From ba4e85a8339b9dd7cdffad31838235f2fe45a8ea Mon Sep 17 00:00:00 2001 From: l3utterfly Date: Mon, 19 Jun 2023 23:20:06 +0800 Subject: [PATCH] llama : use aligned memory during ggml_init call from loading saved sessions (#1934) * fixed issue: memory is not guaranteed to be aligned properly during ggml_init call from loading saved sessions * - removed commented out old code from fix - updated another instance of same issue below original --- llama.cpp | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-) diff --git a/llama.cpp b/llama.cpp index dad31cb..4a7d01b 100644 --- a/llama.cpp +++ b/llama.cpp @@ -3126,9 +3126,7 @@ size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst) { if (kv_size) { const size_t elt_size = ggml_element_size(kv_self.k); - char buffer[4096]; - - ggml_context * cpy_ctx = ggml_init({ sizeof(buffer), buffer, /* no_alloc */ true }); + ggml_context * cpy_ctx = ggml_init({ 4096, NULL, /* no_alloc */ true }); ggml_cgraph gf{}; gf.n_threads = 1; @@ -3234,9 +3232,7 @@ size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src) { const size_t elt_size = ggml_element_size(kv_self.k); - char buffer[4096]; - - ggml_context * cpy_ctx = ggml_init({ sizeof(buffer), buffer, /* no_alloc */ true }); + ggml_context * cpy_ctx = ggml_init({ 4096, NULL, /* no_alloc */ true }); ggml_cgraph gf{}; gf.n_threads = 1;