Pixel-Processing Accelerator (PPA)

Introduction

ESP32-P4 includes a pixel-processing accelerator (PPA) module, to realize hardware-level acceleration of image algorithms, such as image rotation, scaling, mirroring, and blending.

Terminology

The terms used in relation to the PPA driver are given in the table and the diagram below.

Term	Definition
Picture (pic)	A complete image stored in the system memory.
Block	A portion cropped from a picture at a certain size, with the maximum size equivalent to the entire picture.
Pixel	The unit to be used in the PPA context.
PPA Operation	Types of image algorithm accelerations, includes scale-rotate-mirror (SRM), blend, and fill.
PPA Client	Who wants to do the PPA operations. Typically, every PPA client is hold by a specific task.
PPA Transaction	One request from a PPA client to do a PPA operation is one PPA transaction.

Functional Overview

The following sections detail the design of the PPA driver:

Register PPA Client - Covers how to register a PPA client to perform any PPA operations.
Register PPA Event Callbacks - Covers how to hook user specific code to PPA driver event callback function.
Perform PPA Operations - Covers how to perform a PPA operation.
Thread Safety - Covers the usage of the PPA operation APIs in thread safety aspect.
Performance Overview - Covers the performance of PPA operations.

Register PPA Client

Requests to perform PPA operations are made by PPA clients. Therefore, PPA clients need to be registered first before doing any PPA operations. Call ppa_register_client() function to register a new client. ppa_client_config_t structure is used to specific the properties of the client.

ppa_client_config_t::oper_type - Each PPA operation type corresponds to one PPA client type, a registered PPA client can only request one specific type of PPA operations.
ppa_client_config_t::max_pending_trans_num - Decides the maximum number of pending PPA transactions the client can hold.

It is recommended that every task to register its own PPA clients. For example, an application contains two tasks: Task A requires both the PPA SRM and the PPA fill functionalities, so one PPA SRM client and one PPA fill client should be registered in Task A; While Task B also requires the PPA SRM functionality, then another PPA SRM client should be registered in Task B.

If the task no longer needs to do PPA operations, the corresponding PPA clients can be deregistered with ppa_unregister_client() function.

Register PPA Event Callbacks

When an event occurs (e.g., a PPA transaction is completed), the CPU is notified of this event via an interrupt. If some specific functions need to be called when a particular event occurs, a callback can be registered for that event by calling ppa_client_register_event_callbacks(). This can be specifically useful when PPA_TRANS_MODE_NON_BLOCKING mode is selected to perform the PPA operations. It is worth noticing that the event callbacks are bound to PPA clients, but the user context is provided per transaction in the call to the PPA operation APIs. This allows the maximum flexibility in utilizing the event callbacks.

The registered callback functions are called in the interrupt context, therefore, the callback functions should follow common ISR (Interrupt Service Routine) rules.

Perform PPA Operations

Once the PPA client is registered, a PPA operation can be requested with the returned ppa_client_handle_t.

PPA operations includes:

Scale, Rotate, Mirror (SRM)

Call ppa_do_scale_rotate_mirror() to apply one or more of the scaling, rotation, mirroring operations to the target block inside a picture.

Some notes to avoid confusion in configuring ppa_srm_oper_config_t:

ppa_in_pic_blk_config_t::buffer and ppa_out_pic_blk_config_t::buffer have to be the pointers to different picture buffers for a SRM operation.
The precision of ppa_srm_oper_config_t::scale_x and ppa_srm_oper_config_t::scale_y will be truncated to a step size of 1/16.
Output block's width/height is totally determined by the input block's width/height, scaling factor, and rotation angle, so output block's width/height does not need to be configured. However, please make sure the output block can fit at the offset location in the output picture.
If the color mode of the input or output picture is PPA_SRM_COLOR_MODE_YUV420, then its pic_w, pic_h, block_w, block_h, block_offset_x, block_offset_y fields must be even.

Blend

Call ppa_do_blend() to blend the two target blocks of two so-called foreground (FG) and background (BG) pictures.

Blend follows the normal Alpha Blending formula:

\(A_{out} = A_b + A_f - A_b \times A_f\)

\(C_{out} = (C_b \times A_b \times (1 - A_f) + C_f \times A_f) / (A_b + A_f - A_b \times A_f)\)

where \(A_b\) is the Alpha channel of the background layer, \(A_f\) is the Alpha channel of the foreground layer, \(C_b\) corresponds to the R, G, B components of the background layer, and \(C_f\) corresponds to the R, G, B components of the foreground layer.

