Xilinx / finn

Dataflow compiler for QNN inference on FPGAs
https://xilinx.github.io/finn
BSD 3-Clause "New" or "Revised" License
723 stars 230 forks source link

Memory Primitive Issue #1126

Closed Ba1tu3han closed 2 months ago

Ba1tu3han commented 3 months ago

Prerequisites

Please make sure to check off these prerequisites before submitting a bug report.

Quick summary

I run _./run-docker.sh build_dataflow ~/Desktop/dataflow_builddir for W1A1 CNV model and encountered with an error about Memory Primitive in 15/19 of building steps.

Could you help me how to define it or what I should select?

Screenshot from 2024-07-06 23-14-52 image folding_config.json

Steps to Reproduce

Add what needs to be done to reproduce the bug. Add code examples where useful and make sure to include the resulting ONNX files, and the commit hash you are working on.

  1. Clone the FINN repository
  2. Checkout the dev branch, with commit hash: [9f0fad5717ae652eaa324583e957161890964555]
  3. Start the docker container with the command: [./run-docker.sh build_dataflow ~/Desktop/dataflow_build_dir]

Optional

Possible fix

Setting an option about FIFO.

auphelia commented 3 months ago

Hi @Ba1tu3han , from your log files it looks like your build completes successfully. The info message you are seeing is not an error. But if you like to select a specific memory type for the FIFOs (it defaults to "auto"), you can set the ram_style argument of the FIFOs in you folding config json. You can use the final_hw_config.json as starting point.