CN109166169A - A kind of hardware realization of GPGPU texture sampler - Google Patents

A kind of hardware realization of GPGPU texture sampler Download PDF

Info

Publication number
CN109166169A
CN109166169A CN201810853065.9A CN201810853065A CN109166169A CN 109166169 A CN109166169 A CN 109166169A CN 201810853065 A CN201810853065 A CN 201810853065A CN 109166169 A CN109166169 A CN 109166169A
Authority
CN
China
Prior art keywords
texture
coordinate
data
gpgpu
sampler
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201810853065.9A
Other languages
Chinese (zh)
Inventor
龙斌
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Changsha In Blx Ic Design Corp
Original Assignee
Changsha In Blx Ic Design Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Changsha In Blx Ic Design Corp filed Critical Changsha In Blx Ic Design Corp
Priority to CN201810853065.9A priority Critical patent/CN109166169A/en
Publication of CN109166169A publication Critical patent/CN109166169A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T15/003D [Three Dimensional] image rendering
    • G06T15/04Texture mapping
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T1/00General purpose image data processing
    • G06T1/20Processor architectures; Processor configuration, e.g. pipelining

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Graphics (AREA)
  • Image Generation (AREA)
  • Image Processing (AREA)

Abstract

For in texture mapping application process a large amount of in GPGPU, if using common texture sampler, the problems such as upper layer logic needs cross the border according to the wraparound of texture coordinate, texture frame, need repeatedly to send texture reads request, then the operation such as synthesis is filtered to data according to multiple groups texture pixel, cause upper layer logic interface efficiency low, reduces overall performance.In order to solve this problem, the invention discloses a kind of hardware implementations of GPGPU texture sampling unit, the filtering operation of the texture picture pixel of address calculation and return when automatically completing texture wraparound and crossing the border, bottleneck effect of the texture sampler in 3 D image drawing engine is reduced, the overall operation speed of upper layer logic is greatly improved.

Description

A kind of hardware realization of GPGPU texture sampler
Technical field
Present invention relates generally to the 3-D graphic in the design of GPGPU graph processing chips to generate field.Refer in particular to pixel line The design of reason filling and the texture sampling unit of loading section.
Background technique
In the application of GPGPU graphics process, it is often necessary to use mapped pinup picture process (i.e. by the two dimension in caching Image data is loaded into the graph image of 3 D stereo modeling).Under current application, texture mapping process needs frequent visit Memory and the various operations of execution, design efficiency greatly restricts the speed and performance of three-dimensional drawing engine, in reality Often become the bottleneck of three-dimensional drawing engine in design.During texture mapping, texture sampler is responsible for from 2 d texture Multiple groups texture pixel color value is obtained in picture, and is translated into the real-texture pixel color component value of texture function needs Task, designing efficient texture sampler has crucial promotion to make GPGPU entirety 3 D image drawing engine performance With.
Summary of the invention
For in texture mapping application process a large amount of in GPGPU, if using common texture sampler, upper layer logic The problems such as needing the wraparound according to texture coordinate, texture frame to cross the border needs repeatedly to send texture reads request, then according to more Group texture pixel is filtered the operation such as synthesis to data, causes upper layer logic interface efficiency low, reduces overall performance.In order to Solve the problems, such as this, the invention discloses a kind of hardware implementations of GPGPU texture sampling unit, automatically complete texture around The filtering operation of the texture picture pixel of address calculation and return when returning and crossing the border reduces texture sampler in 3-D graphic Bottleneck effect in drawing engine greatly improves the overall operation speed of upper layer logic.
Compared with prior art, advantages of the present invention is that 1, structure is simple: texture sampler list proposed by the present invention Member is all to compare the algorithm realized conducive to chip hardware design, institutional framework is clear not using complicated algorithm;2, property It can be excellent: using the executive mode of full flowing water in texture sampler unit actual design process proposed by the present invention, and protecting In the case where demonstrate,proving operation execution efficiency, hardware resource is reduced as far as possible;3, it is widely used: texture sampler proposed by the present invention Unit not only can be used in GPGPU, can also be widely used in handle the pictures application;3, easy to use: the present invention The texture sampler unit input/output port signal and timing of proposition are clearly penetrating, have readable, reusability well And portability.
Detailed description of the invention
Fig. 1 texture sampler unit internal structure of the present invention;
Fig. 2 texture sampler coordinate system (rectangle of 16x16).
Specific embodiment
The present invention is described in further details below with reference to the drawings and specific embodiments.
As shown in Figure 1, texture sampler internal structure disclosed by the invention includes sample request parsing, sampling address life Five groups of unit modules are had altogether at data buffer storage, data texturing parsing, texture filtering processing in, piece.
Sample request parsing: this module receives the texture coordinate that top level control unit is sent and is parsed and decoupled;Such as Shown in Fig. 2, when overflow and underflow occurs in the coordinate of texture, need to read the border color value and texture picture face of respective texture Color executes filtering operation;When wraparound occurs in texture, need to read current line from texture picture according to different wraparound modes The pixel color value of coordinate and texture wraparound coordinate is managed, and records and send wraparound mode, instructs subsequent texture filtering Module is properly received the texture picture data that data buffer storage returns in piece.
Texture address generates: this module, which is responsible for being generated in corresponding piece according to the texture coordinate that texture resolution unit is sent, to be delayed Corresponding inquiry address is deposited, the tissue for inquiring address need to consider that the particularity of texture sampling algorithm is handled, and guarantee necessity Piece in cache reading efficiency.
Data buffer storage in piece: this module is as the data high-speed caching built in texture sampling unit, reasonable cache Internal data structure can reduce sampler to the number of external memory access, reduce the dependence to external memory bandwidth.
Data texturing resolution unit: according to current texture format, the texture picture data that data buffer storage in piece is exported It is parsed, is split as the texture component letter of the three-dimensional drawings engines such as transparency, red, green, blue coloration, intensity and depth support Breath.
Texture filtering processing unit: the filtering algorithm that this module is specified according to top level control unit, texture filtering algorithm branch The filtering algorithms such as nearest value algorithm and linear interpolation are held, nearest value algorithm is directly rounded according to the fractional part of texture coordinate Choose corresponding pixel;Line style interpolation carries out Weight operation to adjacent pixel according to texture coordinate fractional part, obtains true The real corresponding texture pixel component value of texture coordinate.

