ggml : GPU-accelerated token generation (#1412)

* CUDA kernel for q4_0 dequant. + mat. vec. mult.

* Added q4_1 via template

* Added missing __syncthreads();

* --gpu_layers -> --gpu-layers

* Shorter dequantize_mul_mat_vec line

* q5_0 dequantize_mul_mat kernel

* More readable dequantize_mul_mat_vec logic

* dequantize_mul_mat_vec kernels for q5_1, q8_0, f16

* llama : offload "output" tensor to GPU too + coding style fixes

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This commit is contained in:
Johannes Gäßler 2023-05-13 15:38:36 +02:00 committed by GitHub
parent f954edda93
commit 905d87b70a
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
8 changed files with 336 additions and 42 deletions

View file

@ -277,6 +277,12 @@ bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
params.use_color = true; params.use_color = true;
} else if (arg == "--mlock") { } else if (arg == "--mlock") {
params.use_mlock = true; params.use_mlock = true;
} else if (arg == "--gpu-layers" || arg == "-ngl" || arg == "--n-gpu-layers") {
if (++i >= argc) {
invalid_param = true;
break;
}
params.n_gpu_layers = std::stoi(argv[i]);
} else if (arg == "--no-mmap") { } else if (arg == "--no-mmap") {
params.use_mmap = false; params.use_mmap = false;
} else if (arg == "--mtest") { } else if (arg == "--mtest") {
@ -421,6 +427,8 @@ void gpt_print_usage(int /*argc*/, char ** argv, const gpt_params & params) {
if (llama_mmap_supported()) { if (llama_mmap_supported()) {
fprintf(stderr, " --no-mmap do not memory-map model (slower load but may reduce pageouts if not using mlock)\n"); fprintf(stderr, " --no-mmap do not memory-map model (slower load but may reduce pageouts if not using mlock)\n");
} }
fprintf(stderr, " -ngl N, --n-gpu-layers N\n");
fprintf(stderr, " number of layers to store in VRAM\n");
fprintf(stderr, " --mtest compute maximum memory usage\n"); fprintf(stderr, " --mtest compute maximum memory usage\n");
fprintf(stderr, " --verbose-prompt print prompt before generation\n"); fprintf(stderr, " --verbose-prompt print prompt before generation\n");
fprintf(stderr, " --lora FNAME apply LoRA adapter (implies --no-mmap)\n"); fprintf(stderr, " --lora FNAME apply LoRA adapter (implies --no-mmap)\n");
@ -463,14 +471,15 @@ std::vector<llama_token> llama_tokenize(struct llama_context * ctx, const std::s
struct llama_context * llama_init_from_gpt_params(const gpt_params & params) { struct llama_context * llama_init_from_gpt_params(const gpt_params & params) {
auto lparams = llama_context_default_params(); auto lparams = llama_context_default_params();
lparams.n_ctx = params.n_ctx; lparams.n_ctx = params.n_ctx;
lparams.n_parts = params.n_parts; lparams.n_parts = params.n_parts;
lparams.seed = params.seed; lparams.n_gpu_layers = params.n_gpu_layers;
lparams.f16_kv = params.memory_f16; lparams.seed = params.seed;
lparams.use_mmap = params.use_mmap; lparams.f16_kv = params.memory_f16;
lparams.use_mlock = params.use_mlock; lparams.use_mmap = params.use_mmap;
lparams.logits_all = params.perplexity; lparams.use_mlock = params.use_mlock;
lparams.embedding = params.embedding; lparams.logits_all = params.perplexity;
lparams.embedding = params.embedding;
llama_context * lctx = llama_init_from_file(params.model.c_str(), lparams); llama_context * lctx = llama_init_from_file(params.model.c_str(), lparams);

View file

@ -21,13 +21,14 @@
int32_t get_num_physical_cores(); int32_t get_num_physical_cores();
struct gpt_params { struct gpt_params {
int32_t seed = -1; // RNG seed int32_t seed = -1; // RNG seed
int32_t n_threads = get_num_physical_cores(); int32_t n_threads = get_num_physical_cores();
int32_t n_predict = -1; // new tokens to predict int32_t n_predict = -1; // new tokens to predict
int32_t n_parts = -1; // amount of model parts (-1 = determine from model dimensions) int32_t n_parts = -1; // amount of model parts (-1 = determine from model dimensions)
int32_t n_ctx = 512; // context size int32_t n_ctx = 512; // context size
int32_t n_batch = 512; // batch size for prompt processing (must be >=32 to use BLAS) int32_t n_batch = 512; // batch size for prompt processing (must be >=32 to use BLAS)
int32_t n_keep = 0; // number of tokens to keep from initial prompt int32_t n_keep = 0; // number of tokens to keep from initial prompt
int32_t n_gpu_layers = 0; // number of layers to store in VRAM
// sampling parameters // sampling parameters
std::unordered_map<llama_token, float> logit_bias; // logit bias for specific tokens std::unordered_map<llama_token, float> logit_bias; // logit bias for specific tokens

