meta data for this page
  •  

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
arm:neon:basics [2025/11/16 13:17] – niziakarm:neon:basics [2025/11/17 07:30] (current) – niziak
Line 5: Line 5:
     * load all bytes from pointed memory ``vld1_``     * load all bytes from pointed memory ``vld1_``
     * load from interleaved memory - support up to every 4th element (usefull for RGBA / CMYK): ``vld2_`` ``vld3_`` ``vld4_``.     * load from interleaved memory - support up to every 4th element (usefull for RGBA / CMYK): ``vld2_`` ``vld3_`` ``vld4_``.
-    * load from interleaved/random memory - possible to load one vector element (lane) from given pointer: ``vld1_lane_``+    * load from interleaved/random memory - possible to load one vector element (lane) from given pointer: ''vld1_lane_''
       * random element store ''ret[0]=vgetq_lane_s32(o,0)'', ''ret[1]=vgetq_lane_s32(o,1)'' etc       * random element store ''ret[0]=vgetq_lane_s32(o,0)'', ''ret[1]=vgetq_lane_s32(o,1)'' etc
     * load using lookup tables (permutation) - possible to load 8byte vector using 2nd 8 bytes vector with memory position indexes.     * load using lookup tables (permutation) - possible to load 8byte vector using 2nd 8 bytes vector with memory position indexes.
  
   * combine 8 bit values    * combine 8 bit values