Claims (5)

1. texture sampling request analysis: being crossed the border situation according to texture coordinate and texture wraparound and texture frame, work as texture coordinate When there is overflow (i.e. developed width and height of the coordinate more than texture picture) and underflow (texture coordinate is calculated as negative), carry out The parsing of sample request is sent and the generation of cache access address conversion.
2. texture sampling address generates: the particularity accessed according to texture sampling, by texture transverse and longitudinal coordinate as unit of bit position It is divided into image rectangular block, the adjacent texture picture piecemeal of Low-level coordinate is stored.
3. interior data buffer storage: data high-speed caching is using image rectangular block as basic unit in piece;Carry out the loading of data texturing And replacement.
4. data texturing parses: according to the particularity of texture requirements for access, data buffer storage carries out efficient data access simultaneously in piece Picture pixels value in real-time return cache is completed at the same time image data to pixel transparent degree (Alpha), red (Red), green (Green), the conversion and parsing of blue (Blue) component.
5. texture filtering algorithm: texture filtering algorithm carries out on the basis of using nearest value algorithm and line style interpolation algorithm.
CN201810853065.9A 2018-07-30 2018-07-30 A kind of hardware realization of GPGPU texture sampler Pending CN109166169A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810853065.9A CN109166169A (en) 2018-07-30 2018-07-30 A kind of hardware realization of GPGPU texture sampler

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810853065.9A CN109166169A (en) 2018-07-30 2018-07-30 A kind of hardware realization of GPGPU texture sampler

Publications (1)

Publication Number Publication Date
CN109166169A true CN109166169A (en) 2019-01-08

Family

ID=64898744

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810853065.9A Pending CN109166169A (en) 2018-07-30 2018-07-30 A kind of hardware realization of GPGPU texture sampler

Country Status (1)

Country Link
CN (1) CN109166169A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112233159A (en) * 2020-09-23 2021-01-15 华夏芯(北京)通用处理器技术有限公司 Texture processing method and device
CN113194266A (en) * 2021-04-28 2021-07-30 深圳迪乐普数码科技有限公司 Image sequence frame real-time rendering method and device, computer equipment and storage medium

