mirrors/gnss-sdr
1
0
Fork 0
mirror of https://github.com/gnss-sdr/gnss-sdr synced 2025-02-22 14:00:13 +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

10
src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_16ic_x2_dot_prod_16ic.h
View File

@ -94,7 +94,7 @@ static inline void volk_gnsssdr_16ic_x2_dot_prod_16ic_a_sse2(lv_16sc_t* out, con
if (sse_iters > 0) if (sse_iters > 0)
{ {
__m128i a,b,c, c_sr, mask_imag, mask_real, real, imag, imag1,imag2, b_sl, a_sl, realcacc, imagcacc, result; __m128i a, b, c, c_sr, mask_imag, mask_real, real, imag, imag1, imag2, b_sl, a_sl, realcacc, imagcacc, result;
__VOLK_ATTR_ALIGNED(16) lv_16sc_t dotProductVector[4]; __VOLK_ATTR_ALIGNED(16) lv_16sc_t dotProductVector[4];
realcacc = _mm_setzero_si128(); realcacc = _mm_setzero_si128();
@ -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.
@ -171,7 +173,7 @@ static inline void volk_gnsssdr_16ic_x2_dot_prod_16ic_u_sse2(lv_16sc_t* out, con
if (sse_iters > 0) if (sse_iters > 0)
{ {
__m128i a,b,c, c_sr, mask_imag, mask_real, real, imag, imag1,imag2, b_sl, a_sl, realcacc, imagcacc, result; __m128i a, b, c, c_sr, mask_imag, mask_real, real, imag, imag1, imag2, b_sl, a_sl, realcacc, imagcacc, result;
__VOLK_ATTR_ALIGNED(16) lv_16sc_t dotProductVector[4]; __VOLK_ATTR_ALIGNED(16) lv_16sc_t dotProductVector[4];
realcacc = _mm_setzero_si128(); realcacc = _mm_setzero_si128();
@ -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.
@ -212,7 +216,7 @@ static inline void volk_gnsssdr_16ic_x2_dot_prod_16ic_u_sse2(lv_16sc_t* out, con
result = _mm_or_si128(realcacc, imagcacc); result = _mm_or_si128(realcacc, imagcacc);
_mm_storeu_si128((__m128i*)dotProductVector,result); // Store the results back into the dot product vector _mm_storeu_si128((__m128i*)dotProductVector, result); // Store the results back into the dot product vector
for (i = 0; i < 4; ++i) for (i = 0; i < 4; ++i)
{ {