CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-30 08:42:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

metal : try to utilize more of the shared memory using smaller views

Browse Source
This commit is contained in:
Georgi Gerganov
2023-06-26 22:23:04 +03:00
parent c824d2e368
commit 5cc672a9a5
2 changed files with 6 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
ggml-metal.h
View File

@@ -23,7 +23,7 @@
#include <stdbool.h>
// max memory buffers that can be mapped to the device
#define GGML_METAL_MAX_BUFFERS 16
#define GGML_METAL_MAX_BUFFERS 256
struct ggml_tensor;
struct ggml_cgraph;

8
ggml-metal.m
View File

@@ -262,8 +262,10 @@ bool ggml_metal_add_buffer(
size_aligned += (size_page - (size_aligned % size_page));
}
const size_t max_buffer_length = ctx->device.maxBufferLength/4;
// the buffer fits into the max buffer size allowed by the device
if (size_aligned <= ctx->device.maxBufferLength) {
if (size_aligned <= max_buffer_length) {
ctx->buffers[ctx->n_buffers].name = name;
ctx->buffers[ctx->n_buffers].data = data;
ctx->buffers[ctx->n_buffers].size = size;
@@ -282,8 +284,8 @@ bool ggml_metal_add_buffer(
// this overlap between the views will guarantee that the tensor with the maximum size will fully fit into
// one of the views
const size_t size_ovlp = ((max_size + size_page - 1) / size_page + 1) * size_page; // round-up 2 pages just in case
const size_t size_step = ctx->device.maxBufferLength - size_ovlp;
const size_t size_view = ctx->device.maxBufferLength;
const size_t size_step = max_buffer_length - size_ovlp;
const size_t size_view = max_buffer_length;
for (size_t i = 0; i < size; i += size_step) {
const size_t size_step_aligned = (i + size_view <= size) ? size_view : (size_aligned - i);