2022-10-27 11:11:29 +00:00
// Copyright (c) Facebook, Inc. and its affiliates.
//
// This source code is licensed under the MIT license found in the
2021-10-06 02:16:20 +00:00
// LICENSE file in the root directory of this source tree.
#include <ops.cuh>
#include <kernels.cuh>
#include <cub/device/device_scan.cuh>
#include <limits>
#include <BinSearch.h>
2022-07-22 21:41:05 +00:00
#include <cassert>
2022-07-01 14:16:10 +00:00
#include <common.h>
2021-10-06 02:16:20 +00:00
using namespace BinSearch;
using std::cout;
using std::endl;
2022-07-01 14:41:48 +00:00
void histogramScatterAdd2D(float* histogram, int *index1, int *index2, float *src, int maxidx1, int n)
{
int threads = 512;
2022-08-03 16:05:37 +00:00
int num_blocks = n/threads;
num_blocks = n % threads == 0 ? num_blocks : num_blocks + 1;
kHistogramScatterAdd2D<<<num_blocks, 512>>>(histogram, index1, index2, src, maxidx1, n);
2022-07-01 14:41:48 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
2021-10-06 02:16:20 +00:00
}
2022-07-01 14:41:48 +00:00
template <typename T> void estimateQuantiles(T *A, float *code, float offset, int n)
{
2022-08-03 16:05:37 +00:00
int num_blocks = n/4096;
num_blocks = n % 4096 == 0 ? num_blocks : num_blocks + 1;
2022-07-01 14:41:48 +00:00
CUDA_CHECK_RETURN(cudaMemset(code, 0, 256*sizeof(float)));
2022-08-03 16:05:37 +00:00
kEstimateQuantiles<T><<<num_blocks, 512>>>(A, code, offset, std::numeric_limits<T>::max(), n);
2022-07-01 14:41:48 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
2021-10-06 02:16:20 +00:00
}
2022-07-01 14:41:48 +00:00
void quantize(float *code, float *A, unsigned char *out, int n)
{
2022-08-03 16:05:37 +00:00
int num_blocks = n/1024;
num_blocks = n % 1024 == 0 ? num_blocks : num_blocks + 1;
kQuantize<<<num_blocks, 1024>>>(code, A, out, n);
2022-07-01 14:41:48 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
2021-10-06 02:16:20 +00:00
}
2022-07-01 14:42:58 +00:00
void dequantize(float *code, unsigned char *A, float *out, int n)
{
2022-08-03 16:05:37 +00:00
int num_blocks = n/1024;
num_blocks = n % 1024 == 0 ? num_blocks : num_blocks + 1;
kDequantize<<<num_blocks, 1024>>>(code, A, out, n);
2022-07-01 14:42:58 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
}
2023-04-02 23:10:35 +00:00
template <typename T, int STOCHASTIC, int DATA_TYPE> void quantizeBlockwise(float * code, T *A, float *absmax, unsigned char *out, float *rand, int rand_offset, int blocksize, const int n)
2022-07-01 14:36:30 +00:00
{
2022-11-07 00:27:48 +00:00
int num_blocks = n/blocksize;
num_blocks = n % blocksize == 0 ? num_blocks : num_blocks + 1;
if(STOCHASTIC == 1)
assert(blocksize == 4096);
if(blocksize == 4096)
2023-02-04 22:52:04 +00:00
kQuantizeBlockwise<T, 4096, 4, STOCHASTIC, 0><<<num_blocks, 1024>>>(code, A, absmax, out, rand, rand_offset, n);
2022-11-07 00:27:48 +00:00
else if(blocksize == 2048)
2023-04-02 23:10:35 +00:00
kQuantizeBlockwise<T, 2048, 4, 0, DATA_TYPE><<<num_blocks, 512>>>(code, A, absmax, out, rand, rand_offset, n);
2022-11-07 00:27:48 +00:00
else if(blocksize == 1024)
2023-04-02 23:10:35 +00:00
kQuantizeBlockwise<T, 1024, 4, 0, DATA_TYPE><<<num_blocks, 256>>>(code, A, absmax, out, rand, rand_offset, n);
2022-11-07 00:27:48 +00:00
else if(blocksize == 512)
2023-04-02 23:10:35 +00:00
kQuantizeBlockwise<T, 512, 2, 0, DATA_TYPE><<<num_blocks, 256>>>(code, A, absmax, out, rand, rand_offset, n);
2022-11-20 22:18:15 +00:00
else if(blocksize == 256)
2023-04-02 23:10:35 +00:00
kQuantizeBlockwise<T, 256, 2, 0, DATA_TYPE><<<num_blocks, 128>>>(code, A, absmax, out, rand, rand_offset, n);
2022-11-20 22:18:15 +00:00
else if(blocksize == 128)
2023-04-02 23:10:35 +00:00
kQuantizeBlockwise<T, 128, 2, 0, DATA_TYPE><<<num_blocks, 64>>>(code, A, absmax, out, rand, rand_offset, n);
2022-11-20 22:18:15 +00:00
else if(blocksize == 64)
2023-04-02 23:10:35 +00:00
kQuantizeBlockwise<T, 64, 2, 0, DATA_TYPE><<<num_blocks, 32>>>(code, A, absmax, out, rand, rand_offset, n);
2022-11-07 00:27:48 +00:00
2022-07-01 14:36:30 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
2021-10-06 02:16:20 +00:00
}
2023-04-02 23:10:35 +00:00
template<typename T, int DATA_TYPE> void dequantizeBlockwise(float *code, unsigned char *A, float *absmax, T *out, int blocksize, const int n)
2022-07-01 14:36:30 +00:00
{
2022-08-03 16:05:37 +00:00
int num_blocks = n/blocksize;
num_blocks = n % blocksize == 0 ? num_blocks : num_blocks + 1;
2023-04-02 23:10:35 +00:00
int tile_size = (DATA_TYPE > 0) ? 1024 : 512;
2023-02-04 22:52:04 +00:00
2023-04-02 23:10:35 +00:00
if(DATA_TYPE > 0)
kDequantizeBlockwise<T, 512, 64, 8, DATA_TYPE><<<(n+tile_size-1)/tile_size, 64>>>(code, A, absmax, out, blocksize/2, n);
2023-02-04 22:52:04 +00:00
else
2023-04-02 23:10:35 +00:00
kDequantizeBlockwise<T, 512, 64, 8, DATA_TYPE><<<(n+tile_size-1)/tile_size, 64>>>(code, A, absmax, out, blocksize, n);
2022-11-07 00:27:48 +00:00
2022-07-01 14:36:30 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