View file

@ -32,9 +32,15 @@ static_assert(sizeof(half) == sizeof(ggml_fp16_t), "wrong fp16 size");
} \ } \
} while (0) } while (0)
typedef void (*dequantize_kernel_t)(const void * vx, const int ib, const int iqs, float & v0, float & v1);
typedef void (*to_fp32_cuda_t)(const void * x, float * y, int k, cudaStream_t stream); typedef void (*to_fp32_cuda_t)(const void * x, float * y, int k, cudaStream_t stream);
typedef void (*dequantize_mul_mat_vec_cuda_t)(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream);
// QK = number of values after dequantization
// QR = QK / number of values before dequantization
#define QK4_0 32 #define QK4_0 32
#define QR4_0 2
typedef struct { typedef struct {
float d; // delta float d; // delta
uint8_t qs[QK4_0 / 2]; // nibbles / quants uint8_t qs[QK4_0 / 2]; // nibbles / quants
@ -42,6 +48,7 @@ typedef struct {
static_assert(sizeof(block_q4_0) == sizeof(float) + QK4_0 / 2, "wrong q4_0 block size/padding"); static_assert(sizeof(block_q4_0) == sizeof(float) + QK4_0 / 2, "wrong q4_0 block size/padding");
#define QK4_1 32 #define QK4_1 32
#define QR4_1 2
typedef struct { typedef struct {
float d; // delta float d; // delta
float m; // min float m; // min
@ -50,6 +57,7 @@ typedef struct {
static_assert(sizeof(block_q4_1) == sizeof(float) * 2 + QK4_1 / 2, "wrong q4_1 block size/padding"); static_assert(sizeof(block_q4_1) == sizeof(float) * 2 + QK4_1 / 2, "wrong q4_1 block size/padding");
#define QK5_0 32 #define QK5_0 32
#define QR5_0 2
typedef struct { typedef struct {
half d; // delta half d; // delta
uint8_t qh[4]; // 5-th bit of quants uint8_t qh[4]; // 5-th bit of quants
@ -58,6 +66,7 @@ typedef struct {
static_assert(sizeof(block_q5_0) == sizeof(ggml_fp16_t) + sizeof(uint32_t) + QK5_0 / 2, "wrong q5_0 block size/padding"); static_assert(sizeof(block_q5_0) == sizeof(ggml_fp16_t) + sizeof(uint32_t) + QK5_0 / 2, "wrong q5_0 block size/padding");
#define QK5_1 32 #define QK5_1 32
#define QR5_1 2
typedef struct { typedef struct {
half d; // delta half d; // delta
half m; // min half m; // min
@ -67,12 +76,100 @@ typedef struct {
static_assert(sizeof(block_q5_1) == 2 * sizeof(ggml_fp16_t) + sizeof(uint32_t) + QK5_1 / 2, "wrong q5_1 block size/padding"); static_assert(sizeof(block_q5_1) == 2 * sizeof(ggml_fp16_t) + sizeof(uint32_t) + QK5_1 / 2, "wrong q5_1 block size/padding");
#define QK8_0 32 #define QK8_0 32
#define QR8_0 1
typedef struct { typedef struct {
float d; // delta float d; // delta
int8_t qs[QK8_0]; // quants int8_t qs[QK8_0]; // quants
} block_q8_0; } block_q8_0;
static_assert(sizeof(block_q8_0) == sizeof(float) + QK8_0, "wrong q8_0 block size/padding"); static_assert(sizeof(block_q8_0) == sizeof(float) + QK8_0, "wrong q8_0 block size/padding");
#define CUDA_DMMV_BLOCK_SIZE 32
static __device__ void dequantize_q4_0(const void * vx, const int ib, const int iqs, float & v0, float & v1){
const block_q4_0 * x = (const block_q4_0 *) vx;
const float d = x[ib].d;
const uint8_t vui = x[ib].qs[iqs];
const int8_t vi0 = vui & 0xF;
const int8_t vi1 = vui >> 4;
v0 = (vi0 - 8)*d;
v1 = (vi1 - 8)*d;
}
