error when running inference script

neurosim / DNN_NeuroSim_V1.3

Benchmark framework of compute-in-memory based accelerators for deep neural network (inference engine focused)

62 stars 36 forks source link

error when running inference script #2

Closed saba-er closed 2 years ago

saba-er commented 3 years ago

Hi, When I run inference.py, an error is raised after trace_command.sh is called. The error is "free(): double free detected in tcache 2". My configs are ubuntu 20.04 focal, gcc 9.3, cuda 10.1, cuDNN 7.5.0, python 3.6, pytorch 1.1.0 (according to the suggestions in the user manual). I appreciate it if you could help me resolving the issue.

msabri1372 commented 3 years ago

Hi I have a same error and I have spent a lot of time to solve it.Please help us.

parthoBhoumk commented 3 years ago

Hello, I also ran into the same error. If anyone has resolved the issue please let me know?

msabri1372 commented 3 years ago

I tested in the diffrent server and my problem have been solved! Try another GPU!

parthoBhoumk commented 3 years ago

I dont have another laptop / GPU :(

Can i run it on google colab?? any idea?

msabri1372 commented 3 years ago

I have tested it on university server. I am unfamiliar with Google Colab,you should try it.

parthoBhoumk commented 3 years ago

another question...i have installed exactly the versions mentioned above...did you try the same? if not then please can you tell me about all the versions that you have tried with??

msabri1372 commented 3 years ago

yes.I have instslled the same version that mentioned in the DNN NeuroSim V1.3 Manual . I have used conda to install them. firstly I have face the same problems and spent lots of time to address them but I couldn't solve them. By changing the GPU, it was solved. I recommend you to update your GPU driver and try it agin.

parthoBhoumk commented 3 years ago

Hey can you tell me which GPU you have used in your university server?

ZimingYuan commented 3 years ago

I also met the same error and solved it by modifying the makefile of NeuroSim: remove the "-O3" compile flag. It seems the compiler's optimization caused the error but I don't know why.

datMaffin commented 3 years ago

I was able to fix the crashing by adding missing return statements. (Got the idea because of https://github.com/neurosim/MLP_NeuroSim_V3.0/pull/4 )

Inference_pytorch/NeuroSIM/Chip.cpp at the end of double ChipCalculatePerformance(...) (I added a return 0;)
Inference_pytorch/NeuroSIM/ProcessingUnit.cpp at the end of double ProcessingUnitCalculatePerformance(...) (I, again, added a return 0;)

neurosim commented 3 years ago

Hi guys, thanks for the discussions and suggestions from you all! The return 0 issue has been updated in the codes. You're welcome to reply here whether the problem is solved or any other changes are needed like the compilation.