vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Fixed LLAMA_CUDA_DMMV_Y > 1 for WizardLM

Browse source
This commit is contained in:
JohannesGaessler 2023-06-14 16:36:46 +02:00
parent a47072b85d
commit b783da97a6
1 changed files with 55 additions and 30 deletions
Download patch file Download diff file Expand all files Collapse all files

85
ggml-cuda.cu
View file

@ -660,10 +660,15 @@ static __global__ void dequantize_block(const void * vx, float * y, const int k)
} }
template <int qk, int qr, dequantize_kernel_t dequantize_kernel> template <int qk, int qr, dequantize_kernel_t dequantize_kernel>
static __global__ void dequantize_mul_mat_vec(const void * vx, const float * y, float * dst, const int ncols) { static __global__ void dequantize_mul_mat_vec(const void * vx, const float * y, float * dst, const int ncols, const int nrows) {
// qk = quantized weights per x block // qk = quantized weights per x block
// qr = number of quantized weights per data value in x block // qr = number of quantized weights per data value in x block
const int row = blockIdx.x*blockDim.y + threadIdx.y; const int row = blockIdx.y*blockDim.y + threadIdx.y;
if (row >= nrows) {
return;
}
const int tid = threadIdx.x; const int tid = threadIdx.x;
const int iter_stride = 2*GGML_CUDA_DMMV_X; const int iter_stride = 2*GGML_CUDA_DMMV_X;
@ -708,8 +713,13 @@ static __global__ void dequantize_mul_mat_vec(const void * vx, const float * y,
} }
template <int n_thread, dot_kernel_k_t dot_kernel> template <int n_thread, dot_kernel_k_t dot_kernel>
static __global__ void dequantize_mul_mat_vec_k(const void * vx, const float * y, float * dst, const int ncols) { static __global__ void dequantize_mul_mat_vec_k(const void * vx, const float * y, float * dst, const int ncols, const int nrows) {
const int row = blockIdx.x*blockDim.y + threadIdx.y; const int row = blockIdx.y*blockDim.y + threadIdx.y;
if (row >= nrows) {
return;
}
const int tid = threadIdx.x; const int tid = threadIdx.x;
const int iter_stride = QK_K; const int iter_stride = QK_K;
@ -1035,73 +1045,92 @@ static void dequantize_row_q6_K_cuda(const void * vx, float * y, const int k, cu
static void dequantize_mul_mat_vec_q4_0_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) { static void dequantize_mul_mat_vec_q4_0_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
GGML_ASSERT(ncols % GGML_CUDA_DMMV_X == 0); GGML_ASSERT(ncols % GGML_CUDA_DMMV_X == 0);
GGML_ASSERT(nrows % GGML_CUDA_DMMV_Y == 0); const int block_num_y = (nrows + GGML_CUDA_DMMV_Y - 1) / GGML_CUDA_DMMV_Y;
const dim3 block_nums(1, block_num_y, 1);
const dim3 block_dims(WARP_SIZE, GGML_CUDA_DMMV_Y, 1); const dim3 block_dims(WARP_SIZE, GGML_CUDA_DMMV_Y, 1);
dequantize_mul_mat_vec<QK4_0, QR4_0, dequantize_q4_0> dequantize_mul_mat_vec<QK4_0, QR4_0, dequantize_q4_0>
<<<nrows/GGML_CUDA_DMMV_Y, block_dims, 0, stream>>>(vx, y, dst, ncols); <<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
} }
static void dequantize_mul_mat_vec_q4_1_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) { static void dequantize_mul_mat_vec_q4_1_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
GGML_ASSERT(ncols % GGML_CUDA_DMMV_X == 0); GGML_ASSERT(ncols % GGML_CUDA_DMMV_X == 0);
GGML_ASSERT(nrows % GGML_CUDA_DMMV_Y == 0); const int block_num_y = (nrows + GGML_CUDA_DMMV_Y - 1) / GGML_CUDA_DMMV_Y;
const dim3 block_nums(1, block_num_y, 1);
const dim3 block_dims(WARP_SIZE, GGML_CUDA_DMMV_Y, 1); const dim3 block_dims(WARP_SIZE, GGML_CUDA_DMMV_Y, 1);
dequantize_mul_mat_vec<QK4_1, QR4_1, dequantize_q4_1> dequantize_mul_mat_vec<QK4_1, QR4_1, dequantize_q4_1>
<<<nrows/GGML_CUDA_DMMV_Y, block_dims, 0, stream>>>(vx, y, dst, ncols); <<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
} }
static void dequantize_mul_mat_vec_q5_0_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) { static void dequantize_mul_mat_vec_q5_0_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
