Best way to calculate max sse var mask

Question

(I'm only interested in components of 1st 3)

For instance:
[ 1 2 3 ? ] [ 1 2 3 ? ] should produce [ 0 0 -1 ? ] [ 0 0 -1 ? ]

In addition, it is important to have only one “bit” so that: [ 1 2 2 ? ] [ 1 2 2 ? ] should not create [ 0 -1 -1 ? ] [ 0 -1 -1 ? ]
but most likely [ 0 -1 0 ? ] [ 0 -1 0 ? ] or [ 0 0 -1 ? ] [ 0 0 -1 ? ] (doesn't matter which one)

A later (bad) solution is possible, for example, by extracting a horizontal maximum and comparing with the original:

 __m128 abcd; // input __m128 ccac = _mm_shuffle_ps(abcd, abcd, 0x8A); __m128 abcd_ccac = _mm_max_ps(abcd, ccac); __m128 babb = _mm_shuffle_ps(abcd, abcd, 0x51); __m128 abcd_ccac_babb = _mm_max_ps(abcd_ccac, babb); __m128 mask = _mm_cmpeq_ps(abcd, abcd_ccac_babb); 

Perhaps some bitwise operations to get rid of duplicate set bits?

Update:

Follow-up, I made another (bad) decision.
The key is to compare each component with another, avoiding the equations of equality (not having a >= b and b >= a in another place).

a > b & a >= c
b > c & b >= a
c > a & c >= b

To obtain:

([ abc ? ] > [ bca ? ]) & ([ abc ? ] >= [ cab ? ])

and in code:

 __m128 abcd; // input __m128 bcad = _mm_shuffle_ps(abcd, abcd, 0xC9); __m128 gt = _mm_cmpgt_ps(abcd, bcad); __m128 cabd = _mm_shuffle_ps(abcd, abcd, 0xD2); __m128 ge = _mm_cmpge_ps(abcd, cabd); __m128 mask = _mm_and_ps(gt, ge); 

In the case of [ xxx ? ] [ xxx ? ] it fails (returns [ 0 0 0 ? ] ).

Close :-)

Any ideas?

Update:

Now I am using the following solution:

 __m128 abcd; // input __m128 bcad = _mm_shuffle_ps(abcd, abcd, 0xC9); __m128 gt = _mm_cmpgt_ps(abcd, bcad); __m128 cabd = _mm_shuffle_ps(abcd, abcd, 0xD2); __m128 ge = _mm_cmpge_ps(abcd, cabd); __m128 and = _mm_and_ps(gt, ge); __m128i bits = _mm_setr_epi32(_mm_movemask_ps(and), -1, -1, -1); __m128i dirt = _mm_cmpeq_epi32(bits, _mm_setzero_si128()); __m128i mask = _mm_or_si128(dirt, _mm_castps_si128(and)); 

Answer 1

I have not tested this, but hopefully it will get -1 in only the first (highest order) occurrence of the maximum value:

 __m128 abcd; // input __m128 ccac = _mm_shuffle_ps( abcd,abcd,0x8A ); __m128 abcd_ccac = _mm_max_ps( abcd,ccac ); __m128 babb = _mm_shuffle_ps( abcd,abcd,0x51 ); __m128 abcd_ccac_babb = _mm_max_ps( abcd_ccac,babb ); __m128 mask = _mm_cmpeq_ps( abcd,abcd_ccac_babb ); // set the '?' position in mask to zero mask = _mm_blend_ps( mask,_mm_setzero_ps(),0x08 ); // shift mask left 32 bits shifting in zeros __m128 maskSrl32 = _mm_shuffle_ps( mask,mask,_MM_SHUFFLE( 3,1,0,3 ) ); // shift mask left 64 bits shifting in zeros __m128 maskSrl64 = _mm_shuffle_ps( mask,mask,_MM_SHUFFLE( 3,0,3,3 ) ); // andnot the shifted masks with mask // in doing so, the higher order set bits will suppress any set bits which follow mask = _mm_andnot_ps( maskSrl32,mask ); mask = _mm_andnot_ps( maskSrl64,mask ); // select -1 using the final mask __m128 result = _mm_and_ps( mask,_mm_set1_ps( -1.0f ) ); 

Reverse the shift direction to get -1 at max. smallest line items.