Daniel Morin 9bd3a3be74 gsttensor: adding new datatypes
- Adding datatype for string, boolean, complex numbers and special floating
  point numbers.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9172>
2025-06-10 09:04:48 +00:00

230 lines
6.5 KiB
C

/*
* GStreamer gstreamer-tensor
* Copyright (C) 2024 Collabora Ltd
*
* gsttensor.h
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Library General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Library General Public License for more details.
*
* You should have received a copy of the GNU Library General Public
* License along with this library; if not, write to the
* Free Software Foundation, Inc., 51 Franklin St, Fifth Floor,
* Boston, MA 02110-1301, USA.
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#ifndef __GST_TENSOR_H__
#define __GST_TENSOR_H__
#include <gst/gst.h>
#include <gst/analytics/analytics-meta-prelude.h>
/**
* GstTensorDataType:
* @GST_TENSOR_DATA_TYPE_INT4: signed 4 bit integer tensor data
* @GST_TENSOR_DATA_TYPE_INT8: signed 8 bit integer tensor data
* @GST_TENSOR_DATA_TYPE_INT16: signed 16 bit integer tensor data
* @GST_TENSOR_DATA_TYPE_INT32: signed 32 bit integer tensor data
* @GST_TENSOR_DATA_TYPE_INT64: signed 64 bit integer tensor data
* @GST_TENSOR_DATA_TYPE_UINT4: unsigned 4 bit integer tensor data
* @GST_TENSOR_DATA_TYPE_UINT8: unsigned 8 bit integer tensor data
* @GST_TENSOR_DATA_TYPE_UINT16: unsigned 16 bit integer tensor data
* @GST_TENSOR_DATA_TYPE_UINT32: unsigned 32 bit integer tensor data
* @GST_TENSOR_DATA_TYPE_UINT64: unsigned 64 bit integer tensor data
* @GST_TENSOR_DATA_TYPE_FLOAT16: 16 bit floating point tensor data
* @GST_TENSOR_DATA_TYPE_FLOAT32: 32 bit floating point tensor data
* @GST_TENSOR_DATA_TYPE_FLOAT64: 64 bit floating point tensor data
* @GST_TENSOR_DATA_TYPE_BFLOAT16: "brain" 16 bit floating point tensor data
*
* Describe the type of data contain in the tensor.
*
* Since: 1.26
*/
typedef enum _GstTensorDataType
{
GST_TENSOR_DATA_TYPE_INT4,
GST_TENSOR_DATA_TYPE_INT8,
GST_TENSOR_DATA_TYPE_INT16,
GST_TENSOR_DATA_TYPE_INT32,
GST_TENSOR_DATA_TYPE_INT64,
GST_TENSOR_DATA_TYPE_UINT4,
GST_TENSOR_DATA_TYPE_UINT8,
GST_TENSOR_DATA_TYPE_UINT16,
GST_TENSOR_DATA_TYPE_UINT32,
GST_TENSOR_DATA_TYPE_UINT64,
GST_TENSOR_DATA_TYPE_FLOAT16,
GST_TENSOR_DATA_TYPE_FLOAT32,
GST_TENSOR_DATA_TYPE_FLOAT64,
GST_TENSOR_DATA_TYPE_BFLOAT16,
/**
* GST_TENSOR_DATA_TYPE_STRING:
*
* UTF-8 string
*
* Since: 1.28
*/
GST_TENSOR_DATA_TYPE_STRING,
/**
* GST_TENSOR_DATA_TYPE_BOOL:
*
* A boolean value stored in 1 byte.
*
* Since: 1.28
*/
GST_TENSOR_DATA_TYPE_BOOL,
/**
* GST_TENSOR_DATA_TYPE_COMPLEX64:
*
* A 64-bit complex number stored in 2 32-bit values.
*
* Since: 1.28
*/
GST_TENSOR_DATA_TYPE_COMPLEX64,
/**
* GST_TENSOR_DATA_TYPE_COMPLEX128:
*
* A 128-bit complex number stored in 2 64-bit values.
*
* Since: 1.28
*/
GST_TENSOR_DATA_TYPE_COMPLEX128,
/**
* GST_TENSOR_DATA_TYPE_FLOAT8E4M3FN:
*
* A non-IEEE 8-bit floating point format with 4 exponent bits and 3 mantissa bits, with NaN and no infinite values (FN).
