Skip to content

sync : ggml #12881

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 7 commits into from
Apr 10, 2025
Merged

sync : ggml #12881

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
94ce580
scripts : fix sync-ggml-am.sh
ggerganov Apr 10, 2025
bb7f9e5
ggml : add more generic custom op, remove deprecated custom ops (ggml…
slaren Apr 9, 2025
8ed7176
ggml : add bilinear upscale support (ggml/1185)
slaren Apr 9, 2025
d6e6a49
ggml: don't include arm_neon.h when using CUDA 12 with ARM Neon (ggml…
cmdr2 Apr 10, 2025
209c5b2
sync : ggml
ggerganov Apr 10, 2025
454beea
tests : fix init order (#0)
ggerganov Apr 10, 2025
4cacda8
sync : ggml
ggerganov Apr 10, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
129 changes: 38 additions & 91 deletions ggml/include/ggml.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -507,17 +507,12 @@ extern "C" {

GGML_OP_UNARY,

GGML_OP_MAP_UNARY,
GGML_OP_MAP_BINARY,

GGML_OP_MAP_CUSTOM1_F32,
GGML_OP_MAP_CUSTOM2_F32,
GGML_OP_MAP_CUSTOM3_F32,

GGML_OP_MAP_CUSTOM1,
GGML_OP_MAP_CUSTOM2,
GGML_OP_MAP_CUSTOM3,

GGML_OP_CUSTOM,

GGML_OP_CROSS_ENTROPY_LOSS,
GGML_OP_CROSS_ENTROPY_LOSS_BACK,
GGML_OP_OPT_STEP_ADAMW,
Expand Down Expand Up @@ -1722,24 +1717,29 @@ extern "C" {
float p0,
float p1);

// nearest interpolate
enum ggml_scale_mode {
GGML_SCALE_MODE_NEAREST = 0,
GGML_SCALE_MODE_BILINEAR = 1,
};

// interpolate
// multiplies ne0 and ne1 by scale factor
// used in stable-diffusion
GGML_API struct ggml_tensor * ggml_upscale(
struct ggml_context * ctx,
struct ggml_tensor * a,
int scale_factor);
int scale_factor,
enum ggml_scale_mode mode);

// nearest interpolate
// nearest interpolate to specified dimensions
// used in tortoise.cpp
// interpolate
// interpolate scale to specified dimensions
GGML_API struct ggml_tensor * ggml_upscale_ext(
struct ggml_context * ctx,
struct ggml_tensor * a,
int ne0,
int ne1,
int ne2,
int ne3);
int ne3,
enum ggml_scale_mode mode);

// pad each dimension with zeros: [x, ..., x] -> [x, ..., x, 0, ..., 0]
GGML_API struct ggml_tensor * ggml_pad(
Expand Down Expand Up @@ -1916,83 +1916,6 @@ extern "C" {

// custom operators

typedef void (*ggml_unary_op_f32_t) (const int, float *, const float *);
typedef void (*ggml_binary_op_f32_t)(const int, float *, const float *, const float *);

typedef void (*ggml_custom1_op_f32_t)(struct ggml_tensor *, const struct ggml_tensor *);
typedef void (*ggml_custom2_op_f32_t)(struct ggml_tensor *, const struct ggml_tensor *, const struct ggml_tensor *);
typedef void (*ggml_custom3_op_f32_t)(struct ggml_tensor *, const struct ggml_tensor *, const struct ggml_tensor *, const struct ggml_tensor *);

GGML_DEPRECATED(GGML_API struct ggml_tensor * ggml_map_unary_f32(
struct ggml_context * ctx,
struct ggml_tensor * a,
ggml_unary_op_f32_t fun),
"use ggml_map_custom1 instead");

GGML_DEPRECATED(GGML_API struct ggml_tensor * ggml_map_unary_inplace_f32(
struct ggml_context * ctx,
struct ggml_tensor * a,
ggml_unary_op_f32_t fun),
"use ggml_map_custom1_inplace instead");

