sync : ggml (backend v2) (#3912)
* sync : ggml (backend v2) (wip) * sync : migrate examples and llama.cpp to dynamic graphs (wip) * sync : update tests + fix max op params to 64 ggml-ci * sync : ggml-cuda ggml-ci * llama : fix save/load state context size ggml-ci * sync : try to fix build on tvOS * sync : pass custom graph sizes in training examples * sync : update graph copies to new ggml API * sync : update sync-ggml.sh with new files * scripts : fix header in sync script * train : fix context size calculations * llama : increase inference graph size up to 4096 nodes * train : allocate grads for backward graphs * train : allocate grads for gb_tmp
This commit is contained in:
Georgi Gerganov
GitHub
parent
bb50a792ec
commit
4760e7cc0b
22 changed files with 1994 additions and 864 deletions
89
ggml.h
89
ggml.h
|
|
@ -58,7 +58,8 @@
|
|||
// {
|
||||
// ...
|
||||
//
|
||||
// struct ggml_cgraph gf = ggml_build_forward(f);
|
||||
// struct ggml_cgraph * gf = ggml_new_graph(ctx);
|
||||
// ggml_build_forward_expand(gf, f);
|
||||
//
|
||||
// // set the input variable and parameter values
|
||||
// ggml_set_f32(x, 2.0f);
|
||||
|
|
@ -213,15 +214,14 @@
|
|||
#define GGML_QNT_VERSION 2 // bump this on quantization format changes
|
||||
#define GGML_QNT_VERSION_FACTOR 1000 // do not change this
|
||||
|
||||
#define GGML_MAX_DIMS 4
|
||||
#define GGML_MAX_NODES 16384
|
||||
#define GGML_MAX_PARAMS 1024
|
||||
#define GGML_MAX_CONTEXTS 64
|
||||
#define GGML_MAX_SRC 6
|
||||
#define GGML_MAX_NAME 64
|
||||
#define GGML_MAX_OP_PARAMS 64
|
||||
#define GGML_DEFAULT_N_THREADS 4
|
||||
|
||||
#define GGML_MAX_DIMS 4
|
||||
#define GGML_MAX_PARAMS 1024
|
||||
#define GGML_MAX_CONTEXTS 64
|
||||
#define GGML_MAX_SRC 6
|
||||
#define GGML_MAX_NAME 64
|
||||
#define GGML_MAX_OP_PARAMS 64
|
||||
#define GGML_DEFAULT_N_THREADS 4
|
||||
#define GGML_DEFAULT_GRAPH_SIZE 2048
|
||||
#if UINTPTR_MAX == 0xFFFFFFFF
|
||||
#define GGML_MEM_ALIGN 4
|
||||
#else
|
||||
|
|
@ -245,7 +245,10 @@
|
|||
do { \
|
||||
if (!(x)) { \
|
||||
fprintf(stderr, "GGML_ASSERT: %s:%d: %s\n", __FILE__, __LINE__, #x); \
|
||||
abort(); \
|
||||
fflush(stderr); \
|
||||
fflush(stdout); \
|
||||
ggml_print_backtrace(); \
|
||||
exit(1); \
|
||||
} \
|
||||
} while (0)
|
||||
|
||||
|
|
@ -451,6 +454,7 @@ extern "C" {
|
|||
GGML_UNARY_OP_GELU,
|
||||
GGML_UNARY_OP_GELU_QUICK,
|
||||
GGML_UNARY_OP_SILU,
|
||||
GGML_UNARY_OP_LEAKY
|
||||
};
|
||||
|
||||
enum ggml_object_type {
|
||||
|
|
@ -531,37 +535,33 @@ extern "C" {
|
|||
|
||||
int n_threads;
|
||||
|
||||
// the `n_tasks` of nodes, 1:1 mapping to cgraph nodes
|
||||
int n_tasks[GGML_MAX_NODES];
|
||||
|
||||
// abort ggml_graph_compute when true
|
||||
bool (*abort_callback)(void * data);
|
||||
void * abort_callback_data;
|
||||
};
|
||||
|
||||
// next prime after GGML_MAX_NODES
|
||||
// #define GGML_GRAPH_HASHTABLE_SIZE 4099
|
||||
// next prime after GGML_MAX_NODES * 2 (nodes + leafs)
|
||||
// #define GGML_GRAPH_HASHTABLE_SIZE 8273
|
||||
// #define GGML_GRAPH_HASHTABLE_SIZE 16411
|
||||
#define GGML_GRAPH_HASHTABLE_SIZE 32771
|
||||
|
||||
enum ggml_cgraph_eval_order {
|
||||
GGML_CGRAPH_EVAL_ORDER_LEFT_TO_RIGHT = 0,
|