static __device__ void dequantize_q4_1(const void * vx, const int ib, const int iqs, float & v0, float & v1){
const block_q4_1 * x = (const block_q4_1 *) vx;
const float d = x[ib].d;
const float m = x[ib].m;
const uint8_t vui = x[ib].qs[iqs];
const int8_t vi0 = vui & 0xF;
const int8_t vi1 = vui >> 4;
v0 = vi0*d + m;
v1 = vi1*d + m;
}
static __device__ void dequantize_q5_0(const void * vx, const int ib, const int iqs, float & v0, float & v1){
const block_q5_0 * x = (const block_q5_0 *) vx;
const float d = x[ib].d;
uint32_t qh;
memcpy(&qh, x[ib].qh, sizeof(qh));
const uint8_t xh_0 = ((qh >> (iqs + 0)) << 4) & 0x10;
const uint8_t xh_1 = ((qh >> (iqs + 12)) ) & 0x10;
const int32_t x0 = ((x[ib].qs[iqs] & 0xf) | xh_0) - 16;
const int32_t x1 = ((x[ib].qs[iqs] >> 4) | xh_1) - 16;
v0 = x0*d;
v1 = x1*d;
}
static __device__ void dequantize_q5_1(const void * vx, const int ib, const int iqs, float & v0, float & v1){
const block_q5_1 * x = (const block_q5_1 *) vx;
const float d = x[ib].d;
const float m = x[ib].m;
uint32_t qh;
memcpy(&qh, x[ib].qh, sizeof(qh));
const uint8_t xh_0 = ((qh >> (iqs + 0)) << 4) & 0x10;
const uint8_t xh_1 = ((qh >> (iqs + 12)) ) & 0x10;
const int32_t x0 = ((x[ib].qs[iqs] & 0xf) | xh_0);
const int32_t x1 = ((x[ib].qs[iqs] >> 4) | xh_1);
v0 = x0*d + m;
v1 = x1*d + m;
}
static __device__ void dequantize_q8_0(const void * vx, const int ib, const int iqs, float & v0, float & v1){
const block_q8_0 * x = (const block_q8_0 *) vx;
const float d = x[ib].d;
const int8_t vi0 = x[ib].qs[iqs + 0];
const int8_t vi1 = x[ib].qs[iqs + 1];
v0 = vi0*d;
v1 = vi1*d;
}
static __device__ void convert_f16(const void * vx, const int ib, const int iqs, float & v0, float & v1){
const half * x = (const half *) vx;
v0 = __half2float(x[ib + 0]);
v1 = __half2float(x[ib + 1]);
}
static __global__ void dequantize_block_q4_0(const void * vx, float * y) { static __global__ void dequantize_block_q4_0(const void * vx, float * y) {
static const int qk = QK4_0; static const int qk = QK4_0;
@ -173,6 +270,44 @@ static __global__ void dequantize_block_q8_0(const void * vx, float * y) {
} }
} }
template <int block_size, 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) {
const int row = blockIdx.x;
const int tid = threadIdx.x;
const int y_offset = qr == 1 ? 1 : qk/2;
__shared__ float tmp[block_size]; // separate sum for each thread
tmp[tid] = 0;
for (int i = 0; i < ncols/block_size; i += 2) {
const int col = i*block_size + 2*tid;
const int ib = (row*ncols + col)/qk; // block index
const int iqs = (col%qk)/qr; // quant index
const int iybs = col - col%qk; // y block start index
// dequantize
float v0, v1;
dequantize_kernel(vx, ib, iqs, v0, v1);
// matrix multiplication
tmp[tid] += v0 * y[iybs + iqs + 0];
tmp[tid] += v1 * y[iybs + iqs + y_offset];
}
// sum up partial sums and write back result
__syncthreads();
for (int s=block_size/2; s>0; s>>=1) {
if (tid < s) {
tmp[tid] += tmp[tid + s];
}
__syncthreads();
}
if (tid == 0) {
dst[row] = tmp[0];
}
}
static void dequantize_row_q4_0_cuda(const void * vx, float * y, int k, cudaStream_t stream) { static void dequantize_row_q4_0_cuda(const void * vx, float * y, int k, cudaStream_t stream) {
