/** Documented at declaration */
int
gpujpeg_table_quantization_encoder_init(struct gpujpeg_table_quantization* table, enum gpujpeg_component_type type, int quality)
{
// Load raw table in zig-zag order
gpujpeg_table_quantization_set_default(table->table_raw, type);
// Update raw table by quality
nppiQuantFwdRawTableInit_JPEG_8u(table->table_raw, quality);
// Fix NPP bug before version 4.1 [http://forums.nvidia.com/index.php?showtopic=191896]
const NppLibraryVersion* npp_version = nppGetLibVersion();
if ( npp_version->major < 4 || (npp_version->major == 4 && npp_version->minor == 0) ) {
for ( int i = 0; i < 64; i++ ) {
table->table[gpujpeg_order_natural[i]] = ((1 << 15) / (double)table->table_raw[i]) + 0.5;
}
} else {
// Load forward table from raw table
nppiQuantFwdTableInit_JPEG_8u16u(table->table_raw, table->table);
}
// Copy tables to device memory
if ( cudaSuccess != cudaMemcpy(table->d_table, table->table, 64 * sizeof(uint16_t), cudaMemcpyHostToDevice) )
return -1;
return 0;
}
void printfNPPinfo(int argc, char *argv[])
{
const char *sComputeCap[] =
{
"No CUDA Capable Device Found",
"Compute 1.0", "Compute 1.1", "Compute 1.2", "Compute 1.3",
"Compute 2.0", "Compute 2.1", "Compute 3.0", "Compute 3.5", NULL
};
const NppLibraryVersion *libVer = nppGetLibVersion();
NppGpuComputeCapability computeCap = nppGetGpuComputeCapability();
printf("NPP Library Version %d.%d.%d\n", libVer->major, libVer->minor, libVer->build);
if (computeCap != 0 && g_nDevice == -1)
{
printf("%s using GPU <%s> with %d SM(s) with", argv[0], nppGetGpuName(), nppGetGpuNumSMs());
if (computeCap > 0)
{
printf(" %s\n", sComputeCap[computeCap]);
}
else
{
printf(" Unknown Compute Capabilities\n");
}
}
else
{
printf("%s\n", sComputeCap[computeCap]);
}
}
bool printfNPPinfo(int argc, char *argv[], int cudaVerMajor, int cudaVerMinor)
{
const NppLibraryVersion *libVer = nppGetLibVersion();
printf("NPP Library Version %d.%d.%d\n", libVer->major, libVer->minor, libVer->build);
int driverVersion, runtimeVersion;
cudaDriverGetVersion(&driverVersion);
cudaRuntimeGetVersion(&runtimeVersion);
printf(" CUDA Driver Version: %d.%d\n", driverVersion/1000, (driverVersion%100)/10);
printf(" CUDA Runtime Version: %d.%d\n", runtimeVersion/1000, (runtimeVersion%100)/10);
bool bVal = checkCudaCapabilities(cudaVerMajor, cudaVerMinor);
return bVal;
}
int
gpujpeg_table_quantization_decoder_compute(struct gpujpeg_table_quantization* table)
{
// Fix NPP bug before version 4.1 [http://forums.nvidia.com/index.php?showtopic=191896]
const NppLibraryVersion* npp_version = nppGetLibVersion();
if ( npp_version->major < 4 || (npp_version->major == 4 && npp_version->minor == 0) ) {
for ( int i = 0; i < 64; i++ ) {
table->table[gpujpeg_order_natural[i]] = table->table_raw[i];
}
} else {
// Load inverse table from raw table
nppiQuantInvTableInit_JPEG_8u16u(table->table_raw, table->table);
}
// Copy tables to device memory
if ( cudaSuccess != cudaMemcpy(table->d_table, table->table, 64 * sizeof(uint16_t), cudaMemcpyHostToDevice) )
return -1;
return 0;
}