||||
GGML_CGRAPH_EVAL_ORDER_RIGHT_TO_LEFT,
|
||||
GGML_CGRAPH_EVAL_ORDER_COUNT
|
||||
};
|
||||
|
||||
struct ggml_hash_set {
|
||||
size_t size;
|
||||
struct ggml_tensor ** keys;
|
||||
};
|
||||
|
||||
// computation graph
|
||||
struct ggml_cgraph {
|
||||
int size;
|
||||
int n_nodes;
|
||||
int n_leafs;
|
||||
|
||||
struct ggml_tensor * nodes[GGML_MAX_NODES];
|
||||
struct ggml_tensor * grads[GGML_MAX_NODES];
|
||||
struct ggml_tensor * leafs[GGML_MAX_NODES];
|
||||
struct ggml_tensor ** nodes;
|
||||
struct ggml_tensor ** grads;
|
||||
struct ggml_tensor ** leafs;
|
||||
|
||||
void * visited_hash_table[GGML_GRAPH_HASHTABLE_SIZE];
|
||||
struct ggml_hash_set visited_hash_table;
|
||||
|
||||
enum ggml_cgraph_eval_order order;
|
||||
|
||||
|
|
@ -571,8 +571,6 @@ extern "C" {
|
|||
int64_t perf_time_us;
|
||||
};
|
||||
|
||||
static const size_t GGML_GRAPH_SIZE = sizeof(struct ggml_cgraph);
|
||||
|
||||
// scratch buffer
|
||||
struct ggml_scratch {
|
||||
size_t offs;
|
||||
|
|
@ -617,6 +615,8 @@ extern "C" {
|
|||
GGML_API int64_t ggml_cycles(void);
|
||||
GGML_API int64_t ggml_cycles_per_ms(void);
|
||||
|
||||
GGML_API void ggml_print_backtrace(void);
|
||||
|
||||
GGML_API void ggml_numa_init(void); // call once for better performance on NUMA systems
|
||||
GGML_API bool ggml_is_numa(void); // true if init detected that system has >1 NUMA node
|
||||
|
||||
|
|
@ -709,7 +709,7 @@ extern "C" {
|
|||
// Context tensor enumeration and lookup
|
||||
GGML_API struct ggml_tensor * ggml_get_first_tensor(struct ggml_context * ctx);
|
||||
GGML_API struct ggml_tensor * ggml_get_next_tensor (struct ggml_context * ctx, struct ggml_tensor * tensor);
|
||||
GGML_API struct ggml_tensor * ggml_get_tensor (struct ggml_context * ctx, const char * name);
|
||||
GGML_API struct ggml_tensor * ggml_get_tensor(struct ggml_context * ctx, const char * name);
|
||||
|
||||
GGML_API struct ggml_tensor * ggml_set_zero(struct ggml_tensor * tensor);
|
||||
GGML_API struct ggml_tensor * ggml_set_i32 (struct ggml_tensor * tensor, int32_t value);
|
||||
|
|
@ -943,6 +943,10 @@ extern "C" {
|
|||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a);
|
||||
|
||||
GGML_API struct ggml_tensor * ggml_leaky(
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a);
|
||||
|
||||
GGML_API struct ggml_tensor * ggml_relu_inplace(
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a);
|
||||
|
|
@ -1482,6 +1486,8 @@ extern "C" {
|
|||
int s0, // stride
|
||||
int p0); // padding
|
||||
|
||||
// the result will have 2*p0 padding for the first dimension
|
||||
// and 2*p1 padding for the second dimension
|
||||
GGML_API struct ggml_tensor * ggml_pool_2d(
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a,
|
||||
|
|
@ -1490,8 +1496,8 @@ extern "C" {
|
|||
int k1,
|
||||
int s0,
|
||||
int s1,
|
||||
int p0,
|
||||
int p1);
|
||||
float p0,
|
||||
float p1);
|
||||
|
||||
// nearest interpolate
|
||||
// used in stable-diffusion
|
||||
|
|
@ -1732,19 +1738,22 @@ extern "C" {
|
|||
GGML_API void ggml_build_forward_expand (struct ggml_cgraph * cgraph, struct ggml_tensor * tensor);
|
||||
GGML_API void ggml_build_backward_expand(struct ggml_context * ctx, struct ggml_cgraph * gf, struct ggml_cgraph * gb, bool keep);
