// optimizations version 2 // 74.2 s // remove debug statements from processBlock // rotate 30 bits // 65.2 s // split into 4 loops // _getK // 67.4 // rotate 5 bits // split into 5 loops // 66.3 // _getF // 63.9 // _getW // 63.9 // avoid j*4 while looping // 63.1
// optimizations version 3 // avoid copying data to intermediate buffer // 63.669 64.839
// optimizations version 4 // use heap to store ABCDE and it got slower // 79.008 // use heap for WORDS - this gave a big boost // measuring things in node does not make any sense // as typed arrays are slow in node (node uses buffers)
// optimizations version 5 // removed the DataView when writing the first 16 words to heap