2021-10-06 02:16:20 +00:00
}
2023-04-25 23:13:43 +00:00
2023-04-27 00:12:34 +00:00
//void matmul4bite(half *A, unsigned char *B, half*out, int lda, int ldb, int rowsA, int colsA, int colsB)
//{
// int num_blocks = (colsB+32-1)/32;
// kMatmul_inference_4bit<NF4, half, half, half><<<num_blocks, 256>>>(A, B, out, lda, ldb, rowsA, colsA, colsB);
// CUDA_CHECK_RETURN(cudaPeekAtLastError());
//}
2023-04-25 23:13:43 +00:00
2022-07-01 14:36:30 +00:00
template<typename T, int OPTIMIZER> void optimizer32bit(T* g, T* p,
float* state1, float* state2, float *unorm, float max_unorm, float param_norm,
const float beta1, const float beta2, const float eps, const float weight_decay,
const int step, const float lr, const float gnorm_scale, bool skip_zeros, const int n)
{
2022-08-03 16:05:37 +00:00
int num_blocks = n/4096;
num_blocks = n % 4096 == 0 ? num_blocks : num_blocks + 1;
2022-07-01 14:36:30 +00:00
switch(OPTIMIZER)
{
case ADAM:
if(max_unorm > 0.0f)
{
CUDA_CHECK_RETURN(cudaMemset(unorm, 0, 1*sizeof(float)));
2022-08-03 16:05:37 +00:00
kPreconditionOptimizer32bit2State<T, OPTIMIZER, 4096, 8><<<num_blocks, 512>>>(g, p, state1, state2, unorm, beta1, beta2, eps, weight_decay, step, lr, gnorm_scale, n);
2022-07-01 14:36:30 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
}
2022-08-03 16:05:37 +00:00
kOptimizer32bit2State<T, OPTIMIZER><<<num_blocks, 1024>>>(g, p, state1, state2, unorm, max_unorm, param_norm, beta1, beta2, eps, weight_decay, step, lr, gnorm_scale, skip_zeros, n);
2022-07-01 14:36:30 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
break;
case MOMENTUM:
case RMSPROP:
case ADAGRAD:
if(max_unorm > 0.0f)
{
CUDA_CHECK_RETURN(cudaMemset(unorm, 0, 1*sizeof(float)));
2022-08-03 16:05:37 +00:00
kPreconditionOptimizer32bit1State<T, OPTIMIZER, 4096, 8><<<num_blocks, 512>>>(g, p, state1, unorm, beta1, eps, weight_decay, step, lr, gnorm_scale, n);
2022-07-01 14:36:30 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
}
2022-08-03 16:05:37 +00:00
kOptimizer32bit1State<T, OPTIMIZER><<<num_blocks, 1024>>>(g, p, state1, unorm, max_unorm, param_norm, beta1, eps, weight_decay, step, lr, gnorm_scale, skip_zeros, n);
2022-07-01 14:36:30 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
break;
}
2021-10-06 02:16:20 +00:00
}
2022-07-01 14:36:30 +00:00
template<typename T, int OPTIMIZER> void optimizerStatic8bit(T* p, T* g,
unsigned char* state1, unsigned char* state2,
float *unorm, float max_unorm, float param_norm,
float beta1, float beta2,
float eps, int step, float lr,
float* quantiles1, float* quantiles2,
float* max1, float* max2, float* new_max1, float* new_max2,
float weight_decay,
const float gnorm_scale, int n)
{
2022-08-03 16:05:37 +00:00
int num_blocks = n/4096;
num_blocks = n % 4096 == 0 ? num_blocks : num_blocks + 1;
2022-07-01 14:36:30 +00:00
if(max_unorm > 0.0f){ CUDA_CHECK_RETURN(cudaMemset(unorm, 0, 1*sizeof(float))); }
switch(OPTIMIZER)
{
case ADAM:
CUDA_CHECK_RETURN(cudaMemset(new_max1, 0, 1*sizeof(float)));
CUDA_CHECK_RETURN(cudaMemset(new_max2, 0, 1*sizeof(float)));
2022-08-03 16:05:37 +00:00
kPreconditionOptimizerStatic8bit2State<T, OPTIMIZER><<<num_blocks, 256>>>(p, g, state1, state2, unorm, beta1, beta2, eps, step, quantiles1, quantiles2, max1, max2, new_max1, new_max2, gnorm_scale, n);
2022-07-01 14:36:30 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
2022-08-03 16:05:37 +00:00
kOptimizerStatic8bit2State<T, OPTIMIZER><<<num_blocks, 1024>>>(p, g, state1, state2, unorm, max_unorm, param_norm, beta1, beta2, eps, step, lr,
2022-07-01 14:36:30 +00:00
quantiles1, quantiles2, max1, max2, new_max1, new_max2, weight_decay, gnorm_scale, n);
CUDA_CHECK_RETURN(cudaPeekAtLastError());
break;
case MOMENTUM:
case RMSPROP:
case ADAGRAD:
CUDA_CHECK_RETURN(cudaMemset(new_max1, 0, 1*sizeof(float)));
2022-08-03 16:05:37 +00:00
kPreconditionOptimizerStatic8bit1State<T, OPTIMIZER><<<num_blocks, 256>>>(p, g, state1, unorm, beta1, eps, step, quantiles1, max1, new_max1, weight_decay, gnorm_scale, n);
2022-07-01 14:36:30 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
2022-08-03 16:05:37 +00:00
kOptimizerStatic8bit1State<T, OPTIMIZER><<<num_blocks, 1024>>>(p, g, state1, unorm, max_unorm, param_norm, beta1, eps, step, lr,
2022-07-01 14:36:30 +00:00
quantiles1, max1, new_max1, weight_decay, gnorm_scale, n);
CUDA_CHECK_RETURN(cudaPeekAtLastError());
break;
default:
break;
}
2021-10-06 02:16:20 +00:00
}
#define BLOCKSIZE_2STATE 2048
#define NUM_2STATE 8
#define BLOCKSIZE_1STATE 2048
#define NUM_1STATE 8
2022-07-01 14:36:30 +00:00
template<typename T, int OPTIMIZER> void optimizerStatic8bitBlockwise(T* p, T* g,
unsigned char* state1, unsigned char* state2, float beta1, float beta2, float eps, int step, float lr,
float* quantiles1, float* quantiles2, float* absmax1, float* absmax2, float weight_decay, const float gnorm_scale, bool skip_zeros, int n)
{
2022-08-03 16:05:37 +00:00
int num_blocks = 0;