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5764237A (en) * 1994-10-07 1998-06-09 Kaneko; Koichi Texture mapping apparatus computing texture address by fill address
EP1434171A2 (en) * 1995-08-04 2004-06-30 Microsoft Corporation Method and system for texture mapping a source image to a destination image
CN101810485A (en) * 2009-02-23 2010-08-25 株式会社东芝 Medical image-processing apparatus and medical image processing method
US20110050687A1 (en) * 2008-04-04 2011-03-03 Denis Vladimirovich Alyshev Presentation of Objects in Stereoscopic 3D Displays
US20140215453A1 (en) * 2013-01-31 2014-07-31 Htc Corporation Methods for application management in an electronic device supporting hardware acceleration
CN106201390A (en) * 2015-05-29 2016-12-07 三星电子株式会社 Method and the electronic installation of support the method for output screen

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5764237A (en) * 1994-10-07 1998-06-09 Kaneko; Koichi Texture mapping apparatus computing texture address by fill address
EP1434171A2 (en) * 1995-08-04 2004-06-30 Microsoft Corporation Method and system for texture mapping a source image to a destination image
US20110050687A1 (en) * 2008-04-04 2011-03-03 Denis Vladimirovich Alyshev Presentation of Objects in Stereoscopic 3D Displays
CN101810485A (en) * 2009-02-23 2010-08-25 株式会社东芝 Medical image-processing apparatus and medical image processing method
US20140215453A1 (en) * 2013-01-31 2014-07-31 Htc Corporation Methods for application management in an electronic device supporting hardware acceleration
CN106201390A (en) * 2015-05-29 2016-12-07 三星电子株式会社 Method and the electronic installation of support the method for output screen

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
刘俞辰: "基于CUDA的快速三维超声成像研究", 《中国优秀硕士学位论文全文数据库(电子期刊)信息科技辑》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112233159A (en) * 2020-09-23 2021-01-15 华夏芯(北京)通用处理器技术有限公司 Texture processing method and device
CN113194266A (en) * 2021-04-28 2021-07-30 深圳迪乐普数码科技有限公司 Image sequence frame real-time rendering method and device, computer equipment and storage medium

Similar Documents

Publication Publication Date Title
Rupnow et al. High level synthesis of stereo matching: Productivity, performance, and software constraints
US9691118B2 (en) System for optimizing graphics operations
US11113788B2 (en) Multi-space rendering with configurable transformation parameters
US7847800B2 (en) System for emulating graphics operations
US7788656B2 (en) System for reducing the number of programs necessary to render an image
US8704837B2 (en) High-level program interface for graphics operations
CN109658337B (en) FPGA implementation method for real-time electronic despinning of images
CN1536475A (en) Method for sharing color reference table and device for pattern sample memory
JPH10116356A (en) Floating point processor for threedimensional graphics accelerator having floating point writing and set up core
CN101681502A (en) The pixel cache that is used for the three-dimensional picture circuit
TW200917228A (en) Compression of multiple-sample-anti-aliasing tile data in a graphics pipeline
JP7261226B2 (en) Arithmetic processing unit
CN109166169A (en) A kind of hardware realization of GPGPU texture sampler
US20200184002A1 (en) Hardware accelerated convolution
US20210295607A1 (en) Data reading/writing method and system in 3d image processing, storage medium and terminal
CN103760525A (en) Completion type in-place matrix transposition method
CN105550978B (en) A kind of GPU 3D engine on piece memory hierarchy towards unified dyeing framework
TWI471819B (en) Buffer clearing apparatus and method for computer graphics
CN110738594A (en) FPGA-based onboard electronic instrument image generation method
CN112734897B (en) Graphics processor depth data prefetching method triggered by primitive rasterization
CN105577985B (en) A kind of digital image processing system
JP2005134976A (en) Image processor for appearance inspection apparatus
WO2023202367A1 (en) Graphics processing unit, system, apparatus, device, and method
CN105093283A (en) Three-dimensional observation system surface element attribute multi-thread rapid display method
CN112070651A (en) Texture mapping hardware accelerator based on double Buffer architecture

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20190108