zlib/arm/adler32vec.s

64 	1. vmull.u32   q2,(adler,sum2),(1/BASE)		// *(1/BASE) in Q47
66     3. vpadd.u32   d4,d4,d5						// merge into a double word in d4
67     4. vmls.u32    (adler,sum2),d4,d3[0]        // (adler,sum2) -= floor[(adler,sum2)/BASE]*BASE
241 	vshll.u32	q12, adlersum2, #5			// d25 = (0,32*adler) to be added into (adler,sum2)
246 	vadd.u32	adlersum2,adler16			// sum2 += old adler*32;
261 	vpadd.u32	d16,d16,d17					// 2 32-bit elements for adler
262 	vpadd.u32	d17,d20,d21					// 2 32-bit elements for sum2
264 	vpadd.u32	d4,d17,d16					// s8 : 32-bit elements for sum2, s9 : 32-bit element for adler
265 	vadd.u32	adlersum2,d4				// update adler/sum2 with the new 16 bytes input
269 	vshll.u32	q12, adlersum2, #4			// d25 = (0,16*adler) to be added into (adler,sum2)
282 	vpadd.u32	d16,d16,d17					// 2 32-bit elements for adler
283 	vpadd.u32	d17,d20,d21					// 2 32-bit elements for sum2
284 	vadd.u32	adlersum2,adler16			// sum2 += old adler;
285 	vpadd.u32	d4,d17,d16					// s8 : 32-bit elements for sum2, s9 : 32-bit element for adler
286 	vadd.u32	adlersum2,d4				// update adler/sum2 with the new 16 bytes input
289 	vmull.u32	q2,adlersum2,d3[1]			// alder/BASE, sum2/BASE in Q47
291 	vpadd.u32	d4,d4,d5					// merge into a double word in d4
292 	vmls.u32	adlersum2,d4,d3[0]			// (adler,sum2) -= floor[(adler,sum2)/BASE]*BASE
310 	vshll.u32	q12, adlersum2, #5			// d25 = (0,32*adler) to be added into (adler,sum2)
315 	vadd.u32	adlersum2,adler16			// sum2 += old adler*32;
330 	vpadd.u32	d16,d16,d17					// 2 32-bit elements for adler
331 	vpadd.u32	d17,d20,d21					// 2 32-bit elements for sum2
333 	vpadd.u32	d4,d17,d16					// s8 : 32-bit elements for sum2, s9 : 32-bit element for adler
334 	vadd.u32	adlersum2,d4				// update adler/sum2 with the new 16 bytes input
343 	vshll.u32	q12, adlersum2, #4			// d25 = (0,16*adler) to be added into (adler,sum2)
357 	vpadd.u32	d16,d16,d17					// 2 32-bit elements for adler
358 	vpadd.u32	d17,d20,d21					// 2 32-bit elements for sum2
360 	vadd.u32	adlersum2,adler16			// sum2 += old adler;
361 	vpadd.u32	d4,d17,d16					// s8 : 32-bit elements for sum2, s9 : 32-bit element for adler
362 	vadd.u32	adlersum2,d4				// update adler/sum2 with the new 16 bytes input
384 	vmull.u32	q2,adlersum2,d3[1]			// alder/BASE, sum2/BASE in Q47
386 	vpadd.u32	d4,d4,d5					// merge into a double word in d4
387 	vmls.u32	adlersum2,d4,d3[0]			// (adler,sum2) -= floor[(adler,sum2)/BASE]*BASE