const int nb = k / QK4_0; const int nb = k / QK4_0;
dequantize_block_q4_0<<<nb, 1, 0, stream>>>(vx, y); dequantize_block_q4_0<<<nb, 1, 0, stream>>>(vx, y);
@ -198,6 +333,36 @@ static void dequantize_row_q8_0_cuda(const void * vx, float * y, int k, cudaStre
dequantize_block_q8_0<<<nb, 1, 0, stream>>>(vx, y); dequantize_block_q8_0<<<nb, 1, 0, stream>>>(vx, y);
} }
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 % CUDA_DMMV_BLOCK_SIZE == 0);
dequantize_mul_mat_vec<CUDA_DMMV_BLOCK_SIZE, QK4_0, QR4_0, dequantize_q4_0>
<<<nrows, CUDA_DMMV_BLOCK_SIZE, 0, stream>>>(vx, y, dst, ncols);
}
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 % CUDA_DMMV_BLOCK_SIZE == 0);
dequantize_mul_mat_vec<CUDA_DMMV_BLOCK_SIZE, QK4_1, QR4_1, dequantize_q4_1>
<<<nrows, CUDA_DMMV_BLOCK_SIZE, 0, stream>>>(vx, y, dst, ncols);
}
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 % CUDA_DMMV_BLOCK_SIZE == 0);
dequantize_mul_mat_vec<CUDA_DMMV_BLOCK_SIZE, QK5_0, QR5_0, dequantize_q5_0>
<<<nrows, CUDA_DMMV_BLOCK_SIZE, 0, stream>>>(vx, y, dst, ncols);
}
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 % CUDA_DMMV_BLOCK_SIZE == 0);
dequantize_mul_mat_vec<CUDA_DMMV_BLOCK_SIZE, QK5_1, QR5_1, dequantize_q5_1>
<<<nrows, CUDA_DMMV_BLOCK_SIZE, 0, stream>>>(vx, y, dst, ncols);
}
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 % CUDA_DMMV_BLOCK_SIZE == 0);
dequantize_mul_mat_vec<CUDA_DMMV_BLOCK_SIZE, QK8_0, QR8_0, dequantize_q8_0>
<<<nrows, CUDA_DMMV_BLOCK_SIZE, 0, stream>>>(vx, y, dst, ncols);
}
// TODO: optimize // TODO: optimize
static __global__ void convert_fp16_to_fp32(const void * vx, float * y) { static __global__ void convert_fp16_to_fp32(const void * vx, float * y) {
const half * x = (const half *) vx; const half * x = (const half *) vx;
@ -211,6 +376,12 @@ static void convert_fp16_to_fp32_cuda(const void * x, float * y, int k, cudaStre
convert_fp16_to_fp32<<<k, 1, 0, stream>>>(x, y); convert_fp16_to_fp32<<<k, 1, 0, stream>>>(x, y);
} }
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 % CUDA_DMMV_BLOCK_SIZE == 0);
dequantize_mul_mat_vec<CUDA_DMMV_BLOCK_SIZE, 32, 1, convert_f16>
<<<nrows, CUDA_DMMV_BLOCK_SIZE, 0, stream>>>(vx, y, dst, ncols);
}
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) {
switch (type) { switch (type) {
case GGML_TYPE_Q4_0: case GGML_TYPE_Q4_0:
@ -230,8 +401,27 @@ static to_fp32_cuda_t ggml_get_to_fp32_cuda(ggml_type type) {
} }
} }
static dequantize_mul_mat_vec_cuda_t ggml_get_dequantize_mul_mat_vec_cuda(ggml_type type) {
switch (type) {
case GGML_TYPE_Q4_0:
return dequantize_mul_mat_vec_q4_0_cuda;
case GGML_TYPE_Q4_1:
return dequantize_mul_mat_vec_q4_1_cuda;
case GGML_TYPE_Q5_0:
return dequantize_mul_mat_vec_q5_0_cuda;
case GGML_TYPE_Q5_1:
return dequantize_mul_mat_vec_q5_1_cuda;
case GGML_TYPE_Q8_0:
return dequantize_mul_mat_vec_q8_0_cuda;
case GGML_TYPE_F16:
return dequantize_mul_mat_vec_q8_0_cuda;
default:
return nullptr;
}
}
// buffer pool for cuda // buffer pool for cuda
#define MAX_CUDA_BUFFERS 16 #define MAX_CUDA_BUFFERS 256
struct scoped_spin_lock { struct scoped_spin_lock {
std::atomic_flag& lock; std::atomic_flag& lock;