* See [this paper for more details](https://onnx.ai/onnx/technical/float8.html)
*
* Since: 1.28
*/
GST_TENSOR_DATA_TYPE_FLOAT8E4M3FN,
/**
* GST_TENSOR_DATA_TYPE_FLOAT8E4M3FNUZ:
*
* A non-IEEE 8-bit floating point format with 4 exponent bits and 3 mantissa bits, with NaN, no infinite values (FN) and no negative zero (UZ).
* See [this paper for more details](https://onnx.ai/onnx/technical/float8.html)
*
* Since: 1.28
*/
GST_TENSOR_DATA_TYPE_FLOAT8E4M3FNUZ,
/**
* GST_TENSOR_DATA_TYPE_FLOAT8E5M2:
*
* A non-IEEE 8-bit floating point format with 5 exponent bits and 2 mantissa bits.
* See [this paper for more details](https://onnx.ai/onnx/technical/float8.html)
*
* Since: 1.28
*/
GST_TENSOR_DATA_TYPE_FLOAT8E5M2,
/**
* GST_TENSOR_DATA_TYPE_FLOAT8E5M2FNUZ:
*
* A non-IEEE 8-bit floating point format with 5 exponent bits and 2 mantissa bits, with NaN, no infinite values (FN) and no negative zero (UZ).
* See [this paper for more details](https://onnx.ai/onnx/technical/float8.html)
*
* Since: 1.28
*/
GST_TENSOR_DATA_TYPE_FLOAT8E5M2FNUZ
} GstTensorDataType;
/**
* GstTensorDimOrder:
* @GST_TENSOR_DIM_ORDER_ROW_MAJOR: elements along a row are consecutive in memory
* @GST_TENSOR_DIM_ORDER_COL_MAJOR: elements along a column are consecutive in memory
*
* Indicate to read tensor from memory in row-major or column-major order.
*
* Since: 1.26
*/
typedef enum _GstTensorDimOrder
{
GST_TENSOR_DIM_ORDER_ROW_MAJOR,
GST_TENSOR_DIM_ORDER_COL_MAJOR
} GstTensorDimOrder;
/**
* GstTensorLayout:
* @GST_TENSOR_LAYOUT_CONTIGUOUS: indicate the tensor is stored in a dense format in memory
*
* Indicate tensor storage in memory.
*
* Since: 1.26
*/
typedef enum _GstTensorLayout
{
GST_TENSOR_LAYOUT_CONTIGUOUS
} GstTensorLayout;
/**
* GstTensor:
* @id: semantically identify the contents of the tensor
* @layout: Indicate tensor layout
* @data_type: #GstTensorDataType of tensor data
* @data: #GstBuffer holding tensor data
* @dims_order: Indicate tensor elements layout in memory.
* @num_dims: number of tensor dimensions
* @dims: (array length=num_dims): number of tensor dimensions
*
* Hold tensor data
*
* Since: 1.26
*/
typedef struct _GstTensor
{
GQuark id;
GstTensorLayout layout;
GstTensorDataType data_type;
GstBuffer *data;
GstTensorDimOrder dims_order;
gsize num_dims;
gsize dims[];
} GstTensor;
G_BEGIN_DECLS
#define GST_TYPE_TENSOR (gst_tensor_get_type())
GST_ANALYTICS_META_API
GstTensor * gst_tensor_alloc (gsize num_dims);
GST_ANALYTICS_META_API
GstTensor * gst_tensor_new_simple (GQuark id,
GstTensorDataType data_type,
GstBuffer * data,
GstTensorDimOrder dims_order,
gsize num_dims,
gsize * dims);
GST_ANALYTICS_META_API
void gst_tensor_free (GstTensor * tensor);
GST_ANALYTICS_META_API
GstTensor * gst_tensor_copy (const GstTensor * tensor);
GST_ANALYTICS_META_API
gsize * gst_tensor_get_dims (GstTensor * tensor, gsize * num_dims);
GST_ANALYTICS_META_API
GType gst_tensor_get_type (void);
GST_ANALYTICS_META_API
const gchar *gst_tensor_data_type_get_name (GstTensorDataType data_type);
GST_ANALYTICS_META_API
gboolean gst_tensor_check_type(const GstTensor * tensor, GstTensorDimOrder order,
gsize num_dims, GstTensorDataType data_type, GstBuffer * data);
G_END_DECLS
#endif /* __GST_TENSOR_H__ */