Skip to content

Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.

conditionally enable parallel pipelines

Michael Yang4561fff36e73
diff --git a/ml/backend/ggml/ggml.go b/ml/backend/ggml/ggml.go
index 03b9acb32..cc1b5936b 100644
--- a/ml/backend/ggml/ggml.go
+++ b/ml/backend/ggml/ggml.go
@@ -371,7 +371,7 @@ func New(r *os.File, params ml.BackendParams) (ml.Backend, error) {
(*C.ggml_backend_buffer_type_t)(unsafe.Pointer(&schedBufts[0])),
C.int(len(schedBackends)),
C.size_t(maxGraphNodes),
- true,
+ C._Bool(len(gpus) > 1 && slices.Contains(gpus, output.d)),
),
input: deviceBufferTypes[input.d],
output: deviceBufferTypes[output.d],