mirror of
https://github.com/claunia/cuetools.net.git
synced 2025-12-16 18:14:25 +00:00
experiment with Latice LPC algorithm
This commit is contained in:
chudov
@@ -198,7 +198,7 @@ extern "C" __global__ void cudaComputeLPC(
|
||||
volatile float gen1[32];
|
||||
volatile float parts[128];
|
||||
//volatile float reff[32];
|
||||
int cbits;
|
||||
//int cbits;
|
||||
} shared;
|
||||
const int tid = threadIdx.x;
|
||||
|
||||
@@ -280,6 +280,170 @@ extern "C" __global__ void cudaComputeLPC(
|
||||
}
|
||||
}
|
||||
|
||||
#define SUM256(buf,tid) if (tid < 128) buf[tid] += buf[tid + 128]; __syncthreads(); \
|
||||
if (tid < 64) buf[tid] += buf[tid + 64]; __syncthreads(); \
|
||||
if (tid < 32) { \
|
||||
buf[tid] += buf[tid + 32]; buf[tid] += buf[tid + 16]; buf[tid] += buf[tid + 8]; \
|
||||
buf[tid] += buf[tid + 4]; buf[tid] += buf[tid + 2]; buf[tid] += buf[tid + 1]; \
|
||||
}
|
||||
#define FSQR(s) ((s)*(s))
|
||||
|
||||
extern "C" __global__ void cudaComputeLPCLattice(
|
||||
encodeResidualTaskStruct *tasks,
|
||||
const int taskCount, // tasks per block
|
||||
const int *samples,
|
||||
const int frameSize, // <= 512
|
||||
const int max_order // should be <= 32
|
||||
)
|
||||
{
|
||||
__shared__ struct {
|
||||
encodeResidualTaskStruct task;
|
||||
volatile float F[512];
|
||||
volatile float B[512];
|
||||
volatile float tmp[256];
|
||||
volatile float arp[32];
|
||||
volatile float rc[32];
|
||||
volatile int bits[32];
|
||||
volatile float PE[33];
|
||||
} shared;
|
||||
|
||||
// fetch task data
|
||||
if (threadIdx.x < sizeof(shared.task) / sizeof(int))
|
||||
((int*)&shared.task)[threadIdx.x] = ((int*)(tasks + taskCount * blockIdx.y))[threadIdx.x];
|
||||
__syncthreads();
|
||||
|
||||
// F = samples; B = samples;
|
||||
shared.F[threadIdx.x] = threadIdx.x < frameSize ? samples[shared.task.samplesOffs + threadIdx.x] >> shared.task.wbits : 0.0f;
|
||||
shared.F[threadIdx.x + 256] = threadIdx.x + 256 < frameSize ? samples[shared.task.samplesOffs + threadIdx.x + 256] >> shared.task.wbits : 0.0f;
|
||||
shared.B[threadIdx.x] = shared.F[threadIdx.x];
|
||||
shared.B[threadIdx.x + 256] = shared.F[threadIdx.x + 256];
|
||||
__syncthreads();
|
||||
|
||||
// DEN = F*F'
|
||||
shared.tmp[threadIdx.x] = FSQR(shared.F[threadIdx.x]) + FSQR(shared.F[threadIdx.x + 256]);
|
||||
__syncthreads();
|
||||
SUM256(shared.tmp,threadIdx.x);
|
||||
__syncthreads();
|
||||
float DEN = shared.tmp[0];
|
||||
|
||||
// PE = [DEN./nn,zeros(lr,max_order)];
|
||||
if (threadIdx.x < 32)
|
||||
shared.PE[threadIdx.x+1] = 0.0f;
|
||||
if (threadIdx.x == 0)
|
||||
shared.PE[0] = DEN / frameSize;
|
||||
__syncthreads();
|
||||
|
||||
for (int order = 1; order <= max_order; order++)
|
||||
{
