Lumpiasty/llama.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

ggml-meta: propagate buffer usage and call init on the new tensors (#27586)

Browse Source
This commit is contained in:
Max Krasnyansky
2026-08-26 08:27:51 +03:00
committed by GitHub
parent d222767c7a
commit 5d5cb4c3a4
3 changed files with 21 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-backend-meta.cpp
+18 -2
View File
@@ -1168,7 +1168,6 @@ static struct ggml_backend_meta_split_state ggml_backend_meta_get_split_state(
}
static struct ggml_backend_meta_split_state ggml_backend_meta_get_split_state(const struct ggml_tensor * tensor, bool assume_sync) {
GGML_ASSERT(ggml_backend_buffer_is_meta(tensor->buffer));
ggml_backend_meta_buffer_context * buf_ctx = (ggml_backend_meta_buffer_context *) tensor->buffer->context;
return ggml_backend_meta_get_split_state(buf_ctx->get_simple_tensor_container(tensor), tensor, assume_sync);
}
@@ -1259,7 +1258,14 @@ static enum ggml_status ggml_backend_meta_buffer_init_tensor_impl(ggml_backend_m
t_ij->data = (char *) ggml_backend_buffer_get_base(simple_buf)
+ size_t(tensor->data) - size_t(ggml_backend_buffer_get_base(tensor->buffer));
}
t_ij->extra = tensor->extra;
if (simple_buf) {
// the backend that owns the buffer will set .extra
ggml_backend_buffer_init_tensor(simple_buf, t_ij);
} else {
t_ij->extra = tensor->extra;
}
for (int i = 0; i < GGML_MAX_SRC; i++) {
t_ij->src[i] = tensor->src[i];
if (tensor->src[i] == tensor) {
@@ -1668,6 +1674,16 @@ bool ggml_backend_buffer_is_meta(ggml_backend_buffer_t buf) {
return buf != nullptr && buf->iface.free_buffer == ggml_backend_meta_buffer_iface.free_buffer;
}
void ggml_backend_meta_buffer_set_usage(ggml_backend_buffer_t buffer, enum ggml_backend_buffer_usage usage) {
GGML_ASSERT(ggml_backend_buffer_is_meta(buffer));
ggml_backend_meta_buffer_context * buf_ctx = (ggml_backend_meta_buffer_context *) buffer->context;
for (size_t i = 0; i < buf_ctx->bufs.size(); i++) {
if (buf_ctx->bufs[i]) {
ggml_backend_buffer_set_usage(buf_ctx->bufs[i].get(), usage);
}
}
}
static ggml_backend_buffer_t ggml_backend_meta_buffer_type_alloc_buffer(ggml_backend_buffer_type_t buft, size_t size) {
const size_t n_simple_bufts = ggml_backend_meta_buft_n_bufts(buft);