Note that this formula is not symmetric to FG and BG. When \(A_f = 1\), it calculates \(C_{out} = C_f\), \(A_{out} = 1\), which means if the color mode of the FG picture is PPA_BLEND_COLOR_MODE_RGB565 or PPA_BLEND_COLOR_MODE_RGB888, since a Alpha value of 255 will be filled by the PPA hardware (i.e. \(A_f = 1\)), the blended result will be identical to the FG block.

If ppa_blend_oper_config_t::bg_ck_en or ppa_blend_oper_config_t::fg_ck_en is set to true, the pixels fall into the color-key (also known as Chroma-key) range does not follow Alpha Blending process. Please check ESP32-P4 Technical Reference Manual > Pixel-Processing Accelerator (PPA) > Functional Description > Layer Blending (BLEND) [PDF] for the detailed rules.

Similarly, some notes to avoid confusion in configuring ppa_blend_oper_config_t:

ppa_out_pic_blk_config_t::buffer can be the same pointer to one of the input's ppa_in_pic_blk_config_t::buffer for a blend operation.
The blocks' width/height of FG and BG should be identical, and are the width/height values for the output block.
If the color mode of the input picture is PPA_BLEND_COLOR_MODE_A4, then its block_w and block_offset_x fields must be even.

Fill

Call ppa_do_fill() to fill a target block inside a picture.

ppa_trans_mode_t is a field configurable to all the PPA operation APIs. It decides whether you want the call to the PPA operation API to block until the transaction finishes or to return immediately after the transaction is pushed to the internal queue.

Thread Safety

The PPA driver has guaranteed the thread safety of calling the PPA operation APIs in all following situations:

Among clients of different types in one task
Among clients of same type in different tasks
Among clients of different types in different tasks

Performance Overview

The PPA operations are acted on the target block of an input picture. Therefore, the time it takes to complete a PPA transaction is proportional to the amount of the data in the block. The size of the entire picture has no influence on the performance. More importantly, the PPA performance highly relies on the PSRAM bandwidth if the pictures are located in the PSRAM section. When there are quite a few peripherals reading and writing to the PSRAM at the same time, the performance of PPA operation will be greatly reduced.

API Reference

Header File

components/esp_driver_ppa/include/driver/ppa.h
This header file can be included with:
#include "driver/ppa.h"
This header file is a part of the API provided by the esp_driver_ppa component. To declare that your component depends on esp_driver_ppa, add the following to your CMakeLists.txt:
REQUIRES esp_driver_ppa
or
PRIV_REQUIRES esp_driver_ppa

Functions

esp_err_t ppa_register_client(const ppa_client_config_t *config, ppa_client_handle_t *ret_client)

Register a PPA client to do a specific PPA operation.

Parameters

config -- [in] Pointer to a collection of configurations for the client
ret_client -- [out] Returned client handle

Returns

ESP_OK: Register the PPA client successfully
ESP_ERR_INVALID_ARG: Register the PPA client failed because of invalid argument
ESP_ERR_NO_MEM: Register the PPA client failed because out of memory
ESP_FAIL: Register the PPA client failed because of other error

esp_err_t ppa_unregister_client(ppa_client_handle_t ppa_client)

Unregister a PPA client.

Note

This will also free the resources occupied by the client

Parameters

ppa_client -- [in] PPA client handle, allocated by ppa_register_client

Returns

ESP_OK: Unregister the PPA client successfully
ESP_ERR_INVALID_ARG: Unregister the PPA client failed because of invalid argument
ESP_ERR_INVALID_STATE: Unregister the PPA client failed because there are unfinished transactions

esp_err_t ppa_client_register_event_callbacks(ppa_client_handle_t ppa_client, const ppa_event_callbacks_t *cbs)

Register event callbacks for a PPA client.

Note

Any user private data that wants to be passed directly to callback's user_data is provided per PPA transaction. Please check the user_data field in ppa_xxx_oper_config_t structure.

Parameters

ppa_client -- [in] PPA client handle
cbs -- [in] Structure with all PPA callbacks

Returns

ESP_OK: Register event callbacks for the PPA client successfully
ESP_ERR_INVALID_ARG: Register event callbacks for the PPA client failed because of invalid argument

esp_err_t ppa_do_scale_rotate_mirror(ppa_client_handle_t ppa_client, const ppa_srm_oper_config_t *config)

Perform a scaling-rotating-mirroring (SRM) operation to a picture.