2022-07-01 14:36:30 +00:00
switch(OPTIMIZER)
{
case ADAM:
2022-08-03 16:05:37 +00:00
num_blocks = n/BLOCKSIZE_2STATE;
num_blocks = n % BLOCKSIZE_2STATE == 0 ? num_blocks : num_blocks + 1;
kOptimizerStatic8bit2StateBlockwise<T, OPTIMIZER, BLOCKSIZE_2STATE, NUM_2STATE><<<num_blocks, BLOCKSIZE_2STATE/NUM_2STATE>>>(p, g, state1, state2, beta1, beta2, eps, step, lr,
2022-07-01 14:36:30 +00:00
quantiles1, quantiles2, absmax1, absmax2, weight_decay, gnorm_scale, skip_zeros, n);
CUDA_CHECK_RETURN(cudaPeekAtLastError());
break;
case MOMENTUM:
case RMSPROP:
case ADAGRAD:
2022-08-03 16:05:37 +00:00
num_blocks = n/BLOCKSIZE_1STATE;
num_blocks = n % BLOCKSIZE_1STATE == 0 ? num_blocks : num_blocks + 1;
kOptimizerStatic8bit1StateBlockwise<T, OPTIMIZER, BLOCKSIZE_1STATE, NUM_1STATE><<<num_blocks, BLOCKSIZE_1STATE/NUM_1STATE>>>(p, g, state1, beta1, beta2, eps, step, lr,
2022-07-01 14:36:30 +00:00
quantiles1, absmax1, weight_decay, gnorm_scale, skip_zeros, n);
CUDA_CHECK_RETURN(cudaPeekAtLastError());
break;
}
2021-10-06 02:16:20 +00:00
}
2022-07-01 14:36:30 +00:00
template<typename T> void percentileClipping(T * g, float *gnorm_vec, int step, const int n)
{
2022-08-03 16:05:37 +00:00
int num_blocks = n/2048;
num_blocks = n % 2048 == 0 ? num_blocks : num_blocks + 1;
2022-07-01 14:36:30 +00:00
CUDA_CHECK_RETURN(cudaMemset(&gnorm_vec[step % 100], 0, 1*sizeof(float)));
2022-08-03 16:05:37 +00:00
kPercentileClipping<T, 2048, 4><<<num_blocks, 512>>>(g, gnorm_vec, step, n);
2022-07-01 14:36:30 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
2021-10-06 02:16:20 +00:00
}
2022-07-22 21:41:05 +00:00
void gemmex(Context *context, bool transposeA, bool transposeB, int m, int n, int k, void *A, void *B, void *C, int lda, int ldb, int ldc)
{
const int falpha = 1;
const int fbeta = 0;
const void * alpha = &falpha;
const void * beta = &fbeta;
cublasStatus_t status;
status = cublasGemmEx(context->m_handle,
transposeA ? CUBLAS_OP_T : CUBLAS_OP_N,
transposeB ? CUBLAS_OP_T : CUBLAS_OP_N,
m, n, k,
alpha, A, CUDA_R_8I, lda, B, CUDA_R_8I, ldb, beta,
C, CUDA_R_32I, ldc,
CUDA_R_32I, CUBLAS_GEMM_DEFAULT_TENSOR_OP);
if (status != CUBLAS_STATUS_SUCCESS)
{
std::cout << "CUBLAS ERROR: Status " << status << std::endl;
}
}
2022-10-27 11:11:29 +00:00
void strided_gemmex(Context *context, bool transposeA, bool transposeB, int m, int n, int k, void *A, void *B, void *C, int lda, int ldb, int ldc,
2022-07-22 21:41:05 +00:00
long long int strideA, long long int strideB, long long int strideC, int batchCount)
{
const int falpha = 1;
const int fbeta = 0;
const void * alpha = &falpha;
const void * beta = &fbeta;
cublasStatus_t status;
//cout << transposeA << transposeB << endl;
//printf("%i %i %i\n", m,n,k);
//printf("%i %i %i\n", lda,ldb,ldc);
//printf("%i %i %i\n", strideA, strideB, strideC);
//printf("%i\n", batchCount);
status = cublasGemmStridedBatchedEx(context->m_handle,
transposeA ? CUBLAS_OP_T : CUBLAS_OP_N,
transposeB ? CUBLAS_OP_T : CUBLAS_OP_N,
m, n, k,
alpha, A, CUDA_R_8I, lda, (long long int)strideA, B, CUDA_R_8I, ldb, (long long int)strideB, beta,
C, CUDA_R_32I, ldc, (long long int)strideC, batchCount,
CUDA_R_32I, CUBLAS_GEMM_DEFAULT);
if (status != CUBLAS_STATUS_SUCCESS)
{
std::cout << "CUBLAS ERROR: Status " << status << std::endl;
}
}
int roundoff(int v, int d) {
return (v + d - 1) / d * d;
}
2022-07-26 02:30:37 +00:00
#ifdef NO_CUBLASLT
#else
2022-07-22 21:41:05 +00:00
template<int ORDER> cublasLtOrder_t get_order()
{
switch(ORDER)
{
case ROW:
return CUBLASLT_ORDER_ROW;
break;
case COL:
return CUBLASLT_ORDER_COL;
break;
case COL32:
return CUBLASLT_ORDER_COL32;
break;
case COL_TURING:
return CUBLASLT_ORDER_COL4_4R2_8C;
break;
case COL_AMPERE:
return CUBLASLT_ORDER_COL32_2R_4R4;
break;
2022-07-26 02:30:37 +00:00
default:
break;
2022-07-22 21:41:05 +00:00
}
2022-07-26 02:30:37 +00:00
return CUBLASLT_ORDER_ROW;
2022-07-22 21:41:05 +00:00
}
template cublasLtOrder_t get_order<ROW>();
template cublasLtOrder_t get_order<COL>();
template cublasLtOrder_t get_order<COL32>();
template cublasLtOrder_t get_order<COL_TURING>();
template cublasLtOrder_t get_order<COL_AMPERE>();
2022-07-26 02:30:37 +00:00
#endif
2022-07-22 21:41:05 +00:00
template<int ORDER> int get_leading_dim(int dim1, int dim2)
{
switch(ORDER)
{
case ROW:
return dim2;
break;
case COL:
return dim1;
break;
case COL32:
// 32*row tiles
return dim1*32;
break;
case COL_TURING:
return 32*roundoff(dim1, 8);
break;
case COL_AMPERE:
// 32*32 tiles
return 32*roundoff(dim1, 32);
break;
2022-07-26 02:30:37 +00:00
default:
return 0;
break;
2022-07-22 21:41:05 +00:00
}
}
template int get_leading_dim<ROW>(int dim1, int dim2);
template int get_leading_dim<COL>(int dim1, int dim2);
template int get_leading_dim<COL32>(int dim1, int dim2);
template <typename T, int SRC, int TARGET, bool transpose, int DTYPE> void transform(cublasLtHandle_t ltHandle, T *A, T *out, int dim1, int dim2)
{
2022-07-26 02:30:37 +00:00
