Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix undefined behaviour in dot product kernels #655

Merged
merged 3 commits into from
Oct 22, 2023
Merged

Fix undefined behaviour in dot product kernels #655

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
b9c4020
Fix undefined behaviour in dot product kernels
argilo Oct 15, 2023
db9b37c
Use an absolute tolerance to test the dot product kernels
argilo Oct 15, 2023
221138a
Always initialize returnValue
argilo Oct 22, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
120 changes: 50 additions & 70 deletions kernels/volk/volk_16i_32fc_dot_prod_32fc.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -147,8 +147,7 @@ static inline void volk_16i_32fc_dot_prod_32fc_u_sse(lv_32fc_t* result,
unsigned int number = 0;
const unsigned int eighthPoints = num_points / 8;

float res[2];
float *realpt = &res[0], *imagpt = &res[1];
lv_32fc_t returnValue = lv_cmake(0.0f, 0.0f);
const short* aPtr = input;
const float* bPtr = (float*)taps;

Expand Down Expand Up @@ -207,18 +206,17 @@ static inline void volk_16i_32fc_dot_prod_32fc_u_sse(lv_32fc_t* result,
_mm_store_ps(dotProductVector,
dotProdVal0); // Store the results back into the dot product vector

*realpt = dotProductVector[0];
*imagpt = dotProductVector[1];
*realpt += dotProductVector[2];
*imagpt += dotProductVector[3];
returnValue += lv_cmake(dotProductVector[0], dotProductVector[1]);
returnValue += lv_cmake(dotProductVector[2], dotProductVector[3]);

number = eighthPoints * 8;
for (; number < num_points; number++) {
*realpt += ((*aPtr) * (*bPtr++));
*imagpt += ((*aPtr++) * (*bPtr++));
returnValue += lv_cmake(aPtr[0] * bPtr[0], aPtr[0] * bPtr[1]);
aPtr += 1;
bPtr += 2;
}

*result = *(lv_32fc_t*)(&res[0]);
*result = returnValue;
}

#endif /*LV_HAVE_SSE && LV_HAVE_MMX*/
Expand All @@ -235,8 +233,7 @@ static inline void volk_16i_32fc_dot_prod_32fc_u_avx2_fma(lv_32fc_t* result,
unsigned int number = 0;
const unsigned int sixteenthPoints = num_points / 16;

float res[2];
float *realpt = &res[0], *imagpt = &res[1];
lv_32fc_t returnValue = lv_cmake(0.0f, 0.0f);
const short* aPtr = input;
const float* bPtr = (float*)taps;

Expand Down Expand Up @@ -294,22 +291,19 @@ static inline void volk_16i_32fc_dot_prod_32fc_u_avx2_fma(lv_32fc_t* result,
_mm256_store_ps(dotProductVector,
dotProdVal0); // Store the results back into the dot product vector

*realpt = dotProductVector[0];
*imagpt = dotProductVector[1];
*realpt += dotProductVector[2];
*imagpt += dotProductVector[3];
*realpt += dotProductVector[4];
*imagpt += dotProductVector[5];
*realpt += dotProductVector[6];
*imagpt += dotProductVector[7];
returnValue += lv_cmake(dotProductVector[0], dotProductVector[1]);
returnValue += lv_cmake(dotProductVector[2], dotProductVector[3]);
returnValue += lv_cmake(dotProductVector[4], dotProductVector[5]);
returnValue += lv_cmake(dotProductVector[6], dotProductVector[7]);

number = sixteenthPoints * 16;
for (; number < num_points; number++) {
*realpt += ((*aPtr) * (*bPtr++));
*imagpt += ((*aPtr++) * (*bPtr++));
returnValue += lv_cmake(aPtr[0] * bPtr[0], aPtr[0] * bPtr[1]);
aPtr += 1;
bPtr += 2;
}

*result = *(lv_32fc_t*)(&res[0]);
*result = returnValue;
}

#endif /*LV_HAVE_AVX2 && lV_HAVE_FMA*/
Expand All @@ -326,8 +320,7 @@ static inline void volk_16i_32fc_dot_prod_32fc_u_avx2(lv_32fc_t* result,
unsigned int number = 0;
const unsigned int sixteenthPoints = num_points / 16;

float res[2];
float *realpt = &res[0], *imagpt = &res[1];
lv_32fc_t returnValue = lv_cmake(0.0f, 0.0f);
const short* aPtr = input;
const float* bPtr = (float*)taps;

Expand Down Expand Up @@ -391,22 +384,19 @@ static inline void volk_16i_32fc_dot_prod_32fc_u_avx2(lv_32fc_t* result,
_mm256_store_ps(dotProductVector,
dotProdVal0); // Store the results back into the dot product vector

*realpt = dotProductVector[0];
*imagpt = dotProductVector[1];
*realpt += dotProductVector[2];
*imagpt += dotProductVector[3];
*realpt += dotProductVector[4];
*imagpt += dotProductVector[5];
*realpt += dotProductVector[6];
*imagpt += dotProductVector[7];
returnValue += lv_cmake(dotProductVector[0], dotProductVector[1]);
returnValue += lv_cmake(dotProductVector[2], dotProductVector[3]);
returnValue += lv_cmake(dotProductVector[4], dotProductVector[5]);
returnValue += lv_cmake(dotProductVector[6], dotProductVector[7]);

number = sixteenthPoints * 16;
for (; number < num_points; number++) {
*realpt += ((*aPtr) * (*bPtr++));
*imagpt += ((*aPtr++) * (*bPtr++));
returnValue += lv_cmake(aPtr[0] * bPtr[0], aPtr[0] * bPtr[1]);
aPtr += 1;
bPtr += 2;
}