Parameters

ppa_client -- [in] PPA client handle that has been registered to do SRM operations
config -- [in] Pointer to a collection of configurations for the SRM operation transaction, ppa_srm_oper_config_t

Returns

ESP_OK: Perform a SRM operation successfully
ESP_ERR_INVALID_ARG: Perform a SRM operation failed because of invalid argument
ESP_FAIL: Perform a SRM operation failed because the client's pending transactions has reached its maximum capacity

esp_err_t ppa_do_blend(ppa_client_handle_t ppa_client, const ppa_blend_oper_config_t *config)

Perform a blending operation to a picture.

Parameters

ppa_client -- [in] PPA client handle that has been registered to do blend operations
config -- [in] Pointer to a collection of configurations for the blend operation transaction, ppa_blend_oper_config_t

Returns

ESP_OK: Perform a blend operation successfully
ESP_ERR_INVALID_ARG: Perform a blend operation failed because of invalid argument
ESP_FAIL: Perform a blend operation failed because the client's pending transactions has reached its maximum capacity

esp_err_t ppa_do_fill(ppa_client_handle_t ppa_client, const ppa_fill_oper_config_t *config)

Perform a filling operation to a picture.

Parameters

ppa_client -- [in] PPA client handle that has been registered to do fill operations
config -- [in] Pointer to a collection of configurations for the fill operation transaction, ppa_fill_oper_config_t

Returns

ESP_OK: Perform a fill operation successfully
ESP_ERR_INVALID_ARG: Perform a fill operation failed because of invalid argument
ESP_FAIL: Perform a fill operation failed because the client's pending transactions has reached its maximum capacity

Structures

struct ppa_client_config_t

A collection of configuration items that used for registering a PPA client.

Public Members

ppa_operation_t oper_type: The desired PPA operation for the client

uint32_t max_pending_trans_num: The maximum number of pending transactions for the client. By default, it will be 1, which is sufficient if all transactions are performed with PPA_TRANS_MODE_BLOCKING

ppa_data_burst_length_t data_burst_length: The desired data burst length for all the transactions of the client. Use a small burst length will decrease PPA performance, but can save burst bandwidth for other peripheral usages. By default, it will be at the maximum burst length, PPA_DATA_BURST_LENGTH_128

struct ppa_event_data_t: Type of PPA event data.

struct ppa_event_callbacks_t

Group of supported PPA callbacks.

Public Members

ppa_event_callback_t on_trans_done: Invoked when a PPA transaction finishes

struct ppa_in_pic_blk_config_t

A collection of configuration items for an input picture and the target block inside the picture.

Public Members

const void *buffer: Pointer to the input picture buffer

uint32_t pic_w: Input picture width (unit: pixel)

uint32_t pic_h: Input picture height (unit: pixel)

uint32_t block_w: Target block width (unit: pixel)

uint32_t block_h: Target block height (unit: pixel)

uint32_t block_offset_x: Target block offset in x direction in the picture (unit: pixel)

uint32_t block_offset_y: Target block offset in y direction in the picture (unit: pixel)

ppa_srm_color_mode_t srm_cm: Color mode of the picture in a PPA SRM operation. Supported color mode in ppa_srm_color_mode_t

ppa_blend_color_mode_t blend_cm: Color mode of the picture in a PPA blend operation. Supported color mode in ppa_blend_color_mode_t

ppa_fill_color_mode_t fill_cm: Color mode of the picture in a PPA fill operation. Supported color mode in ppa_fill_color_mode_t

ppa_color_range_t yuv_range: When the color mode is any YUV color space, this field is to describe its color range

ppa_color_conv_std_rgb_yuv_t yuv_std: When the color mode is any YUV color space, this field is to describe its YUV<->RGB conversion standard

struct ppa_out_pic_blk_config_t

A collection of configuration items for an output picture and the target block inside the picture.