|
||||
|
||||
GGML_API struct ggml_cgraph ggml_build_forward (struct ggml_tensor * tensor);
|
||||
GGML_API struct ggml_cgraph ggml_build_backward(struct ggml_context * ctx, struct ggml_cgraph * gf, bool keep);
|
||||
|
||||
// graph allocation in a context
|
||||
GGML_API struct ggml_cgraph * ggml_new_graph (struct ggml_context * ctx);
|
||||
GGML_API struct ggml_cgraph * ggml_build_forward_ctx(struct ggml_context * ctx, struct ggml_tensor * tensor);
|
||||
GGML_API struct ggml_cgraph * ggml_new_graph (struct ggml_context * ctx); // size = GGML_DEFAULT_GRAPH_SIZE, grads = false
|
||||
GGML_API struct ggml_cgraph * ggml_new_graph_custom (struct ggml_context * ctx, size_t size, bool grads);
|
||||
GGML_API struct ggml_cgraph * ggml_graph_dup (struct ggml_context * ctx, struct ggml_cgraph * cgraph);
|
||||
GGML_API struct ggml_cgraph * ggml_graph_view (struct ggml_context * ctx, struct ggml_cgraph * cgraph, int i0, int i1);
|
||||
GGML_API void ggml_graph_cpy (struct ggml_cgraph * src, struct ggml_cgraph * dst);
|
||||
GGML_API void ggml_graph_reset (struct ggml_cgraph * cgraph); // zero grads
|
||||
GGML_API void ggml_graph_clear (struct ggml_cgraph * cgraph);
|
||||
|
||||
GGML_API size_t ggml_graph_overhead(void);
|
||||
GGML_API size_t ggml_graph_overhead_custom(size_t size, bool grads);
|
||||
|
||||
// ggml_graph_plan() has to be called before ggml_graph_compute()
|
||||
// when plan.work_size > 0, caller must allocate memory for plan.work_data
|
||||
GGML_API struct ggml_cplan ggml_graph_plan (struct ggml_cgraph * cgraph, int n_threads /*= GGML_DEFAULT_N_THREADS*/);
|
||||
GGML_API int ggml_graph_compute(struct ggml_cgraph * cgraph, struct ggml_cplan * cplan);
|
||||
GGML_API void ggml_graph_reset (struct ggml_cgraph * cgraph);
|
||||
GGML_API int ggml_graph_compute(struct ggml_cgraph * cgraph, struct ggml_cplan * cplan);
|
||||
|
||||
// same as ggml_graph_compute() but the work data is allocated as a part of the context
|
||||
// note: the drawback of this API is that you must have ensured that the context has enough memory for the work data
|
||||
|
|
@ -1752,8 +1761,8 @@ extern "C" {
|
|||
|
||||
GGML_API struct ggml_tensor * ggml_graph_get_tensor(struct ggml_cgraph * cgraph, const char * name);
|
||||
|
||||
GGML_API void ggml_graph_export(const struct ggml_cgraph * cgraph, const char * fname);
|
||||
GGML_API struct ggml_cgraph ggml_graph_import(const char * fname, struct ggml_context ** ctx_data, struct ggml_context ** ctx_eval);
|
||||
GGML_API void ggml_graph_export(const struct ggml_cgraph * cgraph, const char * fname);
|
||||
GGML_API struct ggml_cgraph * ggml_graph_import(const char * fname, struct ggml_context ** ctx_data, struct ggml_context ** ctx_eval);
|
||||
|
||||
// print info and performance information for the graph
|
||||
GGML_API void ggml_graph_print(const struct ggml_cgraph * cgraph);
|
||||
|
|
@ -1816,6 +1825,8 @@ extern "C" {
|
|||
struct ggml_opt_params {
|
||||
enum ggml_opt_type type;
|
||||
|
||||
size_t graph_size;
|
||||
|
||||
int n_threads;
|
||||
|
||||
// delta-based convergence test
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue