GitPlanet

Cheap and reliable Node.js hosting starts at $3/month, and $1/month static HTML hosting

Created with love in Canada, visit hostnodejs.com today

Feel like to post an Ad? Learn Details
All Projects → hunterlew → Convolution_network_on_fpga

hunterlew / Convolution_network_on_fpga

CNN acceleration on virtex-7 FPGA with verilog HDL

Labels

deep-learningcnnverilogfpga

Projects that are alternatives of or similar to Convolution network on fpga

Symbiflow Arch Defs
FOSS architecture definitions of FPGA hardware useful for doing PnR device generation.
Stars: ✭ 137 (-41.95%)
Mutual labels:  verilog, fpga
Tinytpu
Implementation of a Tensor Processing Unit for embedded systems and the IoT.
Stars: ✭ 153 (-35.17%)
Mutual labels:  verilog, fpga
Openfpgaduino
All open source file and project for OpenFPGAduino project
Stars: ✭ 137 (-41.95%)
Mutual labels:  verilog, fpga
Aes
Verilog implementation of the symmetric block cipher AES (Advanced Encryption Standard) as specified in NIST FIPS 197. This implementation supports 128 and 256 bit keys.
Stars: ✭ 131 (-44.49%)
Mutual labels:  verilog, fpga
Openwifi Hw
FPGA/hardware design of openwifi
Stars: ✭ 181 (-23.31%)
Mutual labels:  verilog, fpga
Wbuart32
A simple, basic, formally verified UART controller
Stars: ✭ 133 (-43.64%)
Mutual labels:  verilog, fpga
Logic
CMake, SystemVerilog and SystemC utilities for creating, building and testing RTL projects for FPGAs and ASICs.
Stars: ✭ 149 (-36.86%)
Mutual labels:  verilog, fpga
Tf2
An Open Source Deep Learning Inference Engine Based on FPGA
Stars: ✭ 113 (-52.12%)
Mutual labels:  fpga, cnn
Wb2axip
Bus bridges and other odds and ends
Stars: ✭ 177 (-25%)
Mutual labels:  verilog, fpga
Fpga Chip8
CHIP-8 console on FPGA
Stars: ✭ 169 (-28.39%)
Mutual labels:  verilog, fpga
Open Register Design Tool
Tool to generate register RTL, models, and docs using SystemRDL or JSpec input
Stars: ✭ 126 (-46.61%)
Mutual labels:  verilog, fpga
Red Pitaya Notes
Notes on the Red Pitaya Open Source Instrument
Stars: ✭ 205 (-13.14%)
Mutual labels:  verilog, fpga
Connectal
Connectal is a framework for software-driven hardware development.
Stars: ✭ 117 (-50.42%)
Mutual labels:  verilog, fpga
Tang e203 mini
LicheeTang 蜂鸟E203 Core
Stars: ✭ 135 (-42.8%)
Mutual labels:  verilog, fpga
Vm80a
i8080 precise replica in Verilog, based on reverse engineering of real die
Stars: ✭ 114 (-51.69%)
Mutual labels:  verilog, fpga
Openwifi
open-source IEEE 802.11 WiFi baseband FPGA (chip) design
Stars: ✭ 2,257 (+856.36%)
Mutual labels:  verilog, fpga
Autofpga
A utility for Composing FPGA designs from Peripherals
Stars: ✭ 108 (-54.24%)
Mutual labels:  verilog, fpga
Livehd
Live Hardware Development (LiveHD), a productive infrastructure for Synthesis and Simulation
Stars: ✭ 110 (-53.39%)
Mutual labels:  verilog, fpga
Fpga readings
Recipe for FPGA cooking
Stars: ✭ 164 (-30.51%)
Mutual labels:  verilog, fpga
Verilog Generator Of Neural Net Digit Detector For Fpga
Verilog Generator of Neural Net Digit Detector for FPGA
Stars: ✭ 187 (-20.76%)
Mutual labels:  verilog, fpga
View All Similar Projects ➔

convolution_network_on_FPGA

The repository is part of my graduation project, but focusing on convolution network inference acceleration on FPGA.

Description

The project was built with ISE 14.7 software and vertix-7 FPGA. It performs a 7-layer network forward computation with certain accelerating strategies. First, train a SAR target classification network on MSTAR dataset with MatConvNet and use early-stop. Then, transform the weights and inputs to FPGA using COE files generated by Matlab. Next, design native verilog-written models to fulfil the forward computation. The system uses 16-bit fix-point data to maintain precision. The result proves that it costs less than 1ms per image, outperforming other computing platforms.

Acceleration

For trade-off between FPGA resource and speed, the first conv. layer uses shift-register structures for acceleration, considering that the size of inputs and weights in the first conv. layer are larger. The other conv. layers only uses pipe-line structures. Note that there are 18 files named with m_conv_1, indicating 18 shift-registers in the first conv. layer.

For the limit of bandwidth, the second conv. layer uses channel-grouping parallelization scheme. I used v7-415t and v7-485t FPGA for comparisons, which use different group numbers (4 vs 1) according to their different amounts of resource. You may notice the scheme in the file CNN_top.v, with the signal 'ram_ready' which controls the channel grouping and merging.

Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].