mask Class — pytorch Architecture
Architecture documentation for the mask class in vec128_uint_aarch64.h from the pytorch codebase.
Entity Profile
Source Code
aten/src/ATen/cpu/vec/vec128/vec128_uint_aarch64.h lines 222–247
template <uint64_t mask>
Vectorized<uint8_t> Vectorized<uint8_t>::blend(
const Vectorized<uint8_t>& a,
const Vectorized<uint8_t>& b) {
// Build an array of flags: each bit of element is 1 if the corresponding bit
// in 'mask' is set, 0 otherwise.
uint8x16_t maskArray = {
(mask & 1LL) ? 0xFF : 0,
(mask & 2LL) ? 0xFF : 0,
(mask & 4LL) ? 0xFF : 0,
(mask & 8LL) ? 0xFF : 0,
(mask & 16LL) ? 0xFF : 0,
(mask & 32LL) ? 0xFF : 0,
(mask & 64LL) ? 0xFF : 0,
(mask & 128LL) ? 0xFF : 0,
(mask & 256LL) ? 0xFF : 0,
(mask & 512LL) ? 0xFF : 0,
(mask & 1024LL) ? 0xFF : 0,
(mask & 2048LL) ? 0xFF : 0,
(mask & 4096LL) ? 0xFF : 0,
(mask & 8192LL) ? 0xFF : 0,
(mask & 16384LL) ? 0xFF : 0,
(mask & 32768LL) ? 0xFF : 0};
// Use BSL to select elements from b where the mask is 1, else from a
return vbslq_u8(maskArray, b.values, a.values);
}
Source
Analyze Your Own Codebase
Get architecture documentation, dependency graphs, and domain analysis for your codebase in minutes.
Try Supermodel Free