Skip to content

Commit 12dcddc

Browse files
committed
kompute : disable LLAMA_SPLIT_LAYER after ggml-org#5321
1 parent 6ff4387 commit 12dcddc

File tree

1 file changed

+4
-1
lines changed

1 file changed

+4
-1
lines changed

src/llama.cpp

Lines changed: 4 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -5972,6 +5972,7 @@ static bool llm_load_tensors(
59725972
model.buft_layer[i] = llama_default_buffer_type_cpu(true);
59735973
}
59745974

5975+
#ifndef GGML_USE_KOMPUTE
59755976
if (split_mode == LLAMA_SPLIT_MODE_LAYER) {
59765977
// calculate the split points
59775978
int device_count = llama_get_device_count(model);
@@ -6009,7 +6010,9 @@ static bool llm_load_tensors(
60096010
} else {
60106011
model.buft_output = llama_default_buffer_type_cpu(true);
60116012
}
6012-
} else {
6013+
} else
6014+
#endif
6015+
{
60136016
ggml_backend_buffer_type_t split_buft;
60146017
if (split_mode == LLAMA_SPLIT_MODE_ROW) {
60156018
split_buft = llama_default_buffer_type_split(model, main_gpu, tensor_split);

0 commit comments

Comments
 (0)