norm.cpp(normL2Sqr_): improve performance of pipeline
authorJojo R <rjiejie@me.com>
Tue, 17 Nov 2020 07:29:03 +0000 (15:29 +0800)
committerJojo R <rjiejie@me.com>
Thu, 19 Nov 2020 01:49:49 +0000 (09:49 +0800)
commit12b8d542b7465f495681d3dc0c50cd27f7e0ee94
treefcdafba415bf088faec5bd9905dca09aa7822274
parent564d1a0f79b3a7a528138b990837290f6cacfdd6
norm.cpp(normL2Sqr_): improve performance of pipeline

The most of target machine use one type cpu unit resource
to execute some one type of instruction, e.g.
all vx_load API use load/store cpu unit,
and v_muladd API use mul/mula cpu unit, we interleave
vx_load and v_muladd to improve performance on most targets like
RISCV or ARM.
modules/core/src/norm.cpp