root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-31 22:04:35 +00:00
Code Issues Actions 7 Packages Projects Releases Wiki Activity

llama : disable MPI for now

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2023-09-20 14:06:41 +03:00
parent e04dc51988
commit 54206962c7
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
1 changed files with 6 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
llama.cpp
View File

@ -4072,7 +4072,8 @@ static int llama_decode_internal(
#ifdef GGML_USE_MPI #ifdef GGML_USE_MPI
// TODO: needs fix after #3228 // TODO: needs fix after #3228
ggml_mpi_eval_init(lctx.ctx_mpi, &n_tokens, &n_past, &n_threads); GGML_ASSERT(false && "not implemented");
//ggml_mpi_eval_init(lctx.ctx_mpi, &n_tokens, &n_past, &n_threads);
#endif #endif
GGML_ASSERT(n_threads > 0); GGML_ASSERT(n_threads > 0);
@ -6846,8 +6847,10 @@ struct llama_context * llama_new_context_with_model(
if (ggml_mpi_rank(ctx->ctx_mpi) > 0) { if (ggml_mpi_rank(ctx->ctx_mpi) > 0) {
// Enter a blocking eval loop with dummy input, letting rank=0 drive the process // Enter a blocking eval loop with dummy input, letting rank=0 drive the process
const std::vector<llama_token> tmp(ctx->model.hparams.n_ctx, llama_token_bos(ctx)); // TODO: needs fix after #3228
while (!llama_eval(ctx, tmp.data(), tmp.size(), 0, 0)) {}; GGML_ASSERT(false && "not implemented");
//const std::vector<llama_token> tmp(ctx->model.hparams.n_ctx, llama_token_bos(ctx));
//while (!llama_eval(ctx, tmp.data(), tmp.size(), 0, 0)) {};
llama_backend_free(); llama_backend_free();
exit(1); exit(1);
} }