@ -528,6 +718,7 @@ static void ggml_cuda_mul_mat_q_f32(const ggml_tensor * src0, const ggml_tensor
const int nb2 = dst->nb[2]; const int nb2 = dst->nb[2];
const int nb3 = dst->nb[3]; const int nb3 = dst->nb[3];
const ggml_type type = src0->type; const ggml_type type = src0->type;
const bool mul_mat_vec = ne11 == 1;
const float alpha = 1.0f; const float alpha = 1.0f;
const float beta = 0.0f; const float beta = 0.0f;
@ -538,12 +729,16 @@ static void ggml_cuda_mul_mat_q_f32(const ggml_tensor * src0, const ggml_tensor
const size_t q_sz = ggml_type_size(type) * x_ne / ggml_blck_size(type); const size_t q_sz = ggml_type_size(type) * x_ne / ggml_blck_size(type);
size_t x_size, y_size, d_size, q_size; size_t x_size, y_size, d_size, q_size;
float * d_X = (float *) ggml_cuda_pool_malloc(n_mm * sizeof(float) * x_ne, &x_size); float * d_X = nullptr;
if (!mul_mat_vec) {
d_X = (float *) ggml_cuda_pool_malloc(n_mm * sizeof(float) * x_ne, &x_size);
}
float * d_Y = (float *) ggml_cuda_pool_malloc(n_mm * sizeof(float) * y_ne, &y_size); float * d_Y = (float *) ggml_cuda_pool_malloc(n_mm * sizeof(float) * y_ne, &y_size);
float * d_D = (float *) ggml_cuda_pool_malloc(n_mm * sizeof(float) * d_ne, &d_size); float * d_D = (float *) ggml_cuda_pool_malloc(n_mm * sizeof(float) * d_ne, &d_size);
char * d_Q = (char *) ggml_cuda_pool_malloc(n_mm * q_sz, &q_size); char * d_Q = (char *) ggml_cuda_pool_malloc(n_mm * q_sz, &q_size);
const to_fp32_cuda_t to_fp32_cuda = ggml_get_to_fp32_cuda(type); const to_fp32_cuda_t to_fp32_cuda = ggml_get_to_fp32_cuda(type);
dequantize_mul_mat_vec_cuda_t dmmv = ggml_get_dequantize_mul_mat_vec_cuda(type);
GGML_ASSERT(to_fp32_cuda != nullptr); GGML_ASSERT(to_fp32_cuda != nullptr);
for (int64_t i03 = 0; i03 < ne03; i03++) { for (int64_t i03 = 0; i03 < ne03; i03++) {
@ -553,31 +748,54 @@ static void ggml_cuda_mul_mat_q_f32(const ggml_tensor * src0, const ggml_tensor
cudaStream_t cudaStream2 = g_cudaStreams2[i % GGML_CUDA_MAX_STREAMS]; cudaStream_t cudaStream2 = g_cudaStreams2[i % GGML_CUDA_MAX_STREAMS];
cudaEvent_t cudaEvent = g_cudaEvents[i % GGML_CUDA_MAX_EVENTS]; cudaEvent_t cudaEvent = g_cudaEvents[i % GGML_CUDA_MAX_EVENTS];
float * c_X = d_X + i * x_ne;
float * c_Y = d_Y + i * y_ne; float * c_Y = d_Y + i * y_ne;
float * c_D = d_D + i * d_ne; float * c_D = d_D + i * d_ne;
char * c_Q = d_Q + i * q_sz; char * c_Q = d_Q + i * q_sz;
// copy src0 and convert to fp32 on device // copy src0 to device if necessary
CUDA_CHECK(ggml_cuda_h2d_tensor_2d(c_Q, src0, i03, i02, cudaStream2)); if (src0->backend == GGML_BACKEND_CPU) {
to_fp32_cuda(c_Q, c_X, x_ne, cudaStream2); CUDA_CHECK(ggml_cuda_h2d_tensor_2d(c_Q, src0, i03, i02, cudaStream2));
CUDA_CHECK(cudaGetLastError()); } else if (src0->backend == GGML_BACKEND_CUDA) {
CUDA_CHECK(cudaEventRecord(cudaEvent, cudaStream2)); c_Q = ((char *) src0->data) + i * q_sz;
} else {
GGML_ASSERT(false);
}
if (mul_mat_vec) { // specialized dequantize_mul_mat_vec kernel
CUDA_CHECK(cudaEventRecord(cudaEvent, cudaStream2));
// copy src1 to device // copy src1 to device
CUDA_CHECK(ggml_cuda_h2d_tensor_2d(c_Y, src1, i03, i02, cudaStream)); CUDA_CHECK(ggml_cuda_h2d_tensor_2d(c_Y, src1, i03, i02, cudaStream));
