diff options
| author | Henri Vasserman <henv@hot.ee> | 2023-06-05 13:43:08 +0300 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2023-06-05 13:43:08 +0300 |
| commit | 5220a991a5e92bddad9542267ab445a2c033681c (patch) | |
| tree | 6881759a7aaa573a5c5e70b24a0516bfd060b94a | |
| parent | d1f563a743a83dabc11e125d4a7d64189c16498c (diff) | |
Increase 3B scratch buffers. (#1698)
The 128 MB was too optimistic.
Too bad it is not dynamically computed.
| -rw-r--r-- | llama.cpp | 4 |
1 files changed, 2 insertions, 2 deletions
@@ -63,7 +63,7 @@ static const size_t MB = 1024*1024; static const std::map<e_model, size_t> & MEM_REQ_SCRATCH0() { static std::map<e_model, size_t> k_sizes = { - { MODEL_3B, 128ull * MB }, + { MODEL_3B, 256ull * MB }, { MODEL_7B, 512ull * MB }, { MODEL_13B, 512ull * MB }, { MODEL_30B, 512ull * MB }, @@ -75,7 +75,7 @@ static const std::map<e_model, size_t> & MEM_REQ_SCRATCH0() static const std::map<e_model, size_t> & MEM_REQ_SCRATCH1() { static std::map<e_model, size_t> k_sizes = { - { MODEL_3B, 128ull * MB }, + { MODEL_3B, 256ull * MB }, { MODEL_7B, 512ull * MB }, { MODEL_13B, 512ull * MB }, { MODEL_30B, 512ull * MB }, |