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: run compute and store into tensor->extra

Browse Source 
Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>
This commit is contained in:
Aaron Teo
2025-07-21 20:30:54 +08:00
parent 02cfcfb270
commit 36d76c30fb
1 changed files with 32 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

50
ggml/src/ggml-zdnn/ggml-zdnn.cpp
View File

@@ -172,29 +172,42 @@ static void ggml_backend_zdnn_mul_mat(ggml_backend_zdnn_context * ctx, const ggm
ZDNN_CHECK(zdnn_transform_ztensor(&ztensor_inputs, inputs->data)); ZDNN_CHECK(zdnn_transform_ztensor(&ztensor_inputs, inputs->data));
ZDNN_CHECK(zdnn_transform_ztensor(&ztensor_bias, bias_data)); ZDNN_CHECK(zdnn_transform_ztensor(&ztensor_bias, bias_data));
zdnn_extra * inputs_extra = (zdnn_extra *)inputs->extra;
zdnn_extra * weights_extra = (zdnn_extra *)weights->extra;
zdnn_extra * output_extra = (zdnn_extra *)output->extra;
ZDNN_CHECK(zdnn_matmul_transpose_op(&inputs_extra->ztensor, &weights_extra->ztensor, &output_extra->extra->ztensor,
false, true, MATMUL_OP_ADDITION, &output_extra->ztensor));
ZDNN_CHECK(zdnn_matmul_transpose_op(&ztensor_inputs, &ztensor_weights, &ztensor_bias, ZDNN_CHECK(zdnn_matmul_transpose_op(&ztensor_inputs, &ztensor_weights, &ztensor_bias,
false, true, MATMUL_OP_ADDITION, &ztensor_output)); false, true, MATMUL_OP_ADDITION, &ztensor_output));
void * check_output_buffer = malloc(ggml_nbytes(output));
ZDNN_CHECK(zdnn_transform_origtensor(&output_extra->ztensor, check_output_buffer));
ZDNN_CHECK(zdnn_transform_origtensor(&ztensor_output, output->data)); ZDNN_CHECK(zdnn_transform_origtensor(&ztensor_output, output->data));
// Compare the first 10 elements of the two buffers // Compare the first 10 elements of the two buffers
GGML_LOG_INFO("%s: Comparing output buffers:\n", __func__); GGML_LOG_INFO("%s: Comparing output buffers:\n", __func__);
GGML_LOG_INFO("Index | output->data\n"); GGML_LOG_INFO("Index | output->data | check_output_buffer\n");
GGML_LOG_INFO("------|-------------\n"); GGML_LOG_INFO("------|--------------|--------------------\n");
for (int i = 0; i < 10 && i < output->ne[0] * output->ne[1]; i++) { for (int i = 0; i < 10 && i < output->ne[0] * output->ne[1]; i++) {
GGML_LOG_INFO("%5d | %12.6f\n", GGML_LOG_INFO("%5d | %12.6f | %18.6f\n",
i, i,
((float *)output->data)[i]); ((float *)output->data)[i],
} ((float *)check_output_buffer)[i]);
}
GGML_LOG_INFO("... (snip) ...\n");
GGML_LOG_INFO("Index | output->data | check_output_buffer\n");
GGML_LOG_INFO("------|--------------|--------------------\n");
const int64_t num_elements = output->ne[0] * output->ne[1];
for (int64_t i = (num_elements > 10 ? num_elements - 10 : 0); i < num_elements; i++) {
GGML_LOG_INFO("%5lld | %12.6f | %18.6f\n",
(long long) i,
((float *)output->data)[i],
((float *)check_output_buffer)[i]);
}
GGML_LOG_INFO("... (snip) ...\n");
GGML_LOG_INFO("Index | output->data\n");
GGML_LOG_INFO("------|-------------\n");
const int64_t num_elements = output->ne[0] * output->ne[1];
for (int64_t i = (num_elements > 10 ? num_elements - 10 : 0); i < num_elements; i++) {
GGML_LOG_INFO("%5lld | %12.6f\n",
(long long) i,
((float *)output->data)[i]);
}
std::raise(SIGINT); std::raise(SIGINT);
@@ -204,6 +217,7 @@ static void ggml_backend_zdnn_mul_mat(ggml_backend_zdnn_context * ctx, const ggm
ZDNN_CHECK(zdnn_free_ztensor_buffer(&ztensor_output)); ZDNN_CHECK(zdnn_free_ztensor_buffer(&ztensor_output));
free(bias_data); free(bias_data);
free(check_output_buffer);
} }
static void ggml_backend_zdnn_mul_mat_dispatch(ggml_backend_zdnn_context * ctx, const ggml_tensor * src0, const ggml_tensor * src1, ggml_tensor * dst) { static void ggml_backend_zdnn_mul_mat_dispatch(ggml_backend_zdnn_context * ctx, const ggml_tensor * src0, const ggml_tensor * src1, ggml_tensor * dst) {