|
||||
// [TMP,nn] = sumskipnan(F(:,order+1:frameSize).*B(:,1:frameSize-order),2);
|
||||
shared.tmp[threadIdx.x] = (threadIdx.x + order < frameSize) * shared.F[threadIdx.x + order]*shared.B[threadIdx.x]
|
||||
+ (threadIdx.x + 256 + order < frameSize) * shared.F[threadIdx.x + 256 + order]*shared.B[threadIdx.x + 256];
|
||||
__syncthreads();
|
||||
SUM256(shared.tmp, threadIdx.x);
|
||||
__syncthreads();
|
||||
float reff = shared.tmp[0] / DEN;
|
||||
__syncthreads();
|
||||
|
||||
// arp(:,order) = TMP./DEN; %Burg
|
||||
// rc(:,order) = arp(:,order);
|
||||
if (threadIdx.x == 0)
|
||||
shared.arp[order - 1] = shared.rc[order - 1] = reff;
|
||||
|
||||
// Levinson-Durbin recursion
|
||||
// arp(:,1:order-1) = arp(:,1:order-1) - arp(:,order*ones(order-1,1)).*arp(:,order-1:-1:1);
|
||||
if (threadIdx.x < 32)
|
||||
shared.arp[threadIdx.x] -= (threadIdx.x < order - 1) * __fmul_rz(reff, shared.arp[order - 2 - threadIdx.x]);
|
||||
|
||||
// tmp = F(:,order+1:frameSize) - rc(:,order*ones(1,frameSize-order)).*B(:,1:frameSize-order);
|
||||
// B(:,1:frameSize-order) = B(:,1:frameSize-order) - rc(:,order*ones(1,frameSize-order)).*F(:,order+1:frameSize);
|
||||
// F(:,order+1:frameSize) = tmp;
|
||||
if (threadIdx.x + order < frameSize)
|
||||
{
|
||||
float f = shared.F[threadIdx.x + order];
|
||||
float b = shared.B[threadIdx.x];
|
||||
shared.F[threadIdx.x + order] = f - reff * b;
|
||||
shared.B[threadIdx.x] = b - reff * f;
|
||||
}
|
||||
if (threadIdx.x + order + 256 < frameSize)
|
||||
{
|
||||
float f = shared.F[threadIdx.x + order + 256];
|
||||
float b = shared.B[threadIdx.x + 256];
|
||||
shared.F[threadIdx.x + order + 256] = f - reff * b;
|
||||
shared.B[threadIdx.x + 256] = b - reff * f;
|
||||
}
|
||||
// [PE(:,order+1),nn] = sumskipnan([F(:,order+1:frameSize).^2,B(:,1:frameSize-order).^2],2);
|
||||
shared.tmp[threadIdx.x] = (threadIdx.x + order < frameSize) * (FSQR(shared.F[threadIdx.x + order]) + FSQR(shared.B[threadIdx.x]))
|
||||
+ (threadIdx.x + 256 + order < frameSize) * (FSQR(shared.F[threadIdx.x + 256 + order]) + FSQR(shared.B[threadIdx.x + 256]));
|
||||
__syncthreads();
|
||||
SUM256(shared.tmp, threadIdx.x);
|
||||
__syncthreads();
|
||||
if (threadIdx.x == 0)
|
||||
shared.PE[order] = shared.tmp[0];
|
||||
__syncthreads();
|
||||
|
||||
// BURG:
|
||||
// DEN = PE(:,order+1);
|
||||
//DEN = PE[order];
|
||||
|
||||
// GEOL:
|
||||
//[f,nf] = sumskipnan(F(:,order+1:frameSize).^2,2);
|
||||
//[b,nb] = sumskipnan(B(:,1:frameSize-order).^2,2);
|
||||
//DEN = sqrt(b.*f);
|
||||
|
||||
shared.tmp[threadIdx.x] = (threadIdx.x + order < frameSize) * FSQR(shared.F[threadIdx.x + order])
|
||||
+ (threadIdx.x + 256 + order < frameSize) * FSQR(shared.F[threadIdx.x + 256 + order]);