*result = *(lv_32fc_t*)(&res[0]);
*result = returnValue;
}

#endif /*LV_HAVE_AVX2*/
Expand All @@ -424,8 +414,7 @@ static inline void volk_16i_32fc_dot_prod_32fc_a_sse(lv_32fc_t* result,
unsigned int number = 0;
const unsigned int eighthPoints = num_points / 8;

float res[2];
float *realpt = &res[0], *imagpt = &res[1];
lv_32fc_t returnValue = lv_cmake(0.0f, 0.0f);
const short* aPtr = input;
const float* bPtr = (float*)taps;

Expand Down Expand Up @@ -484,18 +473,17 @@ static inline void volk_16i_32fc_dot_prod_32fc_a_sse(lv_32fc_t* result,
_mm_store_ps(dotProductVector,
dotProdVal0); // Store the results back into the dot product vector

*realpt = dotProductVector[0];
*imagpt = dotProductVector[1];
*realpt += dotProductVector[2];
*imagpt += dotProductVector[3];
returnValue += lv_cmake(dotProductVector[0], dotProductVector[1]);
returnValue += lv_cmake(dotProductVector[2], dotProductVector[3]);

number = eighthPoints * 8;
for (; number < num_points; number++) {
*realpt += ((*aPtr) * (*bPtr++));
*imagpt += ((*aPtr++) * (*bPtr++));
returnValue += lv_cmake(aPtr[0] * bPtr[0], aPtr[0] * bPtr[1]);
aPtr += 1;
bPtr += 2;
}

*result = *(lv_32fc_t*)(&res[0]);
*result = returnValue;
}

#endif /*LV_HAVE_SSE && LV_HAVE_MMX*/
Expand All @@ -511,8 +499,7 @@ static inline void volk_16i_32fc_dot_prod_32fc_a_avx2(lv_32fc_t* result,
unsigned int number = 0;
const unsigned int sixteenthPoints = num_points / 16;

float res[2];
float *realpt = &res[0], *imagpt = &res[1];
lv_32fc_t returnValue = lv_cmake(0.0f, 0.0f);
const short* aPtr = input;
const float* bPtr = (float*)taps;

Expand Down Expand Up @@ -576,22 +563,19 @@ static inline void volk_16i_32fc_dot_prod_32fc_a_avx2(lv_32fc_t* result,
_mm256_store_ps(dotProductVector,
dotProdVal0); // Store the results back into the dot product vector

*realpt = dotProductVector[0];
*imagpt = dotProductVector[1];
*realpt += dotProductVector[2];
*imagpt += dotProductVector[3];
*realpt += dotProductVector[4];
*imagpt += dotProductVector[5];
*realpt += dotProductVector[6];
*imagpt += dotProductVector[7];
returnValue += lv_cmake(dotProductVector[0], dotProductVector[1]);
returnValue += lv_cmake(dotProductVector[2], dotProductVector[3]);
returnValue += lv_cmake(dotProductVector[4], dotProductVector[5]);
returnValue += lv_cmake(dotProductVector[6], dotProductVector[7]);

number = sixteenthPoints * 16;
for (; number < num_points; number++) {
*realpt += ((*aPtr) * (*bPtr++));
*imagpt += ((*aPtr++) * (*bPtr++));
returnValue += lv_cmake(aPtr[0] * bPtr[0], aPtr[0] * bPtr[1]);
aPtr += 1;
bPtr += 2;
}

*result = *(lv_32fc_t*)(&res[0]);
*result = returnValue;
}


Expand All @@ -608,8 +592,7 @@ static inline void volk_16i_32fc_dot_prod_32fc_a_avx2_fma(lv_32fc_t* result,
unsigned int number = 0;
const unsigned int sixteenthPoints = num_points / 16;

float res[2];
float *realpt = &res[0], *imagpt = &res[1];
lv_32fc_t returnValue = lv_cmake(0.0f, 0.0f);
const short* aPtr = input;
const float* bPtr = (float*)taps;

Expand Down Expand Up @@ -667,22 +650,19 @@ static inline void volk_16i_32fc_dot_prod_32fc_a_avx2_fma(lv_32fc_t* result,
_mm256_store_ps(dotProductVector,
dotProdVal0); // Store the results back into the dot product vector

*realpt = dotProductVector[0];
*imagpt = dotProductVector[1];
*realpt += dotProductVector[2];
*imagpt += dotProductVector[3];
*realpt += dotProductVector[4];
*imagpt += dotProductVector[5];
*realpt += dotProductVector[6];
*imagpt += dotProductVector[7];
returnValue += lv_cmake(dotProductVector[0], dotProductVector[1]);
returnValue += lv_cmake(dotProductVector[2], dotProductVector[3]);
returnValue += lv_cmake(dotProductVector[4], dotProductVector[5]);
returnValue += lv_cmake(dotProductVector[6], dotProductVector[7]);

number = sixteenthPoints * 16;
for (; number < num_points; number++) {
*realpt += ((*aPtr) * (*bPtr++));
*imagpt += ((*aPtr++) * (*bPtr++));
returnValue += lv_cmake(aPtr[0] * bPtr[0], aPtr[0] * bPtr[1]);
aPtr += 1;
bPtr += 2;
}

*result = *(lv_32fc_t*)(&res[0]);
*result = returnValue;
}


Expand Down
Loading
Loading