CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-08 10:07:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

ggml-zdnn: support op out_prod

Browse Source 
Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>
This commit is contained in:
Aaron Teo
2025-07-23 19:28:51 +08:00
parent 11d58d29de
commit 77a753297b
1 changed files with 92 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

105
ggml/src/ggml-zdnn/ggml-zdnn.cpp
View File

@@ -67,6 +67,71 @@ inline void ggml_zdnn_load_tensor(zdnn_ztensor & ztensor,
ZDNN_CHECK(zdnn_transform_ztensor(&ztensor, buffer));
}
static void ggml_backend_zdnn_out_prod(ggml_backend_zdnn_context * ctx, const ggml_tensor * src0, const ggml_tensor * src1, ggml_tensor * dst) {
const ggml_tensor * a = src1;
const ggml_tensor * b = src0;
ggml_tensor * c = dst;
GGML_TENSOR_BINARY_OP_LOCALS
GGML_ASSERT(ne0 == ne00);
GGML_ASSERT(ne1 == ne10);
GGML_ASSERT(ne2 == ne02);
GGML_ASSERT(ne02 == ne12);
GGML_ASSERT(ne3 == ne13);
GGML_ASSERT(ne03 == ne13);
// we don't support permuted src0 or src1
GGML_ASSERT(nb00 == sizeof(float));
// dst cannot be transposed or permuted
GGML_ASSERT(nb0 == sizeof(float));
const int64_t k = ne01;
const int64_t n = ne00;
const int64_t m = ne10;
bool transposeA;
if (!ggml_is_transposed(src1)) {
transposeA = true;
} else {
transposeA = false;
}
zdnn_tensor_desc pre_tfm_desc_a, tfm_desc_a;
zdnn_tensor_desc pre_tfm_desc_b, tfm_desc_b;
zdnn_tensor_desc pre_tfm_desc_bias, tfm_desc_bias;
zdnn_tensor_desc pre_tfm_desc_c, tfm_desc_c;
zdnn_ztensor ztensor_a, ztensor_b, ztensor_bias, ztensor_c;
const int64_t a_dim[GGML_MAX_DIMS] = { 1, 1, m, n };
const int64_t b_dim[GGML_MAX_DIMS] = { 1, 1, n, k };
const int64_t bias_dim[GGML_MAX_DIMS] = { 1, 1, 1, k };
const int64_t c_dim[GGML_MAX_DIMS] = { 1, 1, m, k };
ggml_zdnn_create_tensor(pre_tfm_desc_a, tfm_desc_a, ztensor_a, a, a_dim, ZDNN_2D);
ggml_zdnn_create_tensor(pre_tfm_desc_b, tfm_desc_b, ztensor_b, b, b_dim, ZDNN_2D);
ggml_zdnn_create_tensor(pre_tfm_desc_bias, tfm_desc_bias, ztensor_bias, dst, bias_dim, ZDNN_1D);
ggml_zdnn_create_tensor(pre_tfm_desc_c, tfm_desc_c, ztensor_c, c, c_dim, ZDNN_2D);
void * bias_data = (void *)calloc(k, ggml_element_size(c));
ZDNN_CHECK(zdnn_transform_ztensor(&ztensor_a, a->data));
ZDNN_CHECK(zdnn_transform_ztensor(&ztensor_b, b->data));
ZDNN_CHECK(zdnn_transform_ztensor(&ztensor_bias, bias_data));
ZDNN_CHECK(zdnn_transform_ztensor(&ztensor_c, c->data));
ZDNN_CHECK(zdnn_matmul_transpose_op(&ztensor_a, &ztensor_b, &ztensor_bias,
transposeA, false, MATMUL_OP_ADDITION, &ztensor_c));
ZDNN_CHECK(zdnn_transform_origtensor(&ztensor_c, c->data));
ZDNN_CHECK(zdnn_free_ztensor_buffer(&ztensor_a));
ZDNN_CHECK(zdnn_free_ztensor_buffer(&ztensor_b));
ZDNN_CHECK(zdnn_free_ztensor_buffer(&ztensor_bias));
ZDNN_CHECK(zdnn_free_ztensor_buffer(&ztensor_c));
free(bias_data);
}
static void ggml_backend_zdnn_mul_mat(ggml_backend_zdnn_context * ctx, const ggml_tensor * src0, const ggml_tensor * src1, ggml_tensor * dst) {
GGML_TENSOR_BINARY_OP_LOCALS
@@ -187,6 +252,10 @@ static bool ggml_backend_zdnn_compute_forward(ggml_backend_zdnn_context * ctx, g
ggml_backend_zdnn_mul_mat_dispatch(ctx, dst->src[0], dst->src[1], dst);
break;
case GGML_OP_OUT_PROD:
ggml_backend_zdnn_out_prod(ctx, dst->src[0], dst->src[1], dst);
break;
default:
return false;
}
@@ -518,23 +587,33 @@ static bool ggml_backend_zdnn_device_supports_op(ggml_backend_dev_t dev, const g
return true;
case GGML_OP_MUL_MAT:
{
const ggml_tensor * src0 = op->src[0];
const ggml_tensor * src1 = op->src[1];
{
const ggml_tensor * src0 = op->src[0];
const ggml_tensor * src1 = op->src[1];
const int64_t ne10 = src1->ne[0];
const int64_t ne10 = src1->ne[0];
const int64_t ne0 = op->ne[0];
const int64_t ne1 = op->ne[1];
const int64_t ne0 = op->ne[0];
const int64_t ne1 = op->ne[1];
const int64_t max_batch = zdnn_get_nnpa_max_dim_idx_size();
const int64_t max_batch = zdnn_get_nnpa_max_dim_idx_size();
return ggml_is_contiguous(src0) &&
ggml_is_contiguous(src1) &&
src1->type == GGML_TYPE_F32 &&
(ne0 <= max_batch && ne1 <= max_batch && ne10 <= max_batch) &&
(src0->type == GGML_TYPE_F32 || ggml_get_type_traits(src0->type)->to_float != NULL);
}
return ggml_is_contiguous(src0) &&
ggml_is_contiguous(src1) &&
src1->type == GGML_TYPE_F32 &&
(ne0 <= max_batch && ne1 <= max_batch && ne10 <= max_batch) &&
(src0->type == GGML_TYPE_F32 || ggml_get_type_traits(src0->type)->to_float != NULL);
} break;
case GGML_OP_OUT_PROD:
{
return op->src[0]->type == GGML_TYPE_F32 &&
op->src[1]->type == GGML_TYPE_F32 &&
ggml_is_matrix(src0) &&
ggml_is_matrix(src1) &&
ggml_is_contiguous(src0) &&
(ggml_is_contiguous(src1) || ggml_is_transposed(src1)) &&
(src0->type == GGML_TYPE_F32 || ggml_get_type_traits(src0->type)->to_float != NULL);
} break;
default:
return false;