#ifdef NO_CUBLASLT
#else
2022-07-22 21:41:05 +00:00
cublasLtOrder_t orderA = get_order<SRC>();
cublasLtOrder_t orderOut = get_order<TARGET>();
int ldA = get_leading_dim<SRC>(dim1, dim2);
int ldOut = get_leading_dim<TARGET>(dim1, dim2);
2022-10-27 11:11:29 +00:00
2022-07-22 21:41:05 +00:00
cublasLtMatrixLayout_t A_desc = NULL, out_desc = NULL;
cublasLtMatrixTransformDesc_t A2Out_desc = NULL;
cublasOperation_t opTranspose = CUBLAS_OP_T;
float transformAlpha = 1.0f, transformBeta = 0.0f;
if(DTYPE == 8)
{
checkCublasStatus(cublasLtMatrixLayoutCreate(&A_desc, CUDA_R_8I, dim1, dim2, ldA));
checkCublasStatus(cublasLtMatrixLayoutCreate(&out_desc, CUDA_R_8I, dim1, dim2, ldOut));
}
else if(DTYPE == 32)
{
checkCublasStatus(cublasLtMatrixLayoutCreate(&A_desc, CUDA_R_32I, dim1, dim2, ldA));
checkCublasStatus(cublasLtMatrixLayoutCreate(&out_desc, CUDA_R_32I, dim1, dim2, ldOut));
}
else
{
printf("ERROR WRONG TYPE FOR TRANSFORM: %i\n", DTYPE);
}
checkCublasStatus(cublasLtMatrixLayoutSetAttribute(A_desc, CUBLASLT_MATRIX_LAYOUT_ORDER, &orderA, sizeof(orderA)));
checkCublasStatus(cublasLtMatrixLayoutSetAttribute(out_desc, CUBLASLT_MATRIX_LAYOUT_ORDER, &orderOut, sizeof(orderOut)));
checkCublasStatus(cublasLtMatrixTransformDescCreate(&A2Out_desc, CUDA_R_32F));
if(transpose){ checkCublasStatus(cublasLtMatrixTransformDescSetAttribute(A2Out_desc, CUBLASLT_MATRIX_TRANSFORM_DESC_TRANSA, &opTranspose, sizeof(opTranspose))); }
checkCublasStatus(cublasLtMatrixTransform(ltHandle, A2Out_desc, &transformAlpha, A, A_desc, &transformBeta, NULL, NULL, out, out_desc, 0));
if (A_desc) checkCublasStatus(cublasLtMatrixLayoutDestroy(A_desc));
if (out_desc) checkCublasStatus(cublasLtMatrixLayoutDestroy(out_desc));
if (A2Out_desc) checkCublasStatus(cublasLtMatrixTransformDescDestroy(A2Out_desc));
2022-07-26 02:30:37 +00:00
#endif
2022-07-22 21:41:05 +00:00
}
template void transform<int8_t, ROW, COL, false, 8>(cublasLtHandle_t ltHandle, int8_t *A, int8_t *out, int dim1, int dim2);
template void transform<int8_t, ROW, ROW, false, 8>(cublasLtHandle_t ltHandle, int8_t *A, int8_t *out, int dim1, int dim2);
template void transform<int8_t, ROW, COL32, false, 8>(cublasLtHandle_t ltHandle, int8_t *A, int8_t *out, int dim1, int dim2);
template void transform<int32_t, ROW, COL32, false, 32>(cublasLtHandle_t ltHandle, int32_t *A, int32_t *out, int dim1, int dim2);
template void transform<int8_t, ROW, COL_TURING, false, 8>(cublasLtHandle_t ltHandle, int8_t *A, int8_t *out, int dim1, int dim2);
template void transform<int8_t, ROW, COL_AMPERE, false, 8>(cublasLtHandle_t ltHandle, int8_t *A, int8_t *out, int dim1, int dim2);
template void transform<int8_t, COL32, ROW, false, 8>(cublasLtHandle_t ltHandle, int8_t *A, int8_t *out, int dim1, int dim2);
template void transform<int32_t, COL32, ROW, false, 32>(cublasLtHandle_t ltHandle, int32_t *A, int32_t *out, int dim1, int dim2);
2022-10-27 11:11:29 +00:00
template <int FORMATB, int DTYPE_OUT, int SCALE_ROWS> int igemmlt(cublasLtHandle_t ltHandle, int m, int n, int k, const int8_t *A, const int8_t *B, void *C, float *row_scale, int lda, int ldb, int ldc)
2022-07-22 21:41:05 +00:00
{
2022-07-26 02:30:37 +00:00
#ifdef NO_CUBLASLT
2022-08-23 23:00:26 +00:00
cout << "" << endl;
cout << "=============================================" << endl;
cout << "ERROR: Your GPU does not support Int8 Matmul!" << endl;
cout << "=============================================" << endl;
cout << "" << endl;
2022-08-17 02:03:19 +00:00
assert(false);
2022-07-26 02:30:37 +00:00
return 0;
#else
2022-07-22 21:41:05 +00:00
int has_error = 0;
cublasLtMatmulDesc_t matmulDesc = NULL;
cublasLtMatrixLayout_t Adesc = NULL, Bdesc = NULL, Cdesc = NULL;
cublasOperation_t opT = CUBLAS_OP_T;
cublasLtPointerMode_t alphaVec = CUBLASLT_POINTER_MODE_ALPHA_DEVICE_VECTOR_BETA_ZERO;
cublasLtOrder_t col32 = CUBLASLT_ORDER_COL32;
cublasLtOrder_t col_turing = CUBLASLT_ORDER_COL4_4R2_8C;
cublasLtOrder_t col_ampere = CUBLASLT_ORDER_COL32_2R_4R4;
has_error |= checkCublasStatus(cublasLtMatrixLayoutCreate(&Adesc, CUDA_R_8I, m, k, lda));
has_error |= checkCublasStatus(cublasLtMatrixLayoutCreate(&Bdesc, CUDA_R_8I, n, k, ldb));
has_error |= checkCublasStatus(cublasLtMatrixLayoutSetAttribute(Adesc, CUBLASLT_MATRIX_LAYOUT_ORDER, &col32, sizeof(col32)));
if(FORMATB == COL_TURING)
has_error |= checkCublasStatus(cublasLtMatrixLayoutSetAttribute(Bdesc, CUBLASLT_MATRIX_LAYOUT_ORDER, &col_turing, sizeof(col_turing)));
else
has_error |= checkCublasStatus(cublasLtMatrixLayoutSetAttribute(Bdesc, CUBLASLT_MATRIX_LAYOUT_ORDER, &col_ampere, sizeof(col_ampere)));
if(DTYPE_OUT == 32)
{
has_error |= checkCublasStatus(cublasLtMatmulDescCreate(&matmulDesc, CUBLAS_COMPUTE_32I, CUDA_R_32I));
has_error |= checkCublasStatus(cublasLtMatmulDescSetAttribute(matmulDesc, CUBLASLT_MATMUL_DESC_TRANSB, &opT, sizeof(opT)));
has_error |= checkCublasStatus(cublasLtMatrixLayoutCreate(&Cdesc, CUDA_R_32I, m, n, ldc));
