19 def _AlignForLanes(lanes_count):  argument
20   if lanes_count is 8 or lanes_count is 4:
22   elif lanes_count is 6 or lanes_count is 2:
28 def _AlignForSums(lanes_count):  argument
29   if lanes_count is 8:
31   elif lanes_count in [2, 4, 6]:
37 def _GenerateInputs(emitter, registers, lanes_count, input_address, stride):  argument
41   for i in range(lanes_count):
57 def _GenerateLoadAggregateStore(emitter, registers, lanes_count, elements_count,  argument
61   emitter.EmitComment('Load Aggregate Store: %dx%d.' % (lanes_count,
64   block = [registers.DoubleRegister() for unused_i in range(lanes_count)]
75   emitter.EmitVStoreAE(8, 8 * lanes_count, block, output,
76                        _AlignForLanes(lanes_count))
148   def EmitPack(self, in_type, lanes_count, pack_size, leftovers):  argument
161     inputs = _GenerateInputs(self.asm_emitter, registers, lanes_count,
164     aggregators = [registers.QuadRegister(8) for unused_i in range(lanes_count)]
180     _GenerateLoadAggregateStore(self.asm_emitter, registers, lanes_count, 8,
189       _GenerateLoadAggregateStore(self.asm_emitter, registers, lanes_count,
205 def _GenerateColLoadAggregateStore(emitter, registers, lanes_count,  argument
211                       (lanes_count, elements_count))
213   block = [registers.DoubleRegister() for unused_i in range(lanes_count)]
218   block = emitter.EmitLoadColBlock(registers, 8, lanes_count, elements_count,
224   emitter.EmitVStoreAE(8, 8 * lanes_count, block, output,
225                        _AlignForLanes(lanes_count))
237   def EmitPack(self, in_type, lanes_count, pack_size, leftovers):  argument
252     aggregators = [registers.QuadRegister(8) for unused_i in range(lanes_count)]
255     self.asm_emitter.EmitColBlockStride(lanes_count, stride, stride)
271     _GenerateColLoadAggregateStore(self.asm_emitter, registers, lanes_count, 8,
281       _GenerateColLoadAggregateStore(self.asm_emitter, registers, lanes_count,
293 def GenerateUInt8x8Streams(cc_emitter, asm_emitter, lanes_count):  argument
297   for lanes_count in range(1, 1 + lanes_count):
299       row_major_with_sum.SpecializeStream('uint8_t', lanes_count, 8, leftovers)
301   for lanes_count in range(1, 1 + lanes_count):
303       column_major_with_sum.SpecializeStream('uint8_t', lanes_count, 8,