Aayush-Ankit / puma-simulator

[ASPLOS 2019] PUMA-simulator provides a detailed simulation model of a dataflow architecture built with NVM (non-volatile memory), and runs ML models compiled using the puma compiler.
MIT License
50 stars 43 forks source link

How to get the inference result? #57

Open eAzure opened 2 years ago

eAzure commented 2 years ago

Hello, I'm trying to use this simulator. I want to know how to get the result of inference, like "mnist_l4_mnist". I'm not very clear about the meaning of “output.txt”. Thank you!

qzylalala commented 1 year ago

Hello, I'm trying to use this simulator. I want to know how to get the result of inference, like "mnist_l4_mnist". I'm not very clear about the meaning of “output.txt”. Thank you!

你好,现在这个问题有搞清楚吗,我似乎也找不到精度部分的代码位置?

negishubham commented 1 year ago

Hi,

Please follow the instructions in how_to_run to run the simulator. Further, there are some test scripts (https://github.com/Aayush-Ankit/puma-simulator/tree/training/test/utils) that you can try running to run some MLP and CNN test cases.

Thanks!

qzylalala commented 1 year ago

Hi,

Please follow the instructions in how_to_run to run the simulator. Further, there are some test scripts (https://github.com/Aayush-Ankit/puma-simulator/tree/training/test/utils) that you can try running to run some MLP and CNN test cases.

Thanks!

Thanks for your reply! I follow the instructions in “how_to_run”, and I Got “Success: Hardware results compiled!!” However,I have no idea where is the the result accuracy of inference.

negishubham commented 1 year ago

PUMA is a performance simulator. It estimates the energy, latency, and area for the IMC accelerator more information is in the ASPLOS paper. The performance results after running the tool can be found in test/traces folder. For accuracy, you should use the functional simulator (https://github.com/Aayush-Ankit/puma-functional-model).

qzylalala commented 1 year ago

PUMA is a performance simulator. It estimates the energy, latency, and area for the IMC accelerator more information is in the ASPLOS paper. The performance results after running the tool can be found in test/traces folder. For accuracy, you should use the functional simulator (https://github.com/Aayush-Ankit/puma-functional-model).

Thanks a lot for your reply. BTW, PUMA is great work!