has_error |= checkCublasStatus(cublasLtMatrixLayoutSetAttribute(Cdesc, CUBLASLT_MATRIX_LAYOUT_ORDER, &col32, sizeof(col32)));
int alpha = 1, beta = 0;
has_error |= checkCublasStatus(cublasLtMatmul(ltHandle, matmulDesc,&alpha, A, Adesc, B, Bdesc, &beta, (int32_t*)C, Cdesc, (int32_t*)C, Cdesc, NULL, NULL, 0, 0));
}
else
{
has_error |= checkCublasStatus(cublasLtMatmulDescCreate(&matmulDesc, CUBLAS_COMPUTE_32I, CUDA_R_32F));
has_error |= checkCublasStatus(cublasLtMatmulDescSetAttribute(matmulDesc, CUBLASLT_MATMUL_DESC_TRANSB, &opT, sizeof(opT)));
has_error |= checkCublasStatus(cublasLtMatrixLayoutCreate(&Cdesc, CUDA_R_8I, m, n, ldc));
has_error |= checkCublasStatus(cublasLtMatrixLayoutSetAttribute(Cdesc, CUBLASLT_MATRIX_LAYOUT_ORDER, &col32, sizeof(col32)));
if(!SCALE_ROWS)
{
float alpha = 1.0f, beta = 0.0f;
has_error |= checkCublasStatus(cublasLtMatmul(ltHandle, matmulDesc,&alpha, A, Adesc, B, Bdesc, &beta, (int8_t*)C, Cdesc, (int8_t*)C, Cdesc, NULL, NULL, 0, 0));
}
else
{
has_error |= checkCublasStatus(cublasLtMatmulDescSetAttribute(matmulDesc, CUBLASLT_MATMUL_DESC_POINTER_MODE, &alphaVec, sizeof(alphaVec)));
has_error |= checkCublasStatus(cublasLtMatmul(ltHandle, matmulDesc, row_scale, A, Adesc, B, Bdesc, NULL, (int8_t*)C, Cdesc, (int8_t*)C, Cdesc, NULL, NULL, 0, 0));
}
}
if (Cdesc) has_error |= checkCublasStatus(cublasLtMatrixLayoutDestroy(Cdesc));
if (Bdesc) has_error |= checkCublasStatus(cublasLtMatrixLayoutDestroy(Bdesc));
if (Adesc) has_error |= checkCublasStatus(cublasLtMatrixLayoutDestroy(Adesc));
if (matmulDesc) has_error |= checkCublasStatus(cublasLtMatmulDescDestroy(matmulDesc));
if(has_error == 1)
printf("error detected");
return has_error;
2022-07-26 02:30:37 +00:00
#endif
2022-07-22 21:41:05 +00:00
}
int fill_up_to_nearest_multiple(int value, int multiple)
{
return value + (value % multiple == 0 ? 0 : (multiple - (value % multiple)));
}
2022-08-16 17:56:17 +00:00
void dequant_mm_int32_fp16(int *A, float *rowStats, float *colStats, half *out, float* newRowStats, float* newcolStats, half *bias, int numRows, int numCols)
2022-07-22 21:41:05 +00:00
{
int threads = 512;
int tileCols = fill_up_to_nearest_multiple(numCols, 32);
int n = numRows*tileCols;
int subtile_rows = 128;
int tilesize = 32*subtile_rows;
int num_blocks = numRows/subtile_rows;
num_blocks += (numRows % subtile_rows == 0) ? 0 : 1;
num_blocks = num_blocks*(tileCols/32);
assert(threads <= tilesize);
2022-08-16 17:56:17 +00:00
kdequant_mm_int32_fp16<4, 128, 512><<<num_blocks, threads>>>(A, rowStats, colStats, out, newRowStats, newcolStats, bias, numRows, numCols, tileCols, n);
2022-07-22 21:41:05 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
}
#define STATS_THREADS 64
#define STATS_ITEMS 4
#define STATS_ROWS 16
void getColRowStats(half * A, float *rowStats, float *colStats, int *nnz_count_row, float nnz_threshold, int rows, int cols)
{
int tile_cols = STATS_THREADS*STATS_ITEMS;
int tiledCols = fill_up_to_nearest_multiple(cols, tile_cols);
int tiledRows = fill_up_to_nearest_multiple(rows, STATS_ROWS);
2022-08-03 18:54:01 +00:00
int row_tiles = (tiledRows/STATS_ROWS);
int col_tiles = (tiledCols/tile_cols);
row_tiles = row_tiles > 0 ? row_tiles : 1;
col_tiles = col_tiles > 0 ? col_tiles : 1;
int num_blocks = row_tiles * col_tiles;
2022-07-22 21:41:05 +00:00
if(nnz_threshold == 0.0)
kgetColRowStats<half, STATS_THREADS, STATS_ITEMS, STATS_ROWS, STATS_THREADS*STATS_ITEMS, 0><<<num_blocks, STATS_THREADS>>>(A, rowStats, colStats, nnz_count_row, nnz_threshold, rows, cols, tiledRows, tiledCols);
else if(nnz_threshold != 0.0)
kgetColRowStats<half, STATS_THREADS, STATS_ITEMS, STATS_ROWS, STATS_THREADS*STATS_ITEMS, 1><<<num_blocks, STATS_THREADS>>>(A, rowStats, colStats, nnz_count_row, nnz_threshold, rows, cols, tiledRows, tiledCols);
CUDA_CHECK_RETURN(cudaPeekAtLastError());
}
void doubleRowColQuant(half * A, float *rowStats, float *colStats, char *out_col_normed, char *out_row_normed, int *rowidx, int *colidx, half *val, int *nnz_block_ptr, float threshold, int rows, int cols)
{
int threads = 64;
int items_per_thread = 4;
int tile_cols = threads*items_per_thread;
int tile_rows = 16;
int tiledCols = fill_up_to_nearest_multiple(cols, tile_cols);
int tiledRows = fill_up_to_nearest_multiple(rows, tile_rows);
2022-08-03 18:54:01 +00:00
int row_tiles = (tiledRows/tile_rows);
int col_tiles = (tiledCols/tile_cols);
row_tiles = row_tiles > 0 ? row_tiles : 1;
col_tiles = col_tiles > 0 ? col_tiles : 1;
int num_blocks = row_tiles * col_tiles;
2022-07-22 21:41:05 +00:00
if(threshold > 0.0f)
kDoubleRowColQuant<64, 4, 16, 64*4, 1><<<num_blocks, threads>>>(A, rowStats, colStats, out_col_normed, out_row_normed, rowidx, colidx, val, nnz_block_ptr, threshold, rows, cols, tiledCols);
else
kDoubleRowColQuant<64, 4, 16, 64*4, 0><<<num_blocks, threads>>>(A, rowStats, colStats, out_col_normed, out_row_normed, rowidx, colidx, val, nnz_block_ptr, threshold, rows, cols, tiledCols);
