mirrors
/
gnss-sdr
mirror of https://github.com/gnss-sdr/gnss-sdr
1
0
Fork 0
Code Issues Releases Wiki Activity

Fix typo

This commit is contained in:
Carles Fernandez 2019-03-19 20:39:23 +01:00
parent c8d27eb97c
commit 10d73da839
No known key found for this signature in database
GPG Key ID: 4C583C52B0C3877D
9 changed files with 257 additions and 251 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/algorithms/tracking/libs/cpu_multicorrelator.cc
View File

@ -1,11 +1,11 @@
/*!
* \file cpu_multicorrelator.cc
* \brief High optimized CPU vector multiTAP correlator class
* \brief Highly optimized CPU vector multiTAP correlator class
* \authors <ul>
* <li> Javier Arribas, 2015. jarribas(at)cttc.es
* </ul>
*
* Class that implements a high optimized vector multiTAP correlator class for CPUs
* Class that implements a highly optimized vector multiTAP correlator class for CPUs
*
* -------------------------------------------------------------------------
*

2
src/algorithms/tracking/libs/cpu_multicorrelator.h
View File

@ -65,4 +65,4 @@ private:
};
#endif /* CPU_MULTICORRELATOR_H_ */
#endif /* GNSS_SDR_CPU_MULTICORRELATOR_H_ */

4
src/algorithms/tracking/libs/cpu_multicorrelator_16sc.cc
View File

@ -1,11 +1,11 @@
/*!
* \file cpu_multicorrelator_16sc.cc
* \brief High optimized CPU vector multiTAP correlator class
* \brief Highly optimized CPU vector multiTAP correlator class
* \authors <ul>
* <li> Javier Arribas, 2015. jarribas(at)cttc.es
* </ul>
*
* Class that implements a high optimized vector multiTAP correlator class for CPUs
* Class that implements a highly optimized vector multiTAP correlator class for CPUs
*
* -------------------------------------------------------------------------
*

4
src/algorithms/tracking/libs/cpu_multicorrelator_16sc.h
View File

@ -1,11 +1,11 @@
/*!
* \file cpu_multicorrelator_16sc.h
* \brief High optimized CPU vector multiTAP correlator class for lv_16sc_t (short int complex)
* \brief Highly optimized CPU vector multiTAP correlator class for lv_16sc_t (short int complex)
* \authors <ul>
* <li> Javier Arribas, 2016. jarribas(at)cttc.es
* </ul>
*
* Class that implements a high optimized vector multiTAP correlator class for CPUs
* Class that implements a highly optimized vector multiTAP correlator class for CPUs
*
* -------------------------------------------------------------------------
*

2
src/algorithms/tracking/libs/cpu_multicorrelator_real_codes.cc
View File

@ -6,7 +6,7 @@
* <li> Cillian O'Driscoll, 2017. cillian.odriscoll(at)gmail.com
* </ul>
*
* Class that implements a high optimized vector multiTAP correlator class for CPUs
* Class that implements a highly optimized vector multiTAP correlator class for CPUs
*
* -------------------------------------------------------------------------
*

2
src/algorithms/tracking/libs/cpu_multicorrelator_real_codes.h
View File

@ -6,7 +6,7 @@
* <li> Cillian O'Driscoll, 2017, cillian.odriscoll(at)gmail.com
* </ul>
*
* Class that implements a high optimized vector multiTAP correlator class for CPUs
* Class that implements a highly optimized vector multiTAP correlator class for CPUs
*
* -------------------------------------------------------------------------
*

