First release of my HLS benchmarks. The benchmarks include:
- gSum sums a number of polynomial results from the array elements that meet the given conditions where the difference between two elements from the arrays is non-negative.
- gSumIf is similar to gSum but the SS function returns one of two polynomial expressions based on the value of the difference.
- sparseMatrixPower performs dot product of two matrices, which skips the operation when the weight is zero.
- histogram sums various weight onto the corresponding features but also in a sparse form.
- getTanh performs the approximated function tanh(x) onto an array of integers using the CORDIC algorithm and a polynomial function.
- getTanh(double) is similar to getTanh but uses an array of doubles.
- BNNKernel is a small BNN kernel with LUT function as XOR.