CUDA_CHECK_RETURN(cudaPeekAtLastError());
}
template <int FORMAT, int TRANSPOSE> void transformRowToFormat(char * A, char *out, int rows, int cols)
{
int threads = 256;
int items_per_thread = 8;
// we load 128 column values per warp
int tile_cols = 32*items_per_thread;
int tile_rows = 32;
int tiledCols = fill_up_to_nearest_multiple(cols, tile_cols);
int tiledRows = fill_up_to_nearest_multiple(rows, tile_rows);
2022-08-03 18:54:01 +00:00
int row_tiles = (tiledRows/tile_rows);
int col_tiles = (tiledCols/tile_cols);
row_tiles = row_tiles > 0 ? row_tiles : 1;
col_tiles = col_tiles > 0 ? col_tiles : 1;
int num_blocks = row_tiles * col_tiles;
2022-07-22 21:41:05 +00:00
int outCols = fill_up_to_nearest_multiple(cols, 32);
int outRows = fill_up_to_nearest_multiple(rows, 32);
if(FORMAT == COL_TURING)
{
if(TRANSPOSE)
outRows = fill_up_to_nearest_multiple(cols, 8);
else
outRows = fill_up_to_nearest_multiple(rows, 8);
}
else if(FORMAT == COL_AMPERE)
{
if(TRANSPOSE)
outRows = fill_up_to_nearest_multiple(cols, 32);
else
outRows = fill_up_to_nearest_multiple(rows, 32);
}
else
{
if(TRANSPOSE)
{
outCols = fill_up_to_nearest_multiple(rows, 32);
outRows = cols;
}
}
kTransformRowToFormat<256, 8, 32, 32*8, TRANSPOSE, FORMAT><<<num_blocks, threads>>>(A, out, rows, cols, tiledCols, outRows, outCols);
CUDA_CHECK_RETURN(cudaPeekAtLastError());
}
void spmm_coo(cusparseHandle_t handle, int *A_rowidx, int *A_colidx, half *A_vals, int A_nnz, int A_rows, int A_cols, int B_cols, int ldb, half *B, int ldc, half* C, bool transposed_B)
{
2022-07-26 02:30:37 +00:00
#ifdef NO_CUBLASLT
#else
2022-07-22 21:41:05 +00:00
cusparseSpMatDescr_t descA;
cusparseDnMatDescr_t descB, descC;
float alpha = 1.0f;
float beta = 0.0f;
void *dBuffer = NULL;
size_t bufferSize = 0;
CHECK_CUSPARSE( cusparseCreateCoo(&descA, A_rows, A_cols, A_nnz,
A_rowidx, A_colidx, A_vals,
CUSPARSE_INDEX_32I,
CUSPARSE_INDEX_BASE_ZERO, CUDA_R_16F) );
// Create dense matrix C
CHECK_CUSPARSE( cusparseCreateDnMat(&descC, A_rows, B_cols, ldc, C,
CUDA_R_16F, CUSPARSE_ORDER_ROW) );
// Create dense matrix B
if(transposed_B)
{
int tmp = A_cols;
A_cols = B_cols;
B_cols = tmp;
}
CHECK_CUSPARSE( cusparseCreateDnMat(&descB, A_cols, B_cols, ldb, B,
CUDA_R_16F, CUSPARSE_ORDER_ROW) );
// allocate an external buffer if needed
CHECK_CUSPARSE( cusparseSpMM_bufferSize(
handle,
CUSPARSE_OPERATION_NON_TRANSPOSE,
transposed_B ? CUSPARSE_OPERATION_TRANSPOSE : CUSPARSE_OPERATION_NON_TRANSPOSE,
&alpha, descA, descB, &beta, descC, CUDA_R_32F,
CUSPARSE_SPMM_ALG_DEFAULT, &bufferSize) );
CUDA_CHECK_RETURN( cudaMalloc(&dBuffer, bufferSize) );
// execute SpMM
CHECK_CUSPARSE( cusparseSpMM(handle,
CUSPARSE_OPERATION_NON_TRANSPOSE,
transposed_B ? CUSPARSE_OPERATION_TRANSPOSE : CUSPARSE_OPERATION_NON_TRANSPOSE,
&alpha, descA, descB, &beta, descC, CUDA_R_32F,
CUSPARSE_SPMM_ALG_DEFAULT, dBuffer));
// destroy matrix/vector descriptors
CHECK_CUSPARSE( cusparseDestroySpMat(descA) );
CHECK_CUSPARSE( cusparseDestroyDnMat(descB) );
CHECK_CUSPARSE( cusparseDestroyDnMat(descC) );
CUDA_CHECK_RETURN( cudaFree(dBuffer) );
2022-07-26 02:30:37 +00:00
#endif
2022-07-22 21:41:05 +00:00
}
template <typename T, int BITS> void spmm_coo_very_sparse_naive(int *max_count, int *max_idx, int *offset_rowidx, int *rowidx, int *colidx, half *values, T *B, half *out, float *dequant_stats, int nnz_rows, int nnz, int rowsA, int rowsB, int colsB)
{
kspmm_coo_very_sparse_naive<T, 8, BITS><<<nnz_rows, 256>>>(max_count, max_idx, offset_rowidx, rowidx, colidx, values, B, out, dequant_stats, nnz, rowsA, rowsB, colsB);
CUDA_CHECK_RETURN(cudaPeekAtLastError());
}
2021-10-06 02:16:20 +00:00
2022-07-26 19:12:38 +00:00
template <int FORMAT> void extractOutliers(char * A, int *idx, char *out, int idx_size, int rows, int cols)
{
int threads = 256;
// we load 128 column values per warp
int tiledCols = tiledCols = fill_up_to_nearest_multiple(cols, 32);
int tiledRows = 0;
2022-07-27 00:39:30 +00:00
int num_blocks = idx_size;
2022-07-26 19:12:38 +00:00
if(FORMAT == COL_TURING)
{
tiledRows = fill_up_to_nearest_multiple(rows, 8);
}
else if(FORMAT == COL_AMPERE)
{
tiledRows = fill_up_to_nearest_multiple(rows, 32);
}
2022-07-27 00:39:30 +00:00
kExtractOutliers<FORMAT><<<num_blocks, threads>>>(A, idx, out, idx_size, rows, cols, tiledRows, tiledCols);
2022-07-26 19:12:38 +00:00
CUDA_CHECK_RETURN(cudaPeekAtLastError());
}
2023-04-25 23:13:43 +00:00
2023-04-27 22:12:49 +00:00
void pipeline_test(float *A, float *B, size_t n, size_t batch_size)
{
int threads = 256;
int num_blocks = (n+(256*batch_size)+1)/(batch_size*256);
with_staging_unified<2><<<num_blocks, threads>>>(A, B, n, batch_size);
CUDA_CHECK_RETURN(cudaPeekAtLastError());
}
2023-04-25 23:15:44 +00:00
2023-04-27 00:12:34 +00:00
2023-04-26 00:15:51 +00:00
2023-04-29 05:10:42 +00:00
template <typename T> void gemm_host(int m, int n, int k, T * A, T* B, T * out, int lda, int ldb, int ldc, int bits)
