Merge commit '3e18dba9fd599ace0c3f2bc94bd74e5398d2efb6' into concedo_experimental

# Conflicts:
#	CODEOWNERS
#	ggml/src/ggml-cann/aclnn_ops.cpp
#	ggml/src/ggml-cann/aclnn_ops.h
#	ggml/src/ggml-cann/common.h
#	ggml/src/ggml-cann/ggml-cann.cpp
#	scripts/sync_vendor.py
#	tests/test-backend-ops.cpp
This commit is contained in:
Concedo
2025-11-27 00:07:37 +08:00
29 changed files with 1123 additions and 188 deletions
+14 -6
View File
@@ -536,6 +536,7 @@ extern "C" {
GGML_OP_ARANGE,
GGML_OP_TIMESTEP_EMBEDDING,
GGML_OP_ARGSORT,
GGML_OP_TOP_K,
GGML_OP_LEAKY_RELU,
GGML_OP_TRI,
GGML_OP_FILL,
@@ -2283,18 +2284,25 @@ extern "C" {
struct ggml_tensor * a,
enum ggml_sort_order order);
// similar to ggml_top_k but implemented as `argsort` + `view`
GGML_API struct ggml_tensor * ggml_argsort_top_k(
struct ggml_context * ctx,
struct ggml_tensor * a,
int k);
// top k elements per row
// note: the resulting top k indices are in no particular order
GGML_API struct ggml_tensor * ggml_top_k(
struct ggml_context * ctx,
struct ggml_tensor * a,
int k);
GGML_API struct ggml_tensor * ggml_arange(
struct ggml_context * ctx,
float start,
float stop,
float step);
// top k elements per row
GGML_API struct ggml_tensor * ggml_top_k(
struct ggml_context * ctx,
struct ggml_tensor * a,
int k);
#define GGML_KQ_MASK_PAD 64
// q: [n_embd_k, n_batch, n_head, ne3 ]