Changes for floating point multiplier

[lord-tarun]

Dear authors, Thank you for making your idea open source. I looked through the issues raised earlier and could not find similar points raised. I was going through the repository and saw that we tested the results for 8-bit multipliers. However, for my work, I require 32-bit floating point multipliers. As per my understanding, two changes will be needed. Could you please let me know what more changes might need to be made? Thanks in advance!

1. Removing quantization-related functions in FakeApproxConv2D (present in tf2/python/keras/layers/fake_approx_convolutional.py) to allow floating point multipliers.
2. Creating the binary file for the multiplier by changing the range from 256 (2^8) with the floating point range (2^32), as shown below.

FILE * f = fopen("output.bin", "wb");

for(unsigned int a = 0; a < (2^32); a++)
    for(unsigned int b = 0; b < (2^32); b++) {
      long val = approximate_mult(a, b); // replace by your own function call
      fwrite(&val, sizeof(uint16_t), 1, f);
    }

fclose(f);

[xiaoxixideluoke]

Hello, do you have the full file of this emulator, the "https://ehw.fit.vutbr.cz/tf-approximate/tf-approximate-gpu.sif" website is invalid and the file cannot be found, can you please share it？Thanks a lot