// wait for conversion // wait for data
CUDA_CHECK(cudaStreamWaitEvent(cudaStream, cudaEvent, 0)); CUDA_CHECK(cudaStreamWaitEvent(cudaStream, cudaEvent, 0));
// compute // compute
CUBLAS_CHECK(cublasSetStream(g_cublasH, cudaStream)); dmmv(c_Q, c_Y, c_D, ne00, ne01, cudaStream);
CUBLAS_CHECK( CUDA_CHECK(cudaGetLastError());
cublasSgemm(g_cublasH, CUBLAS_OP_T, CUBLAS_OP_N,
ne01, ne11, ne10, } else { // general dequantization kernel + cuBLAS matrix matrix multiplication
&alpha, c_X, ne00, float * c_X = d_X + i * x_ne;
c_Y, ne10,
&beta, c_D, ne01)); // convert src0 to fp32 on device
to_fp32_cuda(c_Q, c_X, x_ne, cudaStream2);
CUDA_CHECK(cudaGetLastError());
CUDA_CHECK(cudaEventRecord(cudaEvent, cudaStream2));
// copy src1 to device
CUDA_CHECK(ggml_cuda_h2d_tensor_2d(c_Y, src1, i03, i02, cudaStream));
// wait for conversion
CUDA_CHECK(cudaStreamWaitEvent(cudaStream, cudaEvent, 0));
// compute
CUBLAS_CHECK(cublasSetStream(g_cublasH, cudaStream));
CUBLAS_CHECK(
cublasSgemm(g_cublasH, CUBLAS_OP_T, CUBLAS_OP_N,
ne01, ne11, ne10,
&alpha, c_X, ne00,
c_Y, ne10,
&beta, c_D, ne01));
}
// copy dst to host // copy dst to host
float * d = (float *) ((char *) dst->data + i02*nb2 + i03*nb3); float * d = (float *) ((char *) dst->data + i02*nb2 + i03*nb3);
@ -586,7 +804,9 @@ static void ggml_cuda_mul_mat_q_f32(const ggml_tensor * src0, const ggml_tensor
} }
CUDA_CHECK(cudaDeviceSynchronize()); CUDA_CHECK(cudaDeviceSynchronize());
ggml_cuda_pool_free(d_X, x_size); if (!mul_mat_vec) {
ggml_cuda_pool_free(d_X, x_size);
}
ggml_cuda_pool_free(d_Y, y_size); ggml_cuda_pool_free(d_Y, y_size);
ggml_cuda_pool_free(d_D, d_size); ggml_cuda_pool_free(d_D, d_size);
ggml_cuda_pool_free(d_Q, q_size); ggml_cuda_pool_free(d_Q, q_size);
@ -602,8 +822,7 @@ bool ggml_cuda_can_mul_mat(const struct ggml_tensor * src0, const struct ggml_te
if ((src0->type == GGML_TYPE_F32 || src0->type == GGML_TYPE_F16 || ggml_is_quantized(src0->type)) && if ((src0->type == GGML_TYPE_F32 || src0->type == GGML_TYPE_F16 || ggml_is_quantized(src0->type)) &&
src1->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_F32 &&
dst->type == GGML_TYPE_F32 && dst->type == GGML_TYPE_F32 &&
(ne0 >= 32 && ne1 >= 32 && ne10 >= 32)) { ((ne0 >= 32 && ne1 >= 32 && ne10 >= 32) || src0->backend == GGML_BACKEND_CUDA)) {
return true; return true;
} }
@ -655,3 +874,25 @@ size_t ggml_cuda_mul_mat_get_wsize(const struct ggml_tensor * src0, const struct
return 0; return 0;
} }
} }
void ggml_cuda_transform_tensor(ggml_tensor * tensor) {
const int64_t ne0 = tensor->ne[0];
const int64_t ne1 = tensor->ne[1];
const int64_t ne2 = tensor->ne[2];
const int64_t ne3 = tensor->ne[3];
const ggml_type type = tensor->type;
const size_t q_sz = ggml_type_size(type) * ne0 * ne1 * ne2 * ne3 / ggml_blck_size(type);
size_t q_size;
char * d_Q = (char *) ggml_cuda_pool_malloc(q_sz, &q_size);
cudaStream_t cudaStream2 = g_cudaStreams2[0];
// copy tensor to device
CUDA_CHECK(ggml_cuda_h2d_tensor_2d(d_Q, tensor, 0, 0, cudaStream2));
CUDA_CHECK(cudaDeviceSynchronize());
tensor->data = d_Q;
tensor->backend = GGML_BACKEND_CUDA;
}

