x86 - Is it possible to vectorize a multiplication in VC++ without SSE4? -


I want to vector a multiplication operation. I tried to use _mm_mul_epi32 , but my CPU only has support for "MMX, SSE (1,2,3,3 s), EM64T" instruction.

Can someone please tell me I can use any other function?

It depends on the limit of your coefficient - this is 16 beats then SSE 4 (for example < There are several 16 x 16 bit multiple SSE instructions available before code> mm_madd_epi16 , mm_mulhi_epi16 , mm_mullo_epi16 , mm_mulhrs_epi16 , etc.).

If you need 32 bit operands but they are unsigned, then you can use mm_mul_epu32 .

Alternatively, you can find float conversion in float and SSE useful in SEE, and if this requires improvement through 4x SIM then cost can be justified.

Comments

Popular posts from this blog

qt - switch/case statement in C++ with a QString type -

python - sqlite3.OperationalError: near "REFERENCES": syntax error - foreign key creating -

Python's equivalent for Ruby's define_method? -