RPP Log Tensor on HOST and HIP

include/rppt_tensor_arithmetic_operations.h

@@ -258,6 +258,40 @@ RppStatus rppt_magnitude_host(RppPtr_t srcPtr1, RppPtr_t srcPtr2, RpptDescPtr sr
 RppStatus rppt_magnitude_gpu(RppPtr_t srcPtr1, RppPtr_t srcPtr2, RpptDescPtr srcDescPtr, RppPtr_t dstPtr, RpptDescPtr dstDescPtr, RpptROIPtr roiTensorPtrSrc, RpptRoiType roiType, rppHandle_t rppHandle);
 #endif // GPU_SUPPORT
 
+/*! \brief Logarithm operation on HOST backend
+ * \details Computes Log to base e(natural log) of the input for a given ND Tensor.
+ *          Supports u8->f32, i8->f32, f16->f16 and f32->f32 datatypes.
+ *          Uses Absolute of input for log computation and uses nextafter() if input is 0 to avoid undefined result.
+ * \param [in] srcPtr source tensor in HOST memory
+ * \param [in] srcGenericDescPtr source tensor descriptor
+ * \param [out] dstPtr destination tensor in HOST memory
+ * \param [in] dstGenericDescPtr destination tensor descriptor
+ * \param [in] roiTensor values to represent dimensions of input tensor
+ * \param [in] rppHandle RPP HOST handle created with <tt>\ref rppCreateWithBatchSize()</tt>
+ * \return A <tt> \ref RppStatus</tt> enumeration.
+ * \retval RPP_SUCCESS Successful completion.
+ * \retval RPP_ERROR* Unsuccessful completion.
+ */
+RppStatus rppt_log_host(RppPtr_t srcPtr, RpptGenericDescPtr srcGenericDescPtr, RppPtr_t dstPtr, RpptGenericDescPtr dstGenericDescPtr, Rpp32u *roiTensor, rppHandle_t rppHandle);
+
+#ifdef GPU_SUPPORT
+/*! \brief Logarithm operation on HIP backend
+ * \details Computes Log to base e(natural log) of the input for a given ND Tensor.
+ *          Supports u8->f32, i8->f32, f16->f16 and f32->f32 datatypes.
+ *          Uses Absolute of input for log computation and uses nextafter() if input is 0 to avoid undefined result.
+ * \param [in] srcPtr source tensor in HIP memory
+ * \param [in] srcGenericDescPtr source tensor descriptor
+ * \param [out] dstPtr destination tensor in HIP memory
+ * \param [in] dstGenericDescPtr destination tensor descriptor
+ * \param [in] roiTensor values to represent dimensions of input tensor
+ * \param [in] rppHandle RPP HIP handle created with <tt>\ref rppCreateWithStreamAndBatchSize()</tt>
+ * \return A <tt> \ref RppStatus</tt> enumeration.
+ * \retval RPP_SUCCESS Successful completion.
+ * \retval RPP_ERROR* Unsuccessful completion.
+ */
+RppStatus rppt_log_gpu(RppPtr_t srcPtr, RpptGenericDescPtr srcGenericDescPtr, RppPtr_t dstPtr, RpptGenericDescPtr dstGenericDescPtr, Rpp32u *roiTensor, rppHandle_t rppHandle);
+#endif // GPU_SUPPORT
+
 /*! @}
  */
 

src/include/cpu/rpp_cpu_common.hpp

@@ -6501,4 +6501,10 @@ inline void compute_remap_src_loc(Rpp32f rowLoc, Rpp32f colLoc, Rpp32s &srcLoc,
     srcLoc = (rowLoc * stride) + colLoc * channels;
 }
 
+inline void compute_log_16_host(__m256 *p)
+{
+    p[0] = log_ps(p[0]);    // log compute
+    p[1] = log_ps(p[1]);    // log compute
+}
+
 #endif //RPP_CPU_COMMON_H

src/include/hip/rpp_hip_common.hpp

@@ -1869,6 +1869,21 @@ __device__ __forceinline__ float rpp_hip_math_sinc(float x)
     return (fabsf(x) < 1e-5f) ? (1.0f - x * x * ONE_OVER_6) : sinf(x) / x;
 }
 
+__device__ __forceinline__ void rpp_hip_math_log(d_float8 *src_f8, d_float8 *dst_f8)
+{
+    for(int i = 0; i < 8; i++)
+        src_f8->f1[i] = (!src_f8->f1[i]) ? std::nextafter(0.0f, 1.0f) : fabsf(src_f8->f1[i]);
+
+    dst_f8->f1[0] = __logf(src_f8->f1[0]);
+    dst_f8->f1[1] = __logf(src_f8->f1[1]);
+    dst_f8->f1[2] = __logf(src_f8->f1[2]);
+    dst_f8->f1[3] = __logf(src_f8->f1[3]);
+    dst_f8->f1[4] = __logf(src_f8->f1[4]);
+    dst_f8->f1[5] = __logf(src_f8->f1[5]);
+    dst_f8->f1[6] = __logf(src_f8->f1[6]);
+    dst_f8->f1[7] = __logf(src_f8->f1[7]);
+}
+
 // /******************** DEVICE RANDOMIZATION HELPER FUNCTIONS ********************/
 
 template<typename T>

src/modules/cpu/host_tensor_arithmetic_operations.hpp

@@ -30,5 +30,6 @@ SOFTWARE.
 #include "kernel/subtract_scalar.hpp"
 #include "kernel/multiply_scalar.hpp"
 #include "kernel/magnitude.hpp"
+#include "kernel/log.hpp"
 
 #endif // HOST_TENSOR_ARITHMETIC_OPERATIONS_HPP