* process particular chunks of the input data stream in parallel.
*
* For the CRC-32 variants, the constants are precomputed according to
- * these defintions:
+ * these definitions:
*
* R1 = x4*128+64 mod P(x)
* R2 = x4*128 mod P(x)
* Note: To compensate the division by x^32, use the vector unpack
* instruction to move the leftmost word into the leftmost doubleword
* of the vector register. The rightmost doubleword is multiplied
- * with zero to not contribute to the intermedate results.
+ * with zero to not contribute to the intermediate results.
*/
/* T1(x) = floor( R(x) / x^32 ) GF2MUL u */