61
src/algorithms/tracking/libs/cuda_multicorrelator.cu
View File

@ -1,11 +1,11 @@
/*!
* \file cuda_multicorrelator.cu
* \brief High optimized CUDA GPU vector multiTAP correlator class
* \brief Highly optimized CUDA GPU vector multiTAP correlator class
* \authors <ul>
* <li> Javier Arribas, 2015. jarribas(at)cttc.es
* </ul>
*
* Class that implements a high optimized vector multiTAP correlator class for NVIDIA CUDA GPUs
* Class that implements a highly optimized vector multiTAP correlator class for NVIDIA CUDA GPUs
*
* -------------------------------------------------------------------------
*
@ -33,9 +33,8 @@
*/
#include "cuda_multicorrelator.h"
#include <stdio.h>
#include <iostream>
#include <stdio.h>
// For the CUDA runtime routines (prefixed with "cuda_")
#include <cuda_runtime.h>
@ -53,8 +52,7 @@ __global__ void Doppler_wippe_scalarProdGPUCPXxN_shifts_chips(
int vectorN,
int elementN,
float rem_carrier_phase_in_rad,
float phase_step_rad
)
float phase_step_rad)
{
//Accumulators cache
__shared__ GPU_Complex accumResult[ACCUM_N];
@ -90,7 +88,8 @@ __global__ void Doppler_wippe_scalarProdGPUCPXxN_shifts_chips(
for (int iAccum = threadIdx.x; iAccum < ACCUM_N; iAccum += blockDim.x)
{
GPU_Complex sum = GPU_Complex(0, 0);
float local_code_chip_index=0.0;;
float local_code_chip_index = 0.0;
;
//float code_phase;
for (int pos = iAccum; pos < elementN; pos += ACCUM_N)
{
@ -109,7 +108,6 @@ __global__ void Doppler_wippe_scalarProdGPUCPXxN_shifts_chips(
//printf("vec= %i, pos %i, chip_idx=%i chip_shift=%f \r\n",vec, pos,__float2int_rd(local_code_chip_index),local_code_chip_index);
// 2.correlate
sum.multiply_acc(d_sig_wiped[pos], d_local_code_in[__float2int_rd(local_code_chip_index)]);
}
accumResult[iAccum] = sum;
}
@ -135,23 +133,26 @@ __global__ void Doppler_wippe_scalarProdGPUCPXxN_shifts_chips(
}
}
bool cuda_multicorrelator::init_cuda_integrated_resampler(
int signal_length_samples,
int code_length_chips,
int n_correlators
)
int n_correlators)
{
// use command-line specified CUDA device, otherwise use device with highest Gflops/s
// findCudaDevice(argc, (const char **)argv);
cudaDeviceProp prop;
int num_devices, device;
cudaGetDeviceCount(&num_devices);
if (num_devices > 1) {
if (num_devices > 1)
{
int max_multiprocessors = 0, max_device = 0;
for (device = 0; device < num_devices; device++) {
for (device = 0; device < num_devices; device++)
{
cudaDeviceProp properties;
cudaGetDeviceProperties(&properties, device);
if (max_multiprocessors < properties.multiProcessorCount) {
if (max_multiprocessors < properties.multiProcessorCount)
{
max_multiprocessors = properties.multiProcessorCount;
max_device = device;
}
@ -165,7 +166,8 @@ bool cuda_multicorrelator::init_cuda_integrated_resampler(
cudaGetDeviceProperties(&prop, max_device);
//debug code
if (prop.canMapHostMemory != 1) {
if (prop.canMapHostMemory != 1)
{
printf("Device can not map memory.\n");
}
printf("L2 Cache size= %u \n", prop.l2CacheSize);
@ -174,11 +176,14 @@ bool cuda_multicorrelator::init_cuda_integrated_resampler(
printf("sharedMemPerBlock= %lu \n", prop.sharedMemPerBlock);
printf("deviceOverlap= %i \n", prop.deviceOverlap);
printf("multiProcessorCount= %i \n", prop.multiProcessorCount);
}else{
}
else
{
cudaGetDevice(&selected_gps_device);
cudaGetDeviceProperties(&prop, selected_gps_device);
//debug code
if (prop.canMapHostMemory != 1) {
if (prop.canMapHostMemory != 1)
{
printf("Device can not map memory.\n");
}
@ -235,14 +240,13 @@ bool cuda_multicorrelator::init_cuda_integrated_resampler(
return true;
}
bool cuda_multicorrelator::set_local_code_and_taps(
int code_length_chips,
const std::complex<float> *local_codes_in,
float *shifts_chips,
int n_correlators
)
int n_correlators)
{
cudaSetDevice(selected_gps_device);
//********* ZERO COPY VERSION ************
// // Get device pointer from host memory. No allocation or memcpy
@ -272,12 +276,11 @@ bool cuda_multicorrelator::set_local_code_and_taps(
return true;
}
bool cuda_multicorrelator::set_input_output_vectors(
std::complex<float> *corr_out,
std::complex<float>* sig_in
)
std::complex<float> *sig_in)
{
cudaSetDevice(selected_gps_device);
// Save CPU pointers
d_sig_in_cpu = sig_in;
@ -293,10 +296,12 @@ bool cuda_multicorrelator::set_input_output_vectors(
printf("cuda cudaHostGetDevicePointer error \r\n");
}
return true;
}
#define gpuErrchk(ans) { gpuAssert((ans), __FILE__, __LINE__); }
#define gpuErrchk(ans) \
{ \
gpuAssert((ans), __FILE__, __LINE__); \
}
inline void gpuAssert(cudaError_t code, const char *file, int line, bool abort = true)
{
if (code != cudaSuccess)
@ -306,6 +311,7 @@ inline void gpuAssert(cudaError_t code, const char *file, int line, bool abort=t
}
}
bool cuda_multicorrelator::Carrier_wipeoff_multicorrelator_resampler_cuda(
float rem_carrier_phase_in_rad,
float phase_step_rad,
@ -314,7 +320,6 @@ bool cuda_multicorrelator::Carrier_wipeoff_multicorrelator_resampler_cuda(
int signal_length_samples,
int n_correlators)
{
cudaSetDevice(selected_gps_device);
// cudaMemCpy version
//size_t memSize = signal_length_samples * sizeof(std::complex<float>);
@ -337,8 +342,7 @@ bool cuda_multicorrelator::Carrier_wipeoff_multicorrelator_resampler_cuda(
n_correlators,
signal_length_samples,
rem_carrier_phase_in_rad,
phase_step_rad
);
phase_step_rad);
gpuErrchk(cudaPeekAtLastError());
gpuErrchk(cudaStreamSynchronize(stream1));
@ -352,6 +356,7 @@ bool cuda_multicorrelator::Carrier_wipeoff_multicorrelator_resampler_cuda(
return true;
}
cuda_multicorrelator::cuda_multicorrelator()
{
d_sig_in = NULL;
@ -366,6 +371,7 @@ cuda_multicorrelator::cuda_multicorrelator()
d_code_length_chips = 0;
}
bool cuda_multicorrelator::free_cuda()
{
// Free device global memory
@ -385,4 +391,3 @@ bool cuda_multicorrelator::free_cuda()
cudaDeviceReset();
return true;
}