GGML_DEPRECATED(GGML_API struct ggml_tensor * ggml_map_binary_f32(
struct ggml_context * ctx,
struct ggml_tensor * a,
struct ggml_tensor * b,
ggml_binary_op_f32_t fun),
"use ggml_map_custom2 instead");

GGML_DEPRECATED(GGML_API struct ggml_tensor * ggml_map_binary_inplace_f32(
struct ggml_context * ctx,
struct ggml_tensor * a,
struct ggml_tensor * b,
ggml_binary_op_f32_t fun),
"use ggml_map_custom2_inplace instead");

GGML_DEPRECATED(GGML_API struct ggml_tensor * ggml_map_custom1_f32(
struct ggml_context * ctx,
struct ggml_tensor * a,
ggml_custom1_op_f32_t fun),
"use ggml_map_custom1 instead");

GGML_DEPRECATED(GGML_API struct ggml_tensor * ggml_map_custom1_inplace_f32(
struct ggml_context * ctx,
struct ggml_tensor * a,
ggml_custom1_op_f32_t fun),
"use ggml_map_custom1_inplace instead");

GGML_DEPRECATED(GGML_API struct ggml_tensor * ggml_map_custom2_f32(
struct ggml_context * ctx,
struct ggml_tensor * a,
struct ggml_tensor * b,
ggml_custom2_op_f32_t fun),
"use ggml_map_custom2 instead");

GGML_DEPRECATED(GGML_API struct ggml_tensor * ggml_map_custom2_inplace_f32(
struct ggml_context * ctx,
struct ggml_tensor * a,
struct ggml_tensor * b,
ggml_custom2_op_f32_t fun),
"use ggml_map_custom2_inplace instead");

GGML_DEPRECATED(GGML_API struct ggml_tensor * ggml_map_custom3_f32(
struct ggml_context * ctx,
struct ggml_tensor * a,
struct ggml_tensor * b,
struct ggml_tensor * c,
ggml_custom3_op_f32_t fun),
"use ggml_map_custom3 instead");

GGML_DEPRECATED(GGML_API struct ggml_tensor * ggml_map_custom3_inplace_f32(
struct ggml_context * ctx,
struct ggml_tensor * a,
struct ggml_tensor * b,
struct ggml_tensor * c,
ggml_custom3_op_f32_t fun),
"use ggml_map_custom3_inplace instead");

// custom operators v2

typedef void (*ggml_custom1_op_t)(struct ggml_tensor * dst , const struct ggml_tensor * a, int ith, int nth, void * userdata);
typedef void (*ggml_custom2_op_t)(struct ggml_tensor * dst , const struct ggml_tensor * a, const struct ggml_tensor * b, int ith, int nth, void * userdata);
typedef void (*ggml_custom3_op_t)(struct ggml_tensor * dst , const struct ggml_tensor * a, const struct ggml_tensor * b, const struct ggml_tensor * c, int ith, int nth, void * userdata);
Expand Down Expand Up @@ -2048,6 +1971,30 @@ extern "C" {
int n_tasks,
void * userdata);

typedef void (*ggml_custom_op_t)(struct ggml_tensor * dst , int ith, int nth, void * userdata);

GGML_API struct ggml_tensor * ggml_custom_4d(
struct ggml_context * ctx,
enum ggml_type type,
int64_t ne0,
int64_t ne1,
int64_t ne2,
int64_t ne3,
struct ggml_tensor ** args,
int n_args,
ggml_custom_op_t fun,
int n_tasks,
void * userdata);

GGML_API struct ggml_tensor * ggml_custom_inplace(
struct ggml_context * ctx,
struct ggml_tensor * a,
struct ggml_tensor ** args,
int n_args,
ggml_custom_op_t fun,
int n_tasks,
void * userdata);

