tvm.contrib.download import download_testdata
from tvm.driver import tvmc
from PIL import Image
import logging
from pathlib import Path
import numpy as np
def local_compile(onnx:Path):
target_model_p = onnx.with_name(onnx.stem + "_target.tvm")
relay_model_p = onnx.with_name(onnx.stem + "_relay.tvm")
tvm_model = tvmc.load(onnx) # onnx -> TVM Relay，可以使用 model.summary()打印
print(f"get tvm delay model")
## modify input [optional]
tvm_model = tvmc.load(onnx, shape_dict={'input1' : [1, 2, 3, 4],
'input2' : [1, 2, 3, 4]}) # if input model is pt, this step is required!
## Relay -> Target
target_model = tvmc.compile(tvm_model, target="llvm",
package_path=target_model_p) # 第 2 步：编译
print(f"get tvm target model, lib name:{target_model.lib_name}")
# run test
result = tvmc.run(target_model, device="cpu") # 第 3 步：运行
print("run target model success")
onnx_path = Path(r"/Users/zxq/EzvizEngine/tvm/tests/zxq/resnet50-v2-7.onnx")
local_compile(onnx_path)
```
### Triage
* backend: llvm
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
