Spent days debugging instruction decoding,Built my custom pipelined MIPS processor on Basys 3 FPGA with a working dot-product accelerator.
After endless debugging, finally got the correct hardware output: 0032.
Now I want to make it dynamic and want to cnn in it can anyone help
Hello everyone, myself Er. Sandeep Kumar
Electronics and Communication Engineer
Seeking job opportunities in VLSI, Embedded Systems, FPGA Design, and Digital Electronics.
@richnanophd Thatโs a really solid suggestion. I was already hitting BRAM limits on Basys 3, so mapping 2D convolution into serialized 1D MAC operations with weight reuse makes a lot more sense now. Definitely going to experiment with line buffers and sliding window optimization next.
Spent days debugging instruction decoding,Built my custom pipelined MIPS processor on Basys 3 FPGA with a working dot-product accelerator.
After endless debugging, finally got the correct hardware output: 0032.
Now I want to make it dynamic and want to cnn in it can anyone help
@dwasf79850 Thatโs seriously ahead of its time. Dynamic hardware acceleration on Virtex boards mustโve been wild to experiment with back then. Feels like the early roots of what people now market as heterogeneous computing and AI accelerators.
@MaxClerkwell Absolutely. Iโm cleaning up the repo and documentation properly first so itโs easier for students to follow. Iโll share the GitHub link here as soon as itโs ready.
@dwasf79850 MIPS on Spartan FPGA is such a solid learning experience. Those boards introduced an entire generation to real hardware design. What project did you build on it?