Anton Sokolchenko 39e17589a2 Update CMakeLists.txt to fix NDEBUG handling (#537) ...

without my change

| PP  | TG  | N_KV | T_PP s | S_PP t/s | T_TG s | S_TG t/s |
| --- | --- | ---- | ------ | -------- | ------ | -------- |
ggml_backend_cuda_graph_compute: disabling CUDA graphs due to mul_mat_id
ggml_backend_cuda_graph_compute: disabling CUDA graphs due to too many consecutive updates
|  8192 |   2048 |      0 |   54.433 |   150.50 |  414.061 |     4.95 |
|  8192 |   2048 |   8192 |   64.162 |   127.68 |  428.767 |     4.78 |

after my change to CMakeLists.txt

|    PP |     TG |   N_KV |   T_PP s | S_PP t/s |   T_TG s | S_TG t/s |
|-------|--------|--------|----------|----------|----------|----------|
|  8192 |   2048 |      0 |   58.363 |   140.36 |  405.040 |     5.06 |
|  8192 |   2048 |   8192 |   63.752 |   128.50 |  423.548 |     4.84 |
|  8192 |   2048 |  16384 |   69.712 |   117.51 |  431.367 |     4.75 |

2025-06-19 10:18:21 +03:00

..

cmake

Merge mainline llama.cpp (#3)

2024-07-27 07:55:01 +02:00

include

Fix non rpc build error (#506)

2025-06-08 17:27:00 +03:00

src

Update CMakeLists.txt to fix NDEBUG handling (#537)

2025-06-19 10:18:21 +03:00

.gitignore

Merge mainline llama.cpp (#3)

2024-07-27 07:55:01 +02:00

CMakeLists.txt

Better strategy for GPU offload (#520)

2025-06-12 19:25:11 +03:00