I don't understand how to encode PHSUBSW with VEX.256. According to the documentation, VEX.L must be 0, otherwise the instruction will #UD. I thought we must use VEX.L = 1 to force the VEX.256.
- Intel® Advanced Vector Extensions (Intel® AVX)
- Intel® Streaming SIMD Extensions
- Parallel Computing
Hi. Thanks. Yeah that "VEX.L must be zero" thing is left over from before we introduced AVX2 which added the 256b encoding for this instruction. I'll ask our doc team to fix this (and possibly others that might have similar origin).