tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-08-03 07:47:40 -04:00
Code Issues Packages Projects Releases Wiki Activity

llama : add high-throughput mode (#14363)

Browse Source 
* kv-cache : prepare K/V buffers for separation

ggml-ci

* batched-bench : fix oob write

ggml-ci

* llama : add "virtual sequences"

ggml-ci

* llama : use "stream" vs "virtual sequence"

ggml-ci

* graph : fix stream splitting when KV cache is not used

ggml-ci

* kv-cache : add multi-stream save/load support

ggml-ci

* llama : add "--attn-streams" flag

ggml-ci

* kv-cache : fix handling when find_slot fails

ggml-ci

* kv-cache : restore find_slot impl

ggml-ci

* kv-cache : add comments

* kv-cache : add bounds checks for sequence id

ggml-ci

* cont : add n_seq_max to batch allocr

ggml-ci

* kv-cache : perform stream copies lazily after llama_synchronize

ggml-ci

* kv-cache : avoid throwing exceptions across the C boundary

ggml-ci

* CUDA: 4D FlashAttention support (#14628)

* CUDA: 4D FlashAttention support

* CUDA: fix WMMA FA kernel

* llama : rename attn_streams -> kv_unified

ggml-ci

* common : rename kv_split -> kv_unified

ggml-ci

---------

Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
This commit is contained in:
Georgi Gerganov
2025-07-16 16:35:42 +03:00
committed by GitHub
parent ab14019821
commit 225e7a1438
30 changed files with 1080 additions and 460 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
src/llama-graph.h
View File

@@ -255,10 +255,10 @@ public:
ggml_tensor * get_kq_mask() const { return self_kq_mask_cnv; }
ggml_tensor * self_k_idxs = nullptr; // I64 [n_batch]
ggml_tensor * self_v_idxs = nullptr; // I64 [n_batch]
ggml_tensor * self_v_idxs = nullptr; // I64 [n_batch] or [n_batch*n_embd_v_gqa]
ggml_tensor * self_kq_mask = nullptr; // F32 [n_kv, n_batch, 1, 1]
ggml_tensor * self_kq_mask_cnv = nullptr; // [n_kv, n_batch, 1, 1]
ggml_tensor * self_kq_mask = nullptr; // F32 [n_kv, n_batch/n_stream, 1, n_stream]
ggml_tensor * self_kq_mask_cnv = nullptr; // [n_kv, n_batch/n_stream, 1, n_stream]
const llama_hparams & hparams;
const llama_cparams & cparams;
@@ -289,14 +289,14 @@ public:
ggml_tensor * get_kq_mask_swa() const { return self_kq_mask_swa_cnv; }
ggml_tensor * self_k_idxs = nullptr; // I64 [n_batch]
ggml_tensor * self_v_idxs = nullptr; // I64 [n_batch]
ggml_tensor * self_v_idxs = nullptr; // I64 [n_batch] or [n_batch*n_embd_v_gqa]
ggml_tensor * self_k_idxs_swa = nullptr; // I64 [n_batch]
ggml_tensor * self_v_idxs_swa = nullptr; // I64 [n_batch]
ggml_tensor * self_v_idxs_swa = nullptr; // I64 [n_batch] or [n_batch*n_embd_v_gqa]
ggml_tensor * self_kq_mask = nullptr; // F32 [n_kv, n_batch, 1, 1]
ggml_tensor * self_kq_mask_cnv = nullptr; // [n_kv, n_batch, 1, 1]
ggml_tensor * self_kq_mask_swa = nullptr; // F32 [n_kv, n_batch, 1, 1]
ggml_tensor * self_kq_mask_swa_cnv = nullptr; // [n_kv, n_batch, 1, 1]
ggml_tensor * self_kq_mask = nullptr; // F32 [n_kv, n_batch/n_stream, 1, n_stream]
ggml_tensor * self_kq_mask_cnv = nullptr; // [n_kv, n_batch/n_stream, 1, n_stream]
ggml_tensor * self_kq_mask_swa = nullptr; // F32 [n_kv, n_batch/n_stream, 1, n_stream]
ggml_tensor * self_kq_mask_swa_cnv = nullptr; // [n_kv, n_batch/n_stream, 1, n_stream]
const llama_hparams & hparams;
const llama_cparams & cparams;