metal : fix from ptr buffer name (#10189)

diff --git a/ggml/src/ggml-metal.m b/ggml/src/ggml-metal.m
index aee354cdd04168252abad104ce75c678e4a3788d..9966a9e2fb7d7bf07eb81f8669accae8ff045aa5 100644 (file)
--- a/ggml/src/ggml-metal.m
+++ b/ggml/src/ggml-metal.m
@@ -4072,7 +4072,7 @@ static ggml_backend_buffer_t ggml_backend_metal_device_buffer_from_ptr(ggml_back
         }
     }
 
-    return ggml_backend_buffer_init(ggml_backend_metal_buffer_type(), ggml_backend_metal_buffer_i, ctx, size);
+    return ggml_backend_buffer_init(ggml_backend_metal_buffer_from_ptr_type(), ggml_backend_metal_buffer_i, ctx, size);
 }
 
 static bool ggml_backend_metal_device_supports_op(ggml_backend_dev_t dev, const struct ggml_tensor * op) {
@@ -4082,7 +4082,8 @@ static bool ggml_backend_metal_device_supports_op(ggml_backend_dev_t dev, const
 }
 
 static bool ggml_backend_metal_device_supports_buft(ggml_backend_dev_t dev, ggml_backend_buffer_type_t buft) {
-    return buft->iface.get_name == ggml_backend_metal_buffer_type_get_name;
+    return buft->iface.get_name == ggml_backend_metal_buffer_type_get_name ||
+            buft->iface.get_name == ggml_backend_metal_buffer_from_ptr_type_get_name;
 
     UNUSED(dev);
 }

diff --git a/src/llama.cpp b/src/llama.cpp
index 0cdf0c073b19f5ae5f11d177c550e31dc059435e..6719edb38808f5c39307619f59b2092b446cf2b3 100644 (file)
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -9134,7 +9134,7 @@ static bool llm_load_tensors(
 
     // print memory requirements per buffer type
     for (auto & buf : model.bufs) {
-        LLAMA_LOG_INFO("%s: %10s model buffer size = %8.2f MiB\n", __func__, ggml_backend_buffer_name(buf.get()), ggml_backend_buffer_get_size(buf.get()) / 1024.0 / 1024.0);
+        LLAMA_LOG_INFO("%s: %12s model buffer size = %8.2f MiB\n", __func__, ggml_backend_buffer_name(buf.get()), ggml_backend_buffer_get_size(buf.get()) / 1024.0 / 1024.0);
     }
 
     // populate tensors_by_name