// loss function

GGML_API struct ggml_tensor * ggml_cross_entropy_loss(
Expand Down
3 changes: 3 additions & 0 deletions ggml/src/ggml-cann/ggml-cann.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1824,6 +1824,9 @@ static bool ggml_backend_cann_supports_op(ggml_backend_dev_t dev,
if (op->src[0]->ne[2] * op->ne[3] != op->src[0]->ne[3] * op->ne[2]) {
return false;
}
if (op->op_params[0] != GGML_SCALE_MODE_NEAREST) {
return false;
}
return true;
}
case GGML_OP_POOL_2D: {
Expand Down
55 changes: 15 additions & 40 deletions ggml/src/ggml-cpu/ggml-cpu.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2027,41 +2027,6 @@ static void ggml_compute_forward(struct ggml_compute_params * params, struct ggm
{
ggml_compute_forward_rwkv_wkv7(params, tensor);
} break;
case GGML_OP_MAP_UNARY:
{
ggml_unary_op_f32_t fun;
memcpy(&fun, tensor->op_params, sizeof(fun));
ggml_compute_forward_map_unary(params, tensor, fun);
}
break;
case GGML_OP_MAP_BINARY:
{
ggml_binary_op_f32_t fun;
memcpy(&fun, tensor->op_params, sizeof(fun));
ggml_compute_forward_map_binary(params, tensor, fun);
}
break;
case GGML_OP_MAP_CUSTOM1_F32:
{
ggml_custom1_op_f32_t fun;
memcpy(&fun, tensor->op_params, sizeof(fun));
ggml_compute_forward_map_custom1_f32(params, tensor, fun);
}
break;
case GGML_OP_MAP_CUSTOM2_F32:
{
ggml_custom2_op_f32_t fun;
memcpy(&fun, tensor->op_params, sizeof(fun));
ggml_compute_forward_map_custom2_f32(params, tensor, fun);
}
break;
case GGML_OP_MAP_CUSTOM3_F32:
{
ggml_custom3_op_f32_t fun;
memcpy(&fun, tensor->op_params, sizeof(fun));
ggml_compute_forward_map_custom3_f32(params, tensor, fun);
}
break;
case GGML_OP_MAP_CUSTOM1:
{
ggml_compute_forward_map_custom1(params, tensor);
Expand All @@ -2077,6 +2042,11 @@ static void ggml_compute_forward(struct ggml_compute_params * params, struct ggm
ggml_compute_forward_map_custom3(params, tensor);
}
break;
case GGML_OP_CUSTOM:
{
ggml_compute_forward_custom(params, tensor);
}
break;
case GGML_OP_CROSS_ENTROPY_LOSS:
{
ggml_compute_forward_cross_entropy_loss(params, tensor);
Expand Down Expand Up @@ -2328,11 +2298,6 @@ static int ggml_get_n_tasks(struct ggml_tensor * node, int n_threads) {
case GGML_OP_WIN_PART:
case GGML_OP_WIN_UNPART:
case GGML_OP_GET_REL_POS:
case GGML_OP_MAP_UNARY:
case GGML_OP_MAP_BINARY:
case GGML_OP_MAP_CUSTOM1_F32:
case GGML_OP_MAP_CUSTOM2_F32:
case GGML_OP_MAP_CUSTOM3_F32:
{
n_tasks = 1;
} break;
Expand Down Expand Up @@ -2366,6 +2331,16 @@ static int ggml_get_n_tasks(struct ggml_tensor * node, int n_threads) {
n_tasks = MIN(p.n_tasks, n_threads);
}
} break;
case GGML_OP_CUSTOM:
{
struct ggml_custom_op_params p;
memcpy(&p, node->op_params, sizeof(p));
if (p.n_tasks == GGML_N_TASKS_MAX) {
n_tasks = n_threads;
} else {
n_tasks = MIN(p.n_tasks, n_threads);
}
} break;
case GGML_OP_CROSS_ENTROPY_LOSS:
case GGML_OP_CROSS_ENTROPY_LOSS_BACK:
case GGML_OP_OPT_STEP_ADAMW:
Expand Down
Loading
Loading