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

max nodes 8192

Browse source
This commit is contained in:
Concedo 2023-11-18 11:02:35 +08:00
parent a3f708afce
commit 45ad1b97f8
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
otherarch/llama_v3.cpp
View file

@ -3457,7 +3457,7 @@ struct llama_v3_context * llama_v3_new_context_with_model(
#ifdef LLAMA_V3_USE_ALLOCATOR #ifdef LLAMA_V3_USE_ALLOCATOR
{ {
static const size_t tensor_alignment = 32; static const size_t tensor_alignment = 32;
static const size_t GGML_MAX_NODES = 4096; static const size_t GGML_MAX_NODES = 8192;
// the compute buffer is used to store the tensor and graph structs, while the allocator buffer is used for the tensor data // the compute buffer is used to store the tensor and graph structs, while the allocator buffer is used for the tensor data
ctx->buf_compute.resize(ggml_tensor_overhead()*GGML_MAX_NODES + ggml_graph_overhead()); ctx->buf_compute.resize(ggml_tensor_overhead()*GGML_MAX_NODES + ggml_graph_overhead());