hub / github.com/Tencent/ncnn / convert_input_layout

Function convert_input_layout

tests/perf/perfutil.cpp:84–220 · view source on GitHub ↗

convert input mat to the optimal layout that the layer expects this handles fp16/bf16 cast and packing conversion so that these costs are excluded from timing

Source from the content-addressed store, hash-verified

source not stored for this graph (policy: none)

Callers 1

perf_layer_cpuFunction · 0.85

Calls 15

cpu_support_arm_asimdhpFunction · 0.85

cast_float32_to_float16Function · 0.85

cpu_support_arm_vfpv4Function · 0.85

cpu_support_riscv_zvfhFunction · 0.85

cpu_support_riscv_vFunction · 0.85

cpu_support_riscv_zfhFunction · 0.85

cast_float32_to_bfloat16Function · 0.85

cpu_support_x86_avx512Function · 0.85

cpu_support_x86_avxFunction · 0.85

cpu_riscv_vlenbFunction · 0.85

cpu_support_loongarch_lasxFunction · 0.85

cpu_support_loongarch_lsxFunction · 0.85

Tested by

no test coverage detected