GGML_ASSERT(ncols % GGML_CUDA_DMMV_X == 0); GGML_ASSERT(ncols % GGML_CUDA_DMMV_X == 0);
GGML_ASSERT(nrows % GGML_CUDA_DMMV_Y == 0); const int block_num_y = (nrows + GGML_CUDA_DMMV_Y - 1) / GGML_CUDA_DMMV_Y;
const dim3 block_nums(1, block_num_y, 1);
const dim3 block_dims(WARP_SIZE, GGML_CUDA_DMMV_Y, 1); const dim3 block_dims(WARP_SIZE, GGML_CUDA_DMMV_Y, 1);
dequantize_mul_mat_vec<QK5_0, QR5_0, dequantize_q5_0> dequantize_mul_mat_vec<QK5_0, QR5_0, dequantize_q5_0>
<<<nrows/GGML_CUDA_DMMV_Y, block_dims, 0, stream>>>(vx, y, dst, ncols); <<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
} }
static void dequantize_mul_mat_vec_q5_1_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) { static void dequantize_mul_mat_vec_q5_1_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
GGML_ASSERT(ncols % GGML_CUDA_DMMV_X == 0); GGML_ASSERT(ncols % GGML_CUDA_DMMV_X == 0);
GGML_ASSERT(nrows % GGML_CUDA_DMMV_Y == 0); const int block_num_y = (nrows + GGML_CUDA_DMMV_Y - 1) / GGML_CUDA_DMMV_Y;
const dim3 block_nums(1, block_num_y, 1);
const dim3 block_dims(WARP_SIZE, GGML_CUDA_DMMV_Y, 1); const dim3 block_dims(WARP_SIZE, GGML_CUDA_DMMV_Y, 1);
dequantize_mul_mat_vec<QK5_1, QR5_1, dequantize_q5_1> dequantize_mul_mat_vec<QK5_1, QR5_1, dequantize_q5_1>
<<<nrows/GGML_CUDA_DMMV_Y, block_dims, 0, stream>>>(vx, y, dst, ncols); <<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
} }
static void dequantize_mul_mat_vec_q8_0_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) { static void dequantize_mul_mat_vec_q8_0_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
GGML_ASSERT(ncols % GGML_CUDA_DMMV_X == 0); GGML_ASSERT(ncols % GGML_CUDA_DMMV_X == 0);
GGML_ASSERT(nrows % GGML_CUDA_DMMV_Y == 0); const int block_num_y = (nrows + GGML_CUDA_DMMV_Y - 1) / GGML_CUDA_DMMV_Y;
const dim3 block_nums(1, block_num_y, 1);
const dim3 block_dims(WARP_SIZE, GGML_CUDA_DMMV_Y, 1); const dim3 block_dims(WARP_SIZE, GGML_CUDA_DMMV_Y, 1);
dequantize_mul_mat_vec<QK8_0, QR8_0, dequantize_q8_0> dequantize_mul_mat_vec<QK8_0, QR8_0, dequantize_q8_0>
<<<nrows/GGML_CUDA_DMMV_Y, block_dims, 0, stream>>>(vx, y, dst, ncols); <<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
} }
static void dequantize_mul_mat_vec_q2_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) { static void dequantize_mul_mat_vec_q2_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
GGML_ASSERT(ncols % QK_K == 0); GGML_ASSERT(ncols % QK_K == 0);
const int ny = 2; const int ny = 2;
const int block_num_y = (nrows + ny - 1) / ny;
const dim3 block_nums(1, block_num_y, 1);
const dim3 block_dims(32, ny, 1); const dim3 block_dims(32, ny, 1);
dequantize_mul_mat_vec_k<32, vec_dot_q2_K><<<(nrows + ny - 1)/ny, block_dims, 0, stream>>>(vx, y, dst, ncols); dequantize_mul_mat_vec_k<32, vec_dot_q2_K><<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
} }
static void dequantize_mul_mat_vec_q3_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) { static void dequantize_mul_mat_vec_q3_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
GGML_ASSERT(ncols % QK_K == 0); GGML_ASSERT(ncols % QK_K == 0);
const dim3 block_dims(32, 2, 1); const int ny = 2;
dequantize_mul_mat_vec_k<32, vec_dot_q3_K><<<nrows/2, block_dims, 0, stream>>>(vx, y, dst, ncols); const int block_num_y = (nrows + ny - 1) / ny;
const dim3 block_nums(1, block_num_y, 1);
const dim3 block_dims(32, ny, 1);
dequantize_mul_mat_vec_k<32, vec_dot_q3_K><<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
} }
static void dequantize_mul_mat_vec_q4_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) { static void dequantize_mul_mat_vec_q4_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
GGML_ASSERT(ncols % QK_K == 0); GGML_ASSERT(ncols % QK_K == 0);
const dim3 block_dims(32, 2, 1); const int ny = 2;
dequantize_mul_mat_vec_k<32, vec_dot_q4_K><<<nrows/2, block_dims, 0, stream>>>(vx, y, dst, ncols); const int block_num_y = (nrows + ny - 1) / ny;