2023-04-25 23:15:44 +00:00
{
2023-04-27 00:12:34 +00:00
2023-05-01 23:23:45 +00:00
int num_blocks = (m+15)/16;
2023-04-27 00:12:34 +00:00
2023-04-27 22:11:26 +00:00
cout << num_blocks << endl;
2023-04-29 00:19:02 +00:00
cout << lda << endl;
cout << ldb << endl;
cout << ldc << endl;
cout << m << endl;
cout << n << endl;
cout << k << endl;
2023-05-01 00:43:02 +00:00
//if(bits == 32)
//gemm_device<T, 32, 128><<< num_blocks, 128, 0, 0 >>>(m, n, k, A, B, out, lda, ldb, ldc);
//gemm_device<T, 32, 32><<< num_blocks, 32, 0, 0 >>>(m, n, k, A, B, out, lda, ldb, ldc);
if(bits == 16)
2023-05-01 15:21:12 +00:00
//gemm_device<T, 16, 256><<< num_blocks, 256, 0, 0 >>>(m, n, k, A, B, out, lda, ldb, ldc);
gemm_device<T, 16, 128><<< num_blocks, 128, 0, 0 >>>(m, n, k, A, B, out, lda, ldb, ldc);
2023-05-01 01:55:12 +00:00
//gemm_device<T, 16, 32><<< num_blocks, 32, 0, 0 >>>(m, n, k, A, B, out, lda, ldb, ldc);
2023-05-01 15:21:12 +00:00
//gemm_device<T, 16, 64><<< num_blocks, 64, 0, 0 >>>(m, n, k, A, B, out, lda, ldb, ldc);
2023-04-25 23:15:44 +00:00
}
2023-04-30 04:52:47 +00:00
template <typename T> void gemm_4bit_inference(int m, int n, int k, T * A, unsigned char* B, float *absmax, T * out, int lda, int ldb, int ldc, int blocksize)
{
int num_blocks = (m+7)/8;
cout << num_blocks << endl;
cout << lda << endl;
cout << ldb << endl;
cout << ldc << endl;
cout << m << endl;
cout << n << endl;
cout << k << endl;
2023-05-01 00:43:02 +00:00
kgemm_4bit_inference<T, 128><<< num_blocks, 128, 0, 0 >>>(m, n, k, A, B, absmax, out, lda, ldb, ldc, blocksize);
//kgemm_4bit_inference<T, 32><<< num_blocks, 32, 0, 0 >>>(m, n, k, A, B, absmax, out, lda, ldb, ldc, blocksize);
2023-04-30 04:52:47 +00:00
}
2021-10-06 02:16:20 +00:00
//==============================================================
// TEMPLATE DEFINITIONS
//==============================================================
2023-04-30 04:52:47 +00:00
template void gemm_4bit_inference<half>(int m, int n, int k, half * A, unsigned char* B, float *absmax, half * out, int lda, int ldb, int ldc, int blocksize);
2023-05-01 00:43:02 +00:00
//template void gemm_host<float>(int m, int n, int k, float * A, float* B, float * out, int lda, int ldb, int ldc, int bits);
2023-04-29 05:10:42 +00:00
template void gemm_host<half>(int m, int n, int k, half * A, half* B, half * out, int lda, int ldb, int ldc, int bits);
2022-07-26 19:12:38 +00:00
template void extractOutliers<COL_TURING>(char * A, int *idx, char *out, int idx_size, int rows, int cols);
template void extractOutliers<COL_AMPERE>(char * A, int *idx, char *out, int idx_size, int rows, int cols);
2022-07-22 21:41:05 +00:00
template void spmm_coo_very_sparse_naive<half, 16>(int *max_count, int *max_idx, int *offset_rowidx, int *rowidx, int *colidx, half *values, half *B, half *out, float *dequant_stats, int nnz_rows, int nnz, int rowsA, int rowsB, int colsB);
template void spmm_coo_very_sparse_naive<signed char, 8>(int *max_count, int *max_idx, int *offset_rowidx, int *rowidx, int *colidx, half *values, signed char *B, half *out, float *dequant_stats, int nnz_rows, int nnz, int rowsA, int rowsB, int colsB);
template int igemmlt<COL_TURING, 32, 0>(cublasLtHandle_t ltHandle, int m, int n, int k, const int8_t *A, const int8_t *B, void *C, float *row_scale, int lda, int ldb, int ldc);
template int igemmlt<COL_TURING, 8, 0>(cublasLtHandle_t ltHandle, int m, int n, int k, const int8_t *A, const int8_t *B, void *C, float *row_scale, int lda, int ldb, int ldc);
template int igemmlt<COL_TURING, 8, 1>(cublasLtHandle_t ltHandle, int m, int n, int k, const int8_t *A, const int8_t *B, void *C, float *row_scale, int lda, int ldb, int ldc);
template int igemmlt<COL_AMPERE, 32, 0>(cublasLtHandle_t ltHandle, int m, int n, int k, const int8_t *A, const int8_t *B, void *C, float *row_scale, int lda, int ldb, int ldc);
template int igemmlt<COL_AMPERE, 8, 0>(cublasLtHandle_t ltHandle, int m, int n, int k, const int8_t *A, const int8_t *B, void *C, float *row_scale, int lda, int ldb, int ldc);
template int igemmlt<COL_AMPERE, 8, 1>(cublasLtHandle_t ltHandle, int m, int n, int k, const int8_t *A, const int8_t *B, void *C, float *row_scale, int lda, int ldb, int ldc);
template void transformRowToFormat<COL32, 0>(char * A, char *out, int rows, int cols);
template void transformRowToFormat<COL32, 1>(char * A, char *out, int rows, int cols);
template void transformRowToFormat<COL_TURING, 0>(char * A, char *out, int rows, int cols);
template void transformRowToFormat<COL_TURING, 1>(char * A, char *out, int rows, int cols);
template void transformRowToFormat<COL_AMPERE, 0>(char * A, char *out, int rows, int cols);
template void transformRowToFormat<COL_AMPERE, 1>(char * A, char *out, int rows, int cols);
2021-10-06 02:16:20 +00:00
template void estimateQuantiles(half *A, float *code, float offset, int n);
template void estimateQuantiles(float *A, float *code, float offset, int n);
