metal : use shared buffers between CPU and GPU (#1696)

* Use MTLDevice.newBufferWithBytesNoCopy to share buffers between CPU and GPU

* Page-align buffers used by Metal

* Remove trailing whitespace

* Only import unistd.h for Metal builds

* metal : remove unnecessary copies

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

1 files changed, 14 insertions, 3 deletions

diff --git a/ggml-metal.m b/ggml-metal.m
index 3cb423a..82c6596 100644
--- a/ggml-metal.m
+++ b/ggml-metal.m
@@ -195,14 +195,25 @@ bool ggml_metal_add_buffer(
             }
         }
 
+        size_t page_size = getpagesize();
+        size_t aligned_size = size;
+        if ((aligned_size % page_size) != 0) {
+            aligned_size += (page_size - (aligned_size % page_size));
+        }
+
         ctx->buffers[ctx->n_buffers].name = name;
         ctx->buffers[ctx->n_buffers].data = data;
         ctx->buffers[ctx->n_buffers].size = size;
-        ctx->buffers[ctx->n_buffers].metal = [ctx->device newBufferWithBytes:data length:size options:MTLResourceStorageModeShared];
+        ctx->buffers[ctx->n_buffers].metal = [ctx->device newBufferWithBytesNoCopy:data length:aligned_size options:MTLResourceStorageModeShared deallocator:nil];
 
-        ++ctx->n_buffers;
+        if (ctx->buffers[ctx->n_buffers].metal == nil) {
+            fprintf(stderr, "%s: failed to allocate '%-16s' buffer, size = %8.2f MB\n", __func__, name, aligned_size / 1024.0 / 1024.0);
+            return false;
+        } else {
+            fprintf(stderr, "%s: allocated '%-16s' buffer, size = %8.2f MB\n", __func__, name, aligned_size / 1024.0 / 1024.0);
+        }
 
-        fprintf(stderr, "%s: allocated '%-16s' buffer, size = %8.2f MB\n", __func__, name, size / 1024.0 / 1024.0);
+        ++ctx->n_buffers;
     }
 
     return true;