CN116597025B - Compressed texture decoding optimization method based on heterogeneous instruction penetration - Google Patents
Compressed texture decoding optimization method based on heterogeneous instruction penetration Download PDFInfo
- Publication number
- CN116597025B CN116597025B CN202310442918.0A CN202310442918A CN116597025B CN 116597025 B CN116597025 B CN 116597025B CN 202310442918 A CN202310442918 A CN 202310442918A CN 116597025 B CN116597025 B CN 116597025B
- Authority
- CN
- China
- Prior art keywords
- compressed texture
- texture
- compressed
- decoding
- shared memory
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000005457 optimization Methods 0.000 title claims abstract description 16
- 230000035515 penetration Effects 0.000 title claims abstract description 10
- 230000015654 memory Effects 0.000 claims abstract description 65
- 230000008569 process Effects 0.000 claims description 17
- 238000009877 rendering Methods 0.000 claims description 14
- 230000006837 decompression Effects 0.000 claims description 10
- 238000012544 monitoring process Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 abstract description 16
- 238000012545 processing Methods 0.000 abstract description 13
- 238000006243 chemical reaction Methods 0.000 abstract description 6
- 230000006835 compression Effects 0.000 description 3
- 238000007906 compression Methods 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 238000013507 mapping Methods 0.000 description 2
- 101100065701 Arabidopsis thaliana ETC2 gene Proteins 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000003139 buffering effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/54—Interprogram communication
- G06F9/544—Buffers; Shared memory; Pipes
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- General Engineering & Computer Science (AREA)
- Stored Programmes (AREA)
Abstract
The invention discloses a compressed texture decoding optimization method based on heterogeneous instruction penetration, which is characterized in that for a GPU platform supporting processing of compressed textures, a shared memory between An Zhuoduan and a Host end is created according to the decoding condition of the compressed textures when the recorded android application runs for the first time, so that the overhead generated by transmitting the compressed textures is reduced; for a GPU platform which does not support compressed texture processing, parameters of the texture decoding function are transferred to a Host x86 version decoding function instead of an ARM version texture decoding function, and the Host completes decoding operation of compressed texture, so that time consumed by ARM conversion to an x86 instruction is saved.
Description
Technical Field
The invention belongs to the technical field of cross-platform application development, and particularly relates to a compressed texture decoding optimization method based on heterogeneous instruction penetration.
Background
In order to reduce the memory space occupied by mobile applications in a mobile operating system, texture compression format is often adopted to store textures. The texture compression algorithm actually used must meet the following conditions: the method has the advantages of high-speed and real-time decoding capability, and partial decoding of textures. The texture compression algorithm meeting the conditions can be adopted to realize high-speed texture mapping, and meanwhile, less high-speed CACHE can be used for buffering partial texture data currently used at any time in the process of texture mapping.
The compressed texture format adopted by the android system is generally ETC2 or ASTC, and the most currently used mainstream format is ASTC. In an android compatible environment or android simulator, decoding of compressed textures is typically achieved in two modes: firstly, if the GPU supports the processing of the current compressed texture format, the compressed texture is transmitted to a Host end, and decoding is completed by the GPU drive of the Host end; and secondly, if the GPU does not support the processing of the current compressed texture format, the CPU can only be used for decoding.
However, there are problems with android compatible environments or with both of the above approaches of android simulators: the main problem for the manner in which compressed textures need to be transferred between An Zhuoduan and Host is that the process of transferring textures increases the overhead, while the main problem for the manner in which CPU is used for decoding is that the instruction conversion required for decoding android application textures by CPU is long and the processing efficiency is low, for example, compressed textures with size of 1MB may need 500 mm or more to complete decoding.
Disclosure of Invention
In view of the above, the present invention provides a compressed texture decoding optimization method based on heterogeneous instruction penetration, which realizes the compressed texture decoding optimization of supporting heterogeneous instructions, which is transparent to applications.