const dim3 block_nums(1, block_num_y, 1);
const dim3 block_dims(32, ny, 1);
dequantize_mul_mat_vec_k<32, vec_dot_q4_K><<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
} }
static void dequantize_mul_mat_vec_q5_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) { static void dequantize_mul_mat_vec_q5_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
GGML_ASSERT(ncols % QK_K == 0); GGML_ASSERT(ncols % QK_K == 0);
const dim3 block_dims(32, 2, 1); const int ny = 2;
dequantize_mul_mat_vec_k<32, vec_dot_q5_K><<<nrows/2, block_dims, 0, stream>>>(vx, y, dst, ncols); const int block_num_y = (nrows + ny - 1) / ny;
const dim3 block_nums(1, block_num_y, 1);
const dim3 block_dims(32, ny, 1);
dequantize_mul_mat_vec_k<32, vec_dot_q5_K><<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
} }
static void dequantize_mul_mat_vec_q6_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) { static void dequantize_mul_mat_vec_q6_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
GGML_ASSERT(ncols % QK_K == 0); GGML_ASSERT(ncols % QK_K == 0);
const dim3 block_dims(32, 2, 1); const int ny = 2;
dequantize_mul_mat_vec_k<32, vec_dot_q6_K><<<nrows/2, block_dims, 0, stream>>>(vx, y, dst, ncols); const int block_num_y = (nrows + ny - 1) / ny;
const dim3 block_nums(1, block_num_y, 1);
const dim3 block_dims(32, ny, 1);
dequantize_mul_mat_vec_k<32, vec_dot_q6_K><<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
} }
static void convert_fp16_to_fp32_cuda(const void * vx, float * y, const int k, cudaStream_t stream) { static void convert_fp16_to_fp32_cuda(const void * vx, float * y, const int k, cudaStream_t stream) {
@ -1111,10 +1140,11 @@ static void convert_fp16_to_fp32_cuda(const void * vx, float * y, const int k, c
static void convert_mul_mat_vec_f16_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) { static void convert_mul_mat_vec_f16_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
GGML_ASSERT(ncols % GGML_CUDA_DMMV_X == 0); GGML_ASSERT(ncols % GGML_CUDA_DMMV_X == 0);
GGML_ASSERT(nrows % GGML_CUDA_DMMV_Y == 0); const int block_num_y = (nrows + GGML_CUDA_DMMV_Y - 1) / GGML_CUDA_DMMV_Y;
const dim3 block_nums(1, block_num_y, 1);
const dim3 block_dims(WARP_SIZE, GGML_CUDA_DMMV_Y, 1); const dim3 block_dims(WARP_SIZE, GGML_CUDA_DMMV_Y, 1);
dequantize_mul_mat_vec<1, 1, convert_f16> dequantize_mul_mat_vec<1, 1, convert_f16>
<<<nrows/GGML_CUDA_DMMV_Y, block_dims, 0, stream>>>(vx, y, dst, ncols); <<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
} }
static to_fp32_cuda_t ggml_get_to_fp32_cuda(ggml_type type) { static to_fp32_cuda_t ggml_get_to_fp32_cuda(ggml_type type) {
@ -1798,9 +1828,7 @@ static void ggml_cuda_op(const ggml_tensor * src0, const ggml_tensor * src1, ggm
int64_t row_low, row_high; int64_t row_low, row_high;
if (split) { if (split) {
row_low = id == 0 ? 0 : nrows0*g_tensor_split[id]; row_low = id == 0 ? 0 : nrows0*g_tensor_split[id];
row_low -= row_low % GGML_CUDA_DMMV_Y;
row_high = id == g_device_count - 1 ? nrows0 : nrows0*g_tensor_split[id + 1]; row_high = id == g_device_count - 1 ? nrows0 : nrows0*g_tensor_split[id + 1];
row_high -= row_high % GGML_CUDA_DMMV_Y;
} else { } else {
row_low = 0; row_low = 0;
row_high = nrows0; row_high = nrows0;
@ -2223,10 +2251,7 @@ void ggml_cuda_transform_tensor(void * data, struct ggml_tensor * tensor) {
row_high = nrows; row_high = nrows;
} else if (backend == GGML_BACKEND_GPU_SPLIT) { } else if (backend == GGML_BACKEND_GPU_SPLIT) {
row_low = id == 0 ? 0 : nrows*g_tensor_split[id]; row_low = id == 0 ? 0 : nrows*g_tensor_split[id];
row_low -= row_low % GGML_CUDA_DMMV_Y;
row_high = id == g_device_count - 1 ? nrows : nrows*g_tensor_split[id + 1]; row_high = id == g_device_count - 1 ? nrows : nrows*g_tensor_split[id + 1];
row_high -= row_high % GGML_CUDA_DMMV_Y;
GGML_ASSERT(nrows % GGML_CUDA_DMMV_Y == 0);
} else { } else {
GGML_ASSERT(false); GGML_ASSERT(false);
} }