Public Members

void *buffer: Pointer to the output picture buffer (requires alignment: internal memory needs align to L1 cache line size, external memory needs align to L1 and L2 cache line size)

uint32_t buffer_size: Size of the output picture buffer (requires alignment: internal memory needs align to L1 cache line size, external memory needs align to L1 and L2 cache line size)

uint32_t pic_w: Output picture width (unit: pixel)

uint32_t pic_h: Output picture height (unit: pixel)

uint32_t block_offset_x: Target block offset in x direction in the picture (unit: pixel)

uint32_t block_offset_y: Target block offset in y direction in the picture (unit: pixel)

ppa_srm_color_mode_t srm_cm: Color mode of the picture in a PPA SRM operation. Supported color mode in ppa_srm_color_mode_t

ppa_blend_color_mode_t blend_cm: Color mode of the picture in a PPA blend operation. Supported color mode in ppa_blend_color_mode_t

ppa_fill_color_mode_t fill_cm: Color mode of the picture in a PPA fill operation. Supported color mode in ppa_fill_color_mode_t

ppa_color_range_t yuv_range: When the color mode is any YUV color space, this field is to describe its color range

ppa_color_conv_std_rgb_yuv_t yuv_std: When the color mode is any YUV color space, this field is to describe its YUV<->RGB conversion standard

struct ppa_srm_oper_config_t

A collection of configuration items to do a PPA SRM operation transaction.

Public Members

ppa_in_pic_blk_config_t in: Information of the input picture and the target block

ppa_out_pic_blk_config_t out: Information of the output picture and the target block

ppa_srm_rotation_angle_t rotation_angle: Rotation (counter-clockwise) to the target block, select from ppa_srm_rotation_angle_t

float scale_x: Scaling factor to the target block in the x direction

float scale_y: Scaling factor to the target block in the y direction

bool mirror_x: Whether to mirror the target block in the x direction

bool mirror_y: Whether to mirror the target block in the y direction

bool rgb_swap: Whether to swap the input data in RGB (e.g. ARGB becomes BGRA, RGB becomes BGR)

bool byte_swap: Whether to swap the input data in byte. Only available feature if input picture color mode is ARGB8888 or RGB565

ppa_alpha_update_mode_t alpha_update_mode: Select whether the alpha channel of the input picture needs update

uint32_t alpha_fix_val: Range: [0, 255] When PPA_ALPHA_FIX_VALUE mode is selected, alpha_fix_val is the new alpha value to replace the input alpha value (output_alpha = alpha_fix_val)

float alpha_scale_ratio: Range: (0, 1) When PPA_ALPHA_SCALE mode is selected, alpha_scale_ratio is the multiplier to the input alpha value (output_alpha = alpha_scale_ratio * input_alpha) Ratio resolution is 1/256

ppa_trans_mode_t mode: Determines whether to block inside the operation functions, see ppa_trans_mode_t

void *user_data: User registered data to be passed into done_cb callback function

struct ppa_blend_oper_config_t

A collection of configuration items to do a PPA blend operation transaction.

Public Members

ppa_in_pic_blk_config_t in_bg: Information of the input background picture and the target block

ppa_in_pic_blk_config_t in_fg: Information of the input foreground picture and the target block

ppa_out_pic_blk_config_t out: Information of the output picture and the target block

bool bg_rgb_swap: Whether to swap the background input data in RGB (e.g. ARGB becomes BGRA, RGB becomes BGR)

bool bg_byte_swap: Whether to swap the background input data in byte. Only available feature if input BG picture color mode is ARGB8888 or RGB565

ppa_alpha_update_mode_t bg_alpha_update_mode: Select whether the alpha channel of the input background picture needs update

uint32_t bg_alpha_fix_val: Range: [0, 255] When PPA_ALPHA_FIX_VALUE mode is selected, alpha_fix_val is the new alpha value to replace the input alpha value (output_alpha = alpha_fix_val)

float bg_alpha_scale_ratio: Range: (0, 1) When PPA_ALPHA_SCALE mode is selected, alpha_scale_ratio is the multiplier to the input alpha value (output_alpha = alpha_scale_ratio * input_alpha) Ratio resolution is 1/256

bool fg_rgb_swap: Whether to swap the foreground input data in RGB (e.g. ARGB becomes BGRA, RGB becomes BGR)

bool fg_byte_swap: Whether to swap the foreground input data in byte. Only available feature if input FG picture color mode is ARGB8888 or RGB565

ppa_alpha_update_mode_t fg_alpha_update_mode: Select whether the alpha channel of the input foreground picture needs update

uint32_t fg_alpha_fix_val: Range: [0, 255] When PPA_ALPHA_FIX_VALUE mode is selected, alpha_fix_val is the new alpha value to replace the input alpha value (output_alpha = alpha_fix_val)