2023-04-02 23:10:35 +00:00
template void quantizeBlockwise<half, 1, General8bit>(float * code, half *A, float *absmax, unsigned char *out, float* rand, int rand_offset, int blocksize, const int n);
template void quantizeBlockwise<float, 1, General8bit>(float * code, float *A, float *absmax, unsigned char *out, float* rand, int rand_offset, int blocksize, const int n);
template void quantizeBlockwise<half, 0, General8bit>(float * code, half *A, float *absmax, unsigned char *out, float* rand, int rand_offset, int blocksize, const int n);
template void quantizeBlockwise<float, 0, General8bit>(float * code, float *A, float *absmax, unsigned char *out, float* rand, int rand_offset, int blocksize, const int n);
template void quantizeBlockwise<half, 0, FP4>(float * code, half *A, float *absmax, unsigned char *out, float* rand, int rand_offset, int blocksize, const int n);
template void quantizeBlockwise<float, 0, FP4>(float * code, float *A, float *absmax, unsigned char *out, float* rand, int rand_offset, int blocksize, const int n);
template void quantizeBlockwise<half, 0, NF4>(float * code, half *A, float *absmax, unsigned char *out, float* rand, int rand_offset, int blocksize, const int n);
template void quantizeBlockwise<float, 0, NF4>(float * code, float *A, float *absmax, unsigned char *out, float* rand, int rand_offset, int blocksize, const int n);
template void dequantizeBlockwise<half, General8bit>(float *code, unsigned char *A, float *absmax, half *out, int blocksize, const int n);
template void dequantizeBlockwise<float, General8bit>(float *code, unsigned char *A, float *absmax, float *out, int blocksize, const int n);
template void dequantizeBlockwise<half, FP4>(float *code, unsigned char *A, float *absmax, half *out, int blocksize, const int n);
template void dequantizeBlockwise<float, FP4>(float *code, unsigned char *A, float *absmax, float *out, int blocksize, const int n);
template void dequantizeBlockwise<half, NF4>(float *code, unsigned char *A, float *absmax, half *out, int blocksize, const int n);
template void dequantizeBlockwise<float, NF4>(float *code, unsigned char *A, float *absmax, float *out, int blocksize, const int n);
2021-10-06 02:16:20 +00:00
#define MAKE_optimizer32bit(name, gtype) \
template void optimizer32bit<gtype, name>(gtype* g, gtype* p, \
float* state1, float* state2, float* unorm, float max_unorm, float param_norm, \
const float beta1, const float beta2, const float eps, const float weight_decay, \
2021-10-21 01:37:44 +00:00
const int step, const float lr, const float gnorm_scale, const bool skip_zeros, const int n);
2021-10-06 02:16:20 +00:00
MAKE_optimizer32bit(ADAM, half)
MAKE_optimizer32bit(ADAM, float)
2023-04-18 01:01:49 +00:00
MAKE_optimizer32bit(ADAM, __nv_bfloat16)
2021-10-06 02:16:20 +00:00
MAKE_optimizer32bit(MOMENTUM, half)
MAKE_optimizer32bit(MOMENTUM, float)
MAKE_optimizer32bit(RMSPROP, half)
MAKE_optimizer32bit(RMSPROP, float)
2021-11-10 23:10:02 +00:00
MAKE_optimizer32bit(ADAGRAD, half)
MAKE_optimizer32bit(ADAGRAD, float)
2021-10-06 02:16:20 +00:00
#define MAKE_optimizerStatic8bit(name, gtype) \
template void optimizerStatic8bit<gtype, name>(gtype* p, gtype* g, unsigned char* state1, unsigned char* state2, \
float *unorm, float max_unorm, float param_norm, \
float beta1, float beta2, \
float eps, int step, float lr, \
float* quantiles1, float* quantiles2, \
float* max1, float* max2, float* new_max1, float* new_max2, \
float weight_decay, \
const float gnorm_scale, int n); \
MAKE_optimizerStatic8bit(ADAM, half)
MAKE_optimizerStatic8bit(ADAM, float)
MAKE_optimizerStatic8bit(MOMENTUM, half)
MAKE_optimizerStatic8bit(MOMENTUM, float)
MAKE_optimizerStatic8bit(RMSPROP, half)
MAKE_optimizerStatic8bit(RMSPROP, float)
#define MAKE_optimizerStatic8bitBlockwise(gtype, optim_name) \
template void optimizerStatic8bitBlockwise<gtype, optim_name>(gtype* p, gtype* g, \
unsigned char* state1, unsigned char* state2, float beta1, float beta2, float eps, int step, float lr, \
2021-10-21 01:37:44 +00:00
float* quantiles1, float* quantiles2, float* absmax1, float* absmax2, float weight_decay, const float gnorm_scale, bool skip_zeros, int n); \
2021-10-06 02:16:20 +00:00
MAKE_optimizerStatic8bitBlockwise(half, ADAM);
MAKE_optimizerStatic8bitBlockwise(float, ADAM);
MAKE_optimizerStatic8bitBlockwise(half, MOMENTUM);
MAKE_optimizerStatic8bitBlockwise(float, MOMENTUM);
MAKE_optimizerStatic8bitBlockwise(half, RMSPROP);
MAKE_optimizerStatic8bitBlockwise(float, RMSPROP);
2021-11-10 23:10:02 +00:00
MAKE_optimizerStatic8bitBlockwise(half, ADAGRAD);
MAKE_optimizerStatic8bitBlockwise(float, ADAGRAD);
2021-10-06 02:16:20 +00:00
2022-07-01 14:36:30 +00:00
template void percentileClipping(float * g, float *gnorm_vec, int step, const int n);
template void percentileClipping(half * g, float *gnorm_vec, int step, const int n);
2023-04-01 17:33:03 +00:00
MAKE_optimizerStatic8bitBlockwise(__nv_bfloat16, ADAM);
