12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394 |
- void func_name(void *buffers[], void *cl_arg) \
- { \
- unsigned i; \
- float *factor = (float *) cl_arg; \
- \
-
- \
- \
- struct starpu_vector_interface *vector = (struct starpu_vector_interface *) buffers[0]; \
- \
- \
- unsigned n = STARPU_VECTOR_GET_NX(vector); \
- \
-
- \
- float *val = (float *)STARPU_VECTOR_GET_PTR(vector); \
- \
- \
- for (i = 0; i < n; i++) \
- val[i] *= *factor; \
- }
- void func_name(void *buffers[], void *cl_arg) \
- { \
- float *vector = (float *) STARPU_VECTOR_GET_PTR(buffers[0]); \
- unsigned int n = STARPU_VECTOR_GET_NX(buffers[0]); \
- unsigned int n_iterations = n/4; \
- \
- __m128 *VECTOR = (__m128*) vector; \
- __m128 FACTOR STARPU_ATTRIBUTE_ALIGNED(16); \
- float factor = *(float *) cl_arg; \
- FACTOR = _mm_set1_ps(factor); \
- \
- unsigned int i; \
- for (i = 0; i < n_iterations; i++) \
- VECTOR[i] = _mm_mul_ps(FACTOR, VECTOR[i]); \
- \
- unsigned int remainder = n%4; \
- if (remainder != 0) \
- { \
- unsigned int start = 4 * n_iterations; \
- for (i = start; i < start+remainder; ++i) \
- { \
- vector[i] = factor * vector[i]; \
- } \
- } \
- }
|