Compare commits

..

1 Commits

Author SHA1 Message Date
Georgi Gerganov 5934580905 ggml : add and use ggml_cpu_has_llamafile() 2024-07-24 11:31:41 +03:00
5 changed files with 828 additions and 754 deletions
+7 -13
View File
@@ -47,7 +47,7 @@ int main(int argc, char ** argv) {
// save state (rng, logits, embedding and kv_cache) to file
{
std::vector<uint8_t> state_mem(llama_state_get_size(ctx));
const size_t written = llama_state_get_data(ctx, state_mem.data(), state_mem.size());
const size_t written = llama_state_get_data(ctx, state_mem.data());
FILE *fp_write = fopen("dump_state.bin", "wb");
fwrite(state_mem.data(), 1, written, fp_write);
@@ -99,16 +99,13 @@ int main(int argc, char ** argv) {
// load state (rng, logits, embedding and kv_cache) from file
{
std::vector<uint8_t> state_mem;
std::vector<uint8_t> state_mem(llama_state_get_size(ctx2));
FILE * fp_read = fopen("dump_state.bin", "rb");
fseek(fp_read, 0, SEEK_END);
state_mem.resize(ftell(fp_read));
fseek(fp_read, 0, SEEK_SET);
const size_t read = fread(state_mem.data(), 1, state_mem.size(), fp_read);
fclose(fp_read);
if (read != llama_state_set_data(ctx2, state_mem.data(), state_mem.size())) {
if (read != llama_state_set_data(ctx2, state_mem.data())) {
fprintf(stderr, "\n%s : failed to read state\n", __func__);
llama_free(ctx2);
llama_free_model(model);
@@ -162,16 +159,13 @@ int main(int argc, char ** argv) {
// load state (rng, logits, embedding and kv_cache) from file
{
std::vector<uint8_t> state_mem;
std::vector<uint8_t> state_mem(llama_state_get_size(ctx3));
FILE * fp_read = fopen("dump_state.bin", "rb");
fseek(fp_read, 0, SEEK_END);
state_mem.resize(ftell(fp_read));
fseek(fp_read, 0, SEEK_SET);
const size_t read = fread(state_mem.data(), 1, state_mem.size(), fp_read);
fclose(fp_read);
if (read != llama_state_set_data(ctx3, state_mem.data(), state_mem.size())) {
if (read != llama_state_set_data(ctx3, state_mem.data())) {
fprintf(stderr, "\n%s : failed to read state\n", __func__);
llama_free(ctx3);
llama_free_model(model);
@@ -188,7 +182,7 @@ int main(int argc, char ** argv) {
{
// save kv of seq 0
std::vector<uint8_t> seq_store(llama_state_seq_get_size(ctx3, 0));
const size_t ncopy = llama_state_seq_get_data(ctx3, seq_store.data(), seq_store.size(), 0);
const size_t ncopy = llama_state_seq_get_data(ctx3, seq_store.data(), 0);
if (ncopy != seq_store.size()) {
fprintf(stderr, "\n%s : seq copy data length %zd does not match expected length %zd\n", __func__, ncopy, seq_store.size());
llama_free(ctx3);
@@ -202,7 +196,7 @@ int main(int argc, char ** argv) {
fprintf(stderr, "%s : kv cache cleared\n", __func__);
// restore kv into seq 1
const size_t nset = llama_state_seq_set_data(ctx3, seq_store.data(), seq_store.size(), 1);
const size_t nset = llama_state_seq_set_data(ctx3, seq_store.data(), 1);
if (nset != seq_store.size()) {
fprintf(stderr, "\n%s : seq set data length %zd does not match expected length %zd\n", __func__, nset, seq_store.size());
llama_free(ctx3);
+1
View File
@@ -2400,6 +2400,7 @@ extern "C" {
GGML_API int ggml_cpu_has_vsx (void);
GGML_API int ggml_cpu_has_matmul_int8(void);
GGML_API int ggml_cpu_has_cann (void);
GGML_API int ggml_cpu_has_llamafile (void);
//
// Internal types and functions exposed for tests and benchmarks
+8
View File
@@ -22005,6 +22005,14 @@ int ggml_cpu_has_cann(void) {
#endif
}
int ggml_cpu_has_llamafile(void) {
#if defined(GGML_USE_LLAMAFILE)
return 1;
#else
return 0;
#endif
}
int ggml_cpu_has_gpublas(void) {
return ggml_cpu_has_cuda() || ggml_cpu_has_vulkan() || ggml_cpu_has_kompute() || ggml_cpu_has_sycl();
}
+10 -13
View File
@@ -33,15 +33,17 @@
#define LLAMA_DEFAULT_SEED 0xFFFFFFFF
#define LLAMA_MAX_RNG_STATE (64*1024)
#define LLAMA_FILE_MAGIC_GGLA 0x67676c61u // 'ggla'
#define LLAMA_FILE_MAGIC_GGSN 0x6767736eu // 'ggsn'
#define LLAMA_FILE_MAGIC_GGSQ 0x67677371u // 'ggsq'
#define LLAMA_SESSION_MAGIC LLAMA_FILE_MAGIC_GGSN
#define LLAMA_SESSION_VERSION 8
#define LLAMA_SESSION_VERSION 7
#define LLAMA_STATE_SEQ_MAGIC LLAMA_FILE_MAGIC_GGSQ
#define LLAMA_STATE_SEQ_VERSION 2
#define LLAMA_STATE_SEQ_VERSION 1
#ifdef __cplusplus
extern "C" {
@@ -685,11 +687,10 @@ extern "C" {
// State / sessions
//
// Returns the *actual* size in bytes of the state
// (rng, logits, embedding and kv_cache)
// Only use when saving the state, not when restoring it, otherwise the size may be too small.
LLAMA_API size_t llama_state_get_size(struct llama_context * ctx);
LLAMA_API DEPRECATED(size_t llama_get_state_size(struct llama_context * ctx),
// Returns the maximum size in bytes of the state (rng, logits, embedding
// and kv_cache) - will often be smaller after compacting tokens
LLAMA_API size_t llama_state_get_size(const struct llama_context * ctx);
LLAMA_API DEPRECATED(size_t llama_get_state_size(const struct llama_context * ctx),
"use llama_state_get_size instead");
// Copies the state to the specified destination address.
@@ -697,8 +698,7 @@ extern "C" {
// Returns the number of bytes copied
LLAMA_API size_t llama_state_get_data(
struct llama_context * ctx,
uint8_t * dst,
size_t size);
uint8_t * dst);
LLAMA_API DEPRECATED(size_t llama_copy_state_data(
struct llama_context * ctx,
uint8_t * dst),
@@ -708,8 +708,7 @@ extern "C" {
// Returns the number of bytes read
LLAMA_API size_t llama_state_set_data(
struct llama_context * ctx,
const uint8_t * src,
size_t size);
const uint8_t * src);
LLAMA_API DEPRECATED(size_t llama_set_state_data(
struct llama_context * ctx,
const uint8_t * src),
@@ -751,7 +750,6 @@ extern "C" {
LLAMA_API size_t llama_state_seq_get_data(
struct llama_context * ctx,
uint8_t * dst,
size_t size,
llama_seq_id seq_id);
// Copy the sequence data (originally copied with `llama_state_seq_get_data`) into the specified sequence
@@ -761,7 +759,6 @@ extern "C" {
LLAMA_API size_t llama_state_seq_set_data(
struct llama_context * ctx,
const uint8_t * src,
size_t size,
llama_seq_id dest_seq_id);
LLAMA_API size_t llama_state_seq_save_file(
+802 -728
View File
File diff suppressed because it is too large Load Diff