View file

@ -14,6 +14,8 @@ void ggml_cuda_mul_mat(const struct ggml_tensor * src0, const struct ggml_tens
void * ggml_cuda_host_malloc(size_t size); void * ggml_cuda_host_malloc(size_t size);
void ggml_cuda_host_free(void * ptr); void ggml_cuda_host_free(void * ptr);
void ggml_cuda_transform_tensor(struct ggml_tensor * tensor);
#ifdef __cplusplus #ifdef __cplusplus
} }
#endif #endif

1
ggml.c
View file

@ -3882,6 +3882,7 @@ struct ggml_tensor * ggml_new_tensor_impl(
*result = (struct ggml_tensor) { *result = (struct ggml_tensor) {
/*.type =*/ type, /*.type =*/ type,
/*.backend =*/ GGML_BACKEND_CPU,
/*.n_dims =*/ n_dims, /*.n_dims =*/ n_dims,
/*.ne =*/ { 1, 1, 1, 1 }, /*.ne =*/ { 1, 1, 1, 1 },
/*.nb =*/ { 0, 0, 0, 0 }, /*.nb =*/ { 0, 0, 0, 0 },

8
ggml.h
View file

@ -243,6 +243,11 @@ extern "C" {
GGML_TYPE_COUNT, GGML_TYPE_COUNT,
}; };
enum ggml_backend {
GGML_BACKEND_CPU = 0,
GGML_BACKEND_CUDA = 1,
};
// model file types // model file types
enum ggml_ftype { enum ggml_ftype {
GGML_FTYPE_UNKNOWN = -1, GGML_FTYPE_UNKNOWN = -1,
@ -333,6 +338,7 @@ extern "C" {
// n-dimensional tensor // n-dimensional tensor
struct ggml_tensor { struct ggml_tensor {
enum ggml_type type; enum ggml_type type;
enum ggml_backend backend;
int n_dims; int n_dims;
int64_t ne[GGML_MAX_DIMS]; // number of elements int64_t ne[GGML_MAX_DIMS]; // number of elements
@ -363,7 +369,7 @@ extern "C" {
char name[32]; char name[32];
char padding[8]; // TODO: remove and add padding to name? char padding[9]; // TODO: remove and add padding to name?
}; };
// computation graph // computation graph