float fg_alpha_scale_ratio: Range: (0, 1) When PPA_ALPHA_SCALE mode is selected, alpha_scale_ratio is the multiplier to the input alpha value (output_alpha = alpha_scale_ratio * input_alpha) Ratio resolution is 1/256

color_pixel_rgb888_data_t fg_fix_rgb_val: When in_fg.blend_cm is PPA_BLEND_COLOR_MODE_A8/4, this field can be used to set a fixed color for the foreground, in RGB888 format

bool bg_ck_en: Whether to enable color keying for background If not enabled, all background pixels are considered as out of the color-keying range

color_pixel_rgb888_data_t bg_ck_rgb_low_thres: The lower threshold of the color-keying range for the background, in RGB888 format

color_pixel_rgb888_data_t bg_ck_rgb_high_thres: The higher threshold of the color-keying range for the background, in RGB888 format

bool fg_ck_en: Whether to enable color keying for foreground If not enabled, all foreground pixels are considered as out of the color-keying range

color_pixel_rgb888_data_t fg_ck_rgb_low_thres: The lower threshold of the color-keying range for the foreground, in RGB888 format

color_pixel_rgb888_data_t fg_ck_rgb_high_thres: The higher threshold of the color-keying range for the foreground, in RGB888 format

color_pixel_rgb888_data_t ck_rgb_default_val: The color to overwrite when a pixel, where its background element and foreground element are both within their color-keying ranges, in RGB888 format

bool ck_reverse_bg2fg: If this bit is set, in color-keying, for the pixel, where its background element is in the color range, but its foreground element is not in the color range, it will output the foreground element instead of the background element

ppa_trans_mode_t mode: Determines whether to block inside the operation functions, see ppa_trans_mode_t

void *user_data: User registered data to be passed into done_cb callback function

struct ppa_fill_oper_config_t

A collection of configuration items to do a PPA fill operation transaction.

Public Members

ppa_out_pic_blk_config_t out: Information of the output picture and the target block

uint32_t fill_block_w: The width of the block to be filled (unit: pixel)

uint32_t fill_block_h: The height of the block to be filled (unit: pixel)

color_pixel_argb8888_data_t fill_argb_color: The color to be filled, in ARGB8888 format

ppa_trans_mode_t mode: Determines whether to block inside the operation functions, see ppa_trans_mode_t

void *user_data: User registered data to be passed into done_cb callback function

Type Definitions

typedef struct ppa_client_t *ppa_client_handle_t: Type of PPA client handle.

typedef bool (*ppa_event_callback_t)(ppa_client_handle_t ppa_client, ppa_event_data_t *event_data, void *user_data)

Type of PPA event callback.

Param ppa_client: [in] PPA client handle
Param event_data: [in] PPA event data
Param user_data: [in] User registered data from calling ppa_do_xxx to perform an operation
Return: Whether a task switch is needed after the callback function returns, this is usually due to the callback wakes up some high priority task.

Enumerations

enum ppa_operation_t

Enumeration of all PPA available operations.

Values:

enumerator PPA_OPERATION_SRM: Do scale-rotate-mirror operation

enumerator PPA_OPERATION_BLEND: Do blend operation

enumerator PPA_OPERATION_FILL: Do fill operation, use one constant pixel to fill a target window

enumerator PPA_OPERATION_INVALID: Invalid PPA operations, indicates the quantity of available PPA operations

enum ppa_trans_mode_t

Modes to perform the PPA operations.

Values:

enumerator PPA_TRANS_MODE_BLOCKING: ppa_do_xxx function will block until the PPA operation is finished

enumerator PPA_TRANS_MODE_NON_BLOCKING: ppa_do_xxx function will return immediately after the PPA operation is pushed to the internal queue

Header File

components/hal/include/hal/ppa_types.h
This header file can be included with:
#include "hal/ppa_types.h"

Enumerations

enum ppa_engine_type_t

Enumeration of engines in PPA modules.

