topic Hi, in Intel® Integrated Performance Primitives

sing of 32 float

needle__brosh — Tue, 29 Nov 2016 07:38:21 GMT

Hi,

I need to get the sign value of each element in a vector.

is there a simple way to do so.

exp:

input: (32float) 1.26 0 63.54 -8.3 -1.2 1.6 5 -9 ....

output: 1 1 1 -1 -1 1 1 -1

thanks

Chao_Y_Intel — Tue, 29 Nov 2016 07:57:33 GMT

Hello,

Can you check the ippsThreshold_LTValGTVal function?

Thanks,
Chao

Zhen_Z_Intel — Tue, 29 Nov 2016 08:08:05 GMT

Dear customer,

ippsThreshold_LT_32s(pSrc, pDst, len, -1); //replace all value with -1, if <-1
ippsThreshold_GT_32s(pSrc, pDst, len, 1); //otherwise set to 1

Hope it would be useful to you.

Best regards,
Fiona

Igor_A_Intel — Tue, 29 Nov 2016 10:18:23 GMT

Hi,

another, may be the fastest, method is to use ippsRShiftC functions - just cast pointers as (Ipp32s*) and shift on 31:

IPPAPI(IppStatus, ippsRShiftC_32s, (const Ipp32s* pSrc, int val, Ipp32s* pDst, int len))

and then in-place ippsOrC with 1 (cast pointers as Ipp32u):

IPPAPI(IppStatus, ippsOrC_32u_I, ( Ipp32u val, Ipp32u* pSrcDst, int len))

regards, Igor