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

ggml : remove ggml_cplan + rework ggml_cgraph

Browse source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2024-09-11 13:05:10 +03:00
parent ee154457dd
commit 119e0bc9ae
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
10 changed files with 248 additions and 175 deletions
Download patch file Download diff file Expand all files Collapse all files

18
tests/test-rope.cpp
View file

@ -112,17 +112,6 @@ static struct ggml_tensor * get_random_tensor_f32(
return result;
}
static void ggml_graph_compute_helper(std::vector<uint8_t> & buf, ggml_cgraph * graph, int n_threads) {
struct ggml_cplan plan = ggml_graph_plan(graph, n_threads, nullptr);
if (plan.work_size > 0) {
buf.resize(plan.work_size);
plan.work_data = buf.data();
}
ggml_graph_compute(graph, &plan);
}
int main(int /*argc*/, const char ** /*argv*/) {
struct ggml_init_params params = {
/* .mem_size = */ 128*1024*1024,
@ -130,8 +119,6 @@ int main(int /*argc*/, const char ** /*argv*/) {
/* .no_alloc = */ false,
};
std::vector<uint8_t> work_buffer;
struct ggml_context * ctx0 = ggml_init(params);
struct ggml_tensor * x;
@ -175,7 +162,10 @@ int main(int /*argc*/, const char ** /*argv*/) {
ggml_build_forward_expand(gf, r1);
ggml_build_forward_expand(gf, r2);
ggml_graph_compute_helper(work_buffer, gf, 4);
ggml_graph_prepare(gf, 4, nullptr);
ggml_graph_work_init(gf, nullptr);
ggml_graph_compute(gf);
ggml_graph_work_free(gf);
// check that r1 and r2 are the same
{