mirrors/gnss-sdr
1
0
Fork 0
mirror of https://github.com/gnss-sdr/gnss-sdr synced 2025-07-07 04:22:56 +00:00
Code Issues Releases Wiki Activity

prefetching data

Browse Source
This commit is contained in:
Carles Fernandez 2016-03-12 12:22:26 +01:00
parent 268f298fad
commit a93a01e9b1
1 changed files with 7 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_16ic_x2_dot_prod_16ic.h
View File

@ -109,7 +109,9 @@ static inline void volk_gnsssdr_16ic_x2_dot_prod_16ic_a_sse2(lv_16sc_t* out, con
//imaginery part -> reinterpret_cast<cv T*>(a)[2*i + 1] //imaginery part -> reinterpret_cast<cv T*>(a)[2*i + 1]
// a[127:0]=[a3.i,a3.r,a2.i,a2.r,a1.i,a1.r,a0.i,a0.r] // a[127:0]=[a3.i,a3.r,a2.i,a2.r,a1.i,a1.r,a0.i,a0.r]
a = _mm_load_si128((__m128i*)_in_a); //load (2 byte imag, 2 byte real) x 4 into 128 bits reg a = _mm_load_si128((__m128i*)_in_a); //load (2 byte imag, 2 byte real) x 4 into 128 bits reg
__builtin_prefetch(_in_a + 8);
b = _mm_load_si128((__m128i*)_in_b); b = _mm_load_si128((__m128i*)_in_b);
__builtin_prefetch(_in_b + 8);
c = _mm_mullo_epi16(a, b); // a3.i*b3.i, a3.r*b3.r, .... c = _mm_mullo_epi16(a, b); // a3.i*b3.i, a3.r*b3.r, ....
c_sr = _mm_srli_si128(c, 2); // Shift a right by imm8 bytes while shifting in zeros, and store the results in dst. c_sr = _mm_srli_si128(c, 2); // Shift a right by imm8 bytes while shifting in zeros, and store the results in dst.
@ -186,7 +188,9 @@ static inline void volk_gnsssdr_16ic_x2_dot_prod_16ic_u_sse2(lv_16sc_t* out, con
//imaginery part -> reinterpret_cast<cv T*>(a)[2*i + 1] //imaginery part -> reinterpret_cast<cv T*>(a)[2*i + 1]
// a[127:0]=[a3.i,a3.r,a2.i,a2.r,a1.i,a1.r,a0.i,a0.r] // a[127:0]=[a3.i,a3.r,a2.i,a2.r,a1.i,a1.r,a0.i,a0.r]
a = _mm_loadu_si128((__m128i*)_in_a); //load (2 byte imag, 2 byte real) x 4 into 128 bits reg a = _mm_loadu_si128((__m128i*)_in_a); //load (2 byte imag, 2 byte real) x 4 into 128 bits reg
__builtin_prefetch(_in_a + 8);
b = _mm_loadu_si128((__m128i*)_in_b); b = _mm_loadu_si128((__m128i*)_in_b);
__builtin_prefetch(_in_b + 8);
c = _mm_mullo_epi16(a, b); // a3.i*b3.i, a3.r*b3.r, .... c = _mm_mullo_epi16(a, b); // a3.i*b3.i, a3.r*b3.r, ....
c_sr = _mm_srli_si128(c, 2); // Shift a right by imm8 bytes while shifting in zeros, and store the results in dst. c_sr = _mm_srli_si128(c, 2); // Shift a right by imm8 bytes while shifting in zeros, and store the results in dst.