文件列表:

Bender.yml (4385, 2022-11-23)
LICENSE (10328, 2022-11-23)
Makefile (2771, 2022-11-23)
config.sh (1103, 2022-11-23)
requirements.txt (60, 2022-11-23)
rtl/ (0, 2022-11-23)
rtl/evt_conf_reg/ (0, 2022-11-23)
rtl/evt_conf_reg/evt_reggen/ (0, 2022-11-23)
rtl/evt_conf_reg/evt_reggen/evt_reggen.sv (7668, 2022-11-23)
rtl/evt_conf_reg/evt_reggen/evt_reggen_bus_cdc.sv (3823, 2022-11-23)
rtl/evt_conf_reg/evt_reggen/evt_reggen_engine_cdc.sv (4965, 2022-11-23)
rtl/evt_conf_reg/evt_reggen/evt_reggen_system_cdc.sv (4041, 2022-11-23)
rtl/evt_conf_reg/evt_reggen/generated/ (0, 2022-11-23)
rtl/evt_conf_reg/evt_reggen/generated/bus_clock_reg_pkg.sv (6918, 2022-11-23)
rtl/evt_conf_reg/evt_reggen/generated/bus_clock_reg_top.sv (91601, 2022-11-23)
rtl/evt_conf_reg/evt_reggen/generated/engine_clock_reg_pkg.sv (83625, 2022-11-23)
rtl/evt_conf_reg/evt_reggen/generated/engine_clock_reg_top.sv (1128882, 2022-11-23)
rtl/evt_conf_reg/evt_reggen/generated/system_clock_reg_pkg.sv (9099, 2022-11-23)
rtl/evt_conf_reg/evt_reggen/generated/system_clock_reg_top.sv (31702, 2022-11-23)
rtl/evt_conf_reg/hjson/ (0, 2022-11-23)
rtl/evt_conf_reg/hjson/bus_clock.hjson (6670, 2022-11-23)
rtl/evt_conf_reg/hjson/engine_clock.hjson (9552, 2022-11-23)
rtl/evt_conf_reg/hjson/system_clock.hjson (7319, 2022-11-23)
rtl/evt_crossbar/ (0, 2022-11-23)
rtl/evt_crossbar/evt_arbiter.sv (2197, 2022-11-23)
rtl/evt_crossbar/evt_crossbar.sv (3816, 2022-11-23)
rtl/evt_crossbar/evt_crossbar_dst_dev.sv (1951, 2022-11-23)
rtl/evt_crossbar/evt_crossbar_src_dev.sv (2230, 2022-11-23)
rtl/evt_crossbar/evt_fork.sv (1672, 2022-11-23)
rtl/evt_crossbar/evt_synaptic_crossbar.sv (4849, 2022-11-23)
rtl/evt_decoder/ (0, 2022-11-23)
rtl/evt_decoder/evt_decoder.sv (2805, 2022-11-23)
rtl/evt_decoder/evt_engine_router.sv (6522, 2022-11-23)
rtl/evt_decoder/evt_global_router.sv (6630, 2022-11-23)
rtl/evt_dp/ (0, 2022-11-23)
rtl/evt_dp/evt_dp_group.sv (6221, 2022-11-23)
rtl/evt_dp/evt_engine.sv (9479, 2022-11-23)
rtl/evt_dp/evt_filter.sv (3035, 2022-11-23)
... ...

SNE, the sparse neural engine, is a fully digital spiking neural network accelerator that can handle sparse convolutions efficiently. This repository contains the RTL description of the accelerator, as well as a basic test to simulate the accelerator standalone. ## structure The SNE repository is structured as follows: - `rtl` contains the RTL code and Verilog testbench for CUTIE - `sim` contains the files required to start the simulation - `vsne_conv` contains a high level python model of the SNE accelerator - `utils` contains several simulation utility script ### Dependencies This repo uses Bender (https://github.com/pulp-platform/bender) to manage its dependencies and generate compilation scripts. For this reason, the build process of this project will download a current version of the Bender binary. to fetch all the needed dependencies, execute: `make checkout` ### Simulating the SNE platform To build the RTL platform, execute the following command. `source config.sh` And then run: `make build opt` This will generate the required scripts and build the RTL. Then, to run a basic convolutional regression test, run: `make -C sim all` Currently, Modelsim is the only supported simulation platform. ## License SNE is released under permissive open source licenses. SNE's source code is released under the Solderpad v0.51 (`SHL-0.51`) license see [`LICENSE`](https://github.com/pulp-platform/sne/blob/master/LICENSE). The code in `vsne_conv` is released under the Apache License 2.0 (`Apache-2.0`) see [`vsne_conv/LICENSE`](https://github.com/pulp-platform/sne/blob/master/vsne_conv/LICENSE). ## Publication If you find SNE useful in your research, you can cite us: ``` @INPROCEEDINGS{9774552, author={Di Mauro, Alfio and Prasad, Arpan Suravi and Huang, Zhikai and Spallanzani, Matteo and Conti, Francesco and Benini, Luca}, booktitle={2022 Design, Automation & Test in Europe Conference & Exhibition (DATE)}, title={SNE: an Energy-Proportional Digital Accelerator for Sparse Event-Based Convolutions}, year={2022}, volume={}, number={}, pages={825-830}, doi={10.23919/DATE54114.2022.9774552}} ```

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：