|
||||
__syncthreads();
|
||||
SUM256(shared.tmp, threadIdx.x);
|
||||
__syncthreads();
|
||||
float f = shared.tmp[0];
|
||||
__syncthreads();
|
||||
|
||||
shared.tmp[threadIdx.x] = (threadIdx.x + order < frameSize) * FSQR(shared.B[threadIdx.x])
|
||||
+ (threadIdx.x + 256 + order < frameSize) * FSQR(shared.B[threadIdx.x + 256]);
|
||||
__syncthreads();
|
||||
SUM256(shared.tmp, threadIdx.x);
|
||||
__syncthreads();
|
||||
float b = shared.tmp[0];
|
||||
__syncthreads();
|
||||
|
||||
DEN = sqrtf(f * b);
|
||||
|
||||
// PE(:,order+1) = PE(:,order+1)./nn; % estimate of covariance
|
||||
if (threadIdx.x == 0)
|
||||
shared.PE[order] /= 2 * (frameSize - order);
|
||||
|
||||
// Quantization
|
||||
if (threadIdx.x < 32)
|
||||
{
|
||||
int precision = 10 - (order > 8) - min(2, shared.task.wbits);
|
||||
int taskNo = taskCount * blockIdx.y + order - 1;
|
||||
shared.bits[threadIdx.x] = __mul24((33 - __clz(__float2int_rn(fabs(shared.arp[threadIdx.x]) * (1 << 15))) - precision), threadIdx.x < order);
|
||||
shared.bits[threadIdx.x] = max(shared.bits[threadIdx.x], shared.bits[threadIdx.x + 16]);
|
||||
shared.bits[threadIdx.x] = max(shared.bits[threadIdx.x], shared.bits[threadIdx.x + 8]);
|
||||
shared.bits[threadIdx.x] = max(shared.bits[threadIdx.x], shared.bits[threadIdx.x + 4]);
|
||||
shared.bits[threadIdx.x] = max(shared.bits[threadIdx.x], shared.bits[threadIdx.x + 2]);
|
||||
shared.bits[threadIdx.x] = max(shared.bits[threadIdx.x], shared.bits[threadIdx.x + 1]);
|
||||
int sh = max(0,min(15, 15 - shared.bits[0]));
|
||||
|
||||
// reverse coefs
|
||||
int coef = max(-(1 << precision),min((1 << precision)-1,__float2int_rn(shared.arp[order - 1 - threadIdx.x] * (1 << sh))));
|
||||
if (threadIdx.x < order)
|
||||
tasks[taskNo].coefs[threadIdx.x] = coef;
|
||||
if (threadIdx.x == 0)
|
||||
tasks[taskNo].shift = sh;
|
||||
shared.bits[threadIdx.x] = 33 - max(__clz(coef),__clz(-1 ^ coef));
|
||||
shared.bits[threadIdx.x] = max(shared.bits[threadIdx.x], shared.bits[threadIdx.x + 16]);
|
||||
shared.bits[threadIdx.x] = max(shared.bits[threadIdx.x], shared.bits[threadIdx.x + 8]);
|
||||
shared.bits[threadIdx.x] = max(shared.bits[threadIdx.x], shared.bits[threadIdx.x + 4]);
|
||||
shared.bits[threadIdx.x] = max(shared.bits[threadIdx.x], shared.bits[threadIdx.x + 2]);
|
||||
shared.bits[threadIdx.x] = max(shared.bits[threadIdx.x], shared.bits[threadIdx.x + 1]);
|
||||
int cbits = shared.bits[0];
|
||||
if (threadIdx.x == 0)
|
||||
tasks[taskNo].cbits = cbits;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// blockDim.x == 32
|
||||
// blockDim.y == 8
|
||||
extern "C" __global__ void cudaEstimateResidual(
|
||||
|
||||
Reference in New Issue
Block a user