View file

@ -9,6 +9,9 @@
#include "llama.h" #include "llama.h"
#include "ggml.h" #include "ggml.h"
#ifdef GGML_USE_CUBLAS
#include "ggml-cuda.h"
#endif
#include <array> #include <array>
#include <ctime> #include <ctime>
@ -810,6 +813,7 @@ struct llama_context_params llama_context_default_params() {
struct llama_context_params result = { struct llama_context_params result = {
/*.n_ctx =*/ 512, /*.n_ctx =*/ 512,
/*.n_parts =*/ -1, /*.n_parts =*/ -1,
/*.gpu_layers =*/ 0,
/*.seed =*/ -1, /*.seed =*/ -1,
/*.f16_kv =*/ false, /*.f16_kv =*/ false,
/*.logits_all =*/ false, /*.logits_all =*/ false,
@ -876,6 +880,7 @@ static void llama_model_load_internal(
const std::string & fname, const std::string & fname,
llama_context & lctx, llama_context & lctx,
int n_ctx, int n_ctx,
int n_gpu_layers,
ggml_type memory_type, ggml_type memory_type,
bool use_mmap, bool use_mmap,
bool use_mlock, bool use_mlock,
@ -1022,6 +1027,33 @@ static void llama_model_load_internal(
ml->load_all_data(progress_callback, progress_callback_user_data, use_mlock ? &lctx.model.mlock_mmap : NULL); ml->load_all_data(progress_callback, progress_callback_user_data, use_mlock ? &lctx.model.mlock_mmap : NULL);
model.mapping = std::move(ml->mapping); model.mapping = std::move(ml->mapping);
#ifdef GGML_USE_CUBLAS
{
const int n_gpu = std::min(n_gpu_layers, int(hparams.n_layer));
fprintf(stderr, "%s: [cublas] offloading %d layers to GPU\n", __func__, n_gpu);
size_t vram_total = 0;
for (int i = 0; i < n_gpu; ++i) {
const auto & layer = model.layers[i];
ggml_cuda_transform_tensor(layer.wq); vram_total += ggml_nbytes(layer.wq);
ggml_cuda_transform_tensor(layer.wk); vram_total += ggml_nbytes(layer.wk);
ggml_cuda_transform_tensor(layer.wv); vram_total += ggml_nbytes(layer.wv);
ggml_cuda_transform_tensor(layer.wo); vram_total += ggml_nbytes(layer.wo);
ggml_cuda_transform_tensor(layer.w1); vram_total += ggml_nbytes(layer.w1);
ggml_cuda_transform_tensor(layer.w2); vram_total += ggml_nbytes(layer.w2);
ggml_cuda_transform_tensor(layer.w3); vram_total += ggml_nbytes(layer.w3);
}
if (n_gpu_layers > (int) hparams.n_layer) {
fprintf(stderr, "%s: [cublas] offloading output layer to GPU\n", __func__);
ggml_cuda_transform_tensor(model.output); vram_total += ggml_nbytes(model.output);
}
fprintf(stderr, "%s: [cublas] total VRAM used: %zu MB\n", __func__, vram_total / 1024 / 1024);
}
#endif
// loading time will be recalculate after the first eval, so // loading time will be recalculate after the first eval, so
// we take page faults deferred by mmap() into consideration // we take page faults deferred by mmap() into consideration
@ -1032,6 +1064,7 @@ static bool llama_model_load(
const std::string & fname, const std::string & fname,
llama_context & lctx, llama_context & lctx,
int n_ctx, int n_ctx,
int n_gpu_layers,
ggml_type memory_type, ggml_type memory_type,
bool use_mmap, bool use_mmap,
bool use_mlock, bool use_mlock,
@ -1039,7 +1072,7 @@ static bool llama_model_load(
llama_progress_callback progress_callback, llama_progress_callback progress_callback,
void *progress_callback_user_data) { void *progress_callback_user_data) {
try { try {
llama_model_load_internal(fname, lctx, n_ctx, memory_type, use_mmap, use_mlock, llama_model_load_internal(fname, lctx, n_ctx, n_gpu_layers, memory_type, use_mmap, use_mlock,
vocab_only, progress_callback, progress_callback_user_data); vocab_only, progress_callback, progress_callback_user_data);
return true; return true;
} catch (const std::string & err) { } catch (const std::string & err) {
@ -2111,7 +2144,7 @@ struct llama_context * llama_init_from_file(
ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32; ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
if (!llama_model_load(path_model, *ctx, params.n_ctx, memory_type, if (!llama_model_load(path_model, *ctx, params.n_ctx, params.n_gpu_layers, memory_type,
params.use_mmap, params.use_mlock, params.vocab_only, params.use_mmap, params.use_mlock, params.vocab_only,
params.progress_callback, params.progress_callback_user_data)) { params.progress_callback, params.progress_callback_user_data)) {
fprintf(stderr, "%s: failed to load model\n", __func__); fprintf(stderr, "%s: failed to load model\n", __func__);

View file

@ -54,9 +54,10 @@ extern "C" {
typedef void (*llama_progress_callback)(float progress, void *ctx); typedef void (*llama_progress_callback)(float progress, void *ctx);
struct llama_context_params { struct llama_context_params {
int n_ctx; // text context int n_ctx; // text context
int n_parts; // -1 for default int n_parts; // -1 for default
int seed; // RNG seed, -1 for random int n_gpu_layers; // number of layers to store in VRAM
int seed; // RNG seed, -1 for random
bool f16_kv; // use fp16 for KV cache bool f16_kv; // use fp16 for KV cache
bool logits_all; // the llama_eval() call computes all logits, not just the last one bool logits_all; // the llama_eval() call computes all logits, not just the last one