# instructions. Several futher instruction sets are built on core
# %XMM support, without specific inter-dependencies. Additionally
# AMD has a special mis-alignment sub-mode.
- SSE: [SSE2, SSE3, SSSE3, SSE4A, MISALIGNSSE],
+ SSE: [SSE2, MISALIGNSSE],
# SSE2 was re-specified as core instructions for 64bit. Also ISA
# extensions dealing with vectors of integers are added here rather
# than to SSE.
- SSE2: [LM, AESNI, PCLMULQDQ, SHA],
+ SSE2: [SSE3, LM, AESNI, PCLMULQDQ, SHA],
- # SSE4.1 explicitly depends on SSE3 and SSSE3
- SSE3: [SSE4_1],
- SSSE3: [SSE4_1],
-
- # SSE4.2 explicitly depends on SSE4.1
+ # Other SSEn each depend on their predecessor versions.
+ SSE3: [SSSE3],
+ SSSE3: [SSE4_1, SSE4A],
SSE4_1: [SSE4_2],
# AMD specify no relationship between POPCNT and SSE4.2. Intel