thek0tyara/llama-cpp-turboquant
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 4

llama : remove KV cache defragmentation logic (#15473)

Browse source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2025-08-22 12:22:13 +03:00 • committed by GitHub
parent ad5c975c2d
commit 9ebebef62f
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
16 changed files with 32 additions and 440 deletions
Download patch file Download diff file Expand all files Collapse all files

3
src/llama-context.cpp
View file

@ -39,7 +39,6 @@ llama_context::llama_context(
cparams.yarn_attn_factor = params.yarn_attn_factor;
cparams.yarn_beta_fast = params.yarn_beta_fast;
cparams.yarn_beta_slow = params.yarn_beta_slow;
cparams.defrag_thold = params.defrag_thold;
cparams.embeddings = params.embeddings;
cparams.offload_kqv = params.offload_kqv;
cparams.flash_attn = params.flash_attn;
@ -978,7 +977,7 @@ int llama_context::decode(const llama_batch & batch_inp) {
bool did_optimize = false;
// handle any pending defrags/shifts
// handle any pending shifts/copies
memory_update(false);
llama_memory_context_ptr mctx;