Values:

enumerator PPA_ENGINE_TYPE_SRM: PPA Scaling-Rotating-Mirroring (SRM) engine, used to perform scale, rotate, mirror

enumerator PPA_ENGINE_TYPE_BLEND: PPA Blending engine, used to perform blend or fill

enum ppa_srm_rotation_angle_t

Enumeration of PPA Scaling-Rotating-Mirroring available rotation angle (in the counterclockwise direction)

Values:

enumerator PPA_SRM_ROTATION_ANGLE_0: Picture does no rotation

enumerator PPA_SRM_ROTATION_ANGLE_90: Picture rotates 90 degrees CCW

enumerator PPA_SRM_ROTATION_ANGLE_180: Picture rotates 180 degrees CCW

enumerator PPA_SRM_ROTATION_ANGLE_270: Picture rotates 270 degrees CCW

enum ppa_srm_color_mode_t

Enumeration of PPA Scaling-Rotating-Mirroring available color mode.

Values:

enumerator PPA_SRM_COLOR_MODE_ARGB8888: PPA SRM color mode: ARGB8888

enumerator PPA_SRM_COLOR_MODE_RGB888: PPA SRM color mode: RGB888

enumerator PPA_SRM_COLOR_MODE_RGB565: PPA SRM color mode: RGB565

enumerator PPA_SRM_COLOR_MODE_YUV420: PPA SRM color mode: YUV420

enumerator PPA_SRM_COLOR_MODE_YUV444: PPA SRM color mode: YUV444 (limited range only)

enum ppa_blend_color_mode_t

Enumeration of PPA blend available color mode.

Values:

enumerator PPA_BLEND_COLOR_MODE_ARGB8888: PPA blend color mode: ARGB8888

enumerator PPA_BLEND_COLOR_MODE_RGB888: PPA blend color mode: RGB888

enumerator PPA_BLEND_COLOR_MODE_RGB565: PPA blend color mode: RGB565

enumerator PPA_BLEND_COLOR_MODE_A8: PPA blend color mode: A8, only available on blend foreground input

enumerator PPA_BLEND_COLOR_MODE_A4: PPA blend color mode: A4, only available on blend foreground input

enum ppa_fill_color_mode_t

Enumeration of PPA fill available color mode.

Values:

enumerator PPA_FILL_COLOR_MODE_ARGB8888: PPA fill color mode: ARGB8888

enumerator PPA_FILL_COLOR_MODE_RGB888: PPA fill color mode: RGB888

enumerator PPA_FILL_COLOR_MODE_RGB565: PPA fill color mode: RGB565

enum ppa_alpha_update_mode_t

Enumeration of PPA alpha compositing update mode.

Values:

enumerator PPA_ALPHA_NO_CHANGE: Do not replace alpha value (A' = A). If input format does not contain alpha info, alpha value 255 will be used.

enumerator PPA_ALPHA_FIX_VALUE: Replace the alpha value in received pixel with a new, fixed alpha value (A' = val)

enumerator PPA_ALPHA_SCALE: Scale the alpha value in received pixel to be a new alpha value (A' = (A * val) >> 8). If input format does not contain alpha info, A' = (255 * val) >> 8.

enumerator PPA_ALPHA_INVERT: Invert the alpha value in received pixel (A' = 255 - A). If input format does not contain alpha info, A' = 0, i.e. a layer with 0% opacity.

enum ppa_color_conv_std_rgb_yuv_t

Enumeration of PPA supported color conversion standard between RGB and YUV (determines the YUV<->RGB conversion equation)

Values:

enumerator PPA_COLOR_CONV_STD_RGB_YUV_BT601: YUV<->RGB conversion standard: BT.601

enumerator PPA_COLOR_CONV_STD_RGB_YUV_BT709: YUV<->RGB conversion standard: BT.709

enum ppa_color_range_t

Enumeration of PPA supported color range (determines the YUV<->RGB conversion equation)

Values:

enumerator PPA_COLOR_RANGE_LIMIT: Limited color range, 16 is the darkest black and 235 is the brightest white

enumerator PPA_COLOR_RANGE_FULL: Full color range, 0 is the darkest black and 255 is the brightest white

enum ppa_data_burst_length_t

Enumeration of PPA supported data burst length.

Values:

enumerator PPA_DATA_BURST_LENGTH_8: Data burst length: 8 bytes

enumerator PPA_DATA_BURST_LENGTH_16: Data burst length: 16 bytes

enumerator PPA_DATA_BURST_LENGTH_32: Data burst length: 32 bytes

enumerator PPA_DATA_BURST_LENGTH_64: Data burst length: 64 bytes

enumerator PPA_DATA_BURST_LENGTH_128: Data burst length: 128 bytes

Provide feedback about this document