6
src/algorithms/tracking/libs/cuda_multicorrelator.h
View File

@ -1,11 +1,11 @@
/*!
* \file cuda_multicorrelator.h
* \brief High optimized CUDA GPU vector multiTAP correlator class
* \brief Highly optimized CUDA GPU vector multiTAP correlator class
* \authors <ul>
* <li> Javier Arribas, 2015. jarribas(at)cttc.es
* </ul>
*
* Class that implements a high optimized vector multiTAP correlator class for NVIDIA CUDA GPUs
* Class that implements a highly optimized vector multiTAP correlator class for NVIDIA CUDA GPUs
*
* -------------------------------------------------------------------------
*
@ -92,6 +92,7 @@ struct GPU_Complex
}
};
struct GPU_Complex_Short
{
float r;
@ -149,7 +150,6 @@ private:
GPU_Complex* d_local_codes_in;
GPU_Complex* d_corr_out;
//
std::complex<float>* d_sig_in_cpu;
std::complex<float>* d_corr_out_cpu;

3
src/algorithms/tracking/libs/tracking_discriminators.cc
View File

@ -36,6 +36,7 @@
#include <cmath>
// All the outputs are in RADIANS
/*
* FLL four quadrant arctan discriminator:
* \f{equation}
@ -45,7 +46,6 @@
* \f$I_{PS1},Q_{PS1}\f$ are the inphase and quadrature prompt correlator outputs respectively at sample time \f$t_1\f$, and
* \f$I_{PS2},Q_{PS2}\f$ are the inphase and quadrature prompt correlator outputs respectively at sample time \f$t_2\f$. The output is in [radians/second].
*/
double fll_four_quadrant_atan(gr_complex prompt_s1, gr_complex prompt_s2, double t1, double t2)
{
double cross, dot;
@ -105,6 +105,7 @@ double dll_nc_e_minus_l_normalized(gr_complex early_s1, gr_complex late_s1)
return 0.5 * (P_early - P_late) / (P_early + P_late);
}
/*
* DLL Noncoherent Very Early Minus Late Power (VEMLP) normalized discriminator, using the outputs
* of four correlators, Very Early (VE), Early (E), Late (L) and Very Late (VL):