The invention provides a compressed texture decoding optimization method based on heterogeneous instruction penetration, which comprises the following steps:
step 1, acquiring a compressed texture format supported by a Host GPU when starting An Zhuoduan, and taking the compressed texture format as a first compressed texture format;
step 2, when the android application in An Zhuoduan starts a rendering process of compressed textures, obtaining an android application ID, a compressed texture size, a compressed texture block offset to be decoded and a compressed texture block size to be decoded, if the android application ID exists in compressed texture understanding pressure history information, reading compressed texture understanding pressure history information of the android application, and then executing step 3, otherwise, executing step 4; the compressed texture decompression history information stores an android application ID, a second compressed texture format, a compressed texture ID, a compressed texture size, a decoded compressed texture block offset and a decoded compressed texture block size; the second compressed texture format is a compressed texture format of a compressed texture;
step 3, if the second compressed texture format recorded in the compressed texture understanding pressure history information exists in the first compressed texture format, executing step 4, otherwise executing step 9;
step 4, an Zhuoduan searches whether related information exists in the compressed texture understanding pressure history information according to the compressed texture ID, and if so, step 5 is executed; otherwise, executing the step 6;
step 5, distributing a shared memory according to the compressed texture size recorded in the compressed texture decompression history information, recording a shared memory pointer, and writing the compressed texture block to be decoded into the shared memory; taking the decoded compressed texture block offset corresponding to the compressed texture block to be decoded recorded in the compressed texture decompression history information as the texture block offset in the shared memory; transmitting the shared memory pointer and the texture block offset to a Host end, and executing the step 7;
step 6, adding the compressed texture ID, the compressed texture size, the compressed texture block offset to be decoded and the compressed texture block size to be decoded into compressed texture understanding and compressing history information, distributing shared memory record shared memory pointers according to the compressed texture block size, writing the compressed texture block to be decoded into the shared memory, and taking the compressed texture block offset to be decoded as the texture block offset in the shared memory; transmitting the shared memory pointer and the texture block offset to a Host end, and executing the step 7;
step 7, the Host informs the GPU to read the texture block from the shared memory according to the received shared memory pointer and the texture block offset to execute decoding operation, after decoding is completed, a decoding completion message is sent to An Zhuoduan, and then rendering operation of the texture block is executed;
step 8, after receiving the decoding completion message, the android receives the decoding and rendering operation of the bypass An Zhuoduan, and then executes the subsequent operation of the android application;
step 9, an Zhuoduan starts the monitoring of the ARM library loading process, if the compressed texture decoding operation occurs in the ARM library loading process, the texture parameters of the compressed texture decoding operation are obtained, and the Host end is adopted to finish the decoding process of the compressed texture by taking the texture parameters as input based on the compressed texture decoding operation realized by the CPU.
Further, in the step 2, the android application ID is a path of the android application.
Further, the compressed texture ID is a path of an image used by the compressed texture.
Further, the step 5 further includes: and obtaining the sizes of all decoded compressed texture blocks recorded in the compressed texture decompression history information, summing the sizes of all decoded compressed texture blocks to obtain the size of an allocation space, and allocating the shared memory according to the size of the allocation space.
Further, in the step 6, a shared memory is allocated according to the compressed texture block size obtained in the step 2.
Advantageous effects
For the GPU platform supporting processing of the compressed texture, the shared memory between An Zhuoduan and the Host end is created according to the decoding condition of the compressed texture when the recorded android application runs for the first time, so that the overhead generated by transmitting the compressed texture is reduced; for a GPU platform which does not support compressed texture processing, parameters of the texture decoding function are transferred to a Host x86 version decoding function instead of an ARM version texture decoding function, and the Host completes decoding operation of compressed texture, so that time consumed by ARM conversion to an x86 instruction is saved.
Detailed Description
The present invention will be described in detail with reference to the following examples.
The invention provides a compressed texture decoding optimization method based on heterogeneous instruction penetration, which has the following core ideas: determining an optimization scheme according to the supporting condition of the GPU platform on the compressed texture, wherein for the GPU platform supporting processing the compressed texture, a shared memory between An Zhuoduan and a Host end is created according to the decoding condition of the compressed texture when the recorded android application runs for the first time so as to reduce the overhead generated by transmitting the compressed texture; for a GPU platform which does not support compressed texture processing, parameters of the texture decoding function are transferred to a Host x86 version decoding function instead of an ARM version texture decoding function, and the Host completes decoding operation of compressed texture, so that time consumed by ARM conversion to an x86 instruction is saved.
The invention provides a compressed texture decoding optimization method based on heterogeneous instruction penetration, which specifically comprises the following steps:
step 1, acquiring a compressed texture format supported by a GPU currently used by a Host terminal when starting An Zhuoduan, and taking the compressed texture format as a first compressed texture format.
And 2, when the android application in An Zhuoduan starts a rendering process of the compressed texture, acquiring an android application ID, a compressed texture size, a compressed texture block offset to be decoded and a compressed texture block size to be decoded, judging whether related information of the android application exists in compressed texture understanding pressure historical information according to the android application ID, if so, reading the compressed texture understanding pressure historical information of the android application, and then executing the step 3, otherwise, executing the step 4.
The compressed texture understanding and compressing history information stores an android application ID, a second compressed texture format, a compressed texture ID, a compressed texture size, a decoded compressed texture block offset and a decoded compressed texture block size, wherein the second compressed texture format is a compressed texture format of a compressed texture currently processed by the android application. The android application ID is a path of the android application, and the compressed texture ID is a path of an image used by the compressed texture.
And step 3, judging whether the second compressed texture format recorded in the compressed texture understanding pressure history information exists in the first compressed texture format, if so, executing the step 4, otherwise, executing the step 9.
Step 4, an Zhuoduan searches whether related information exists in the compressed texture understanding pressure history information according to the compressed texture ID, and if so, step 5 is executed; otherwise, executing step 6.
Step 5, obtaining the compressed texture size recorded in the compressed texture understanding and compressing history information, distributing a shared memory according to the compressed texture size, recording a shared memory pointer, and writing the compressed texture block to be decoded into the shared memory; taking the offset of the decoded texture block corresponding to the compressed texture block to be decoded recorded in the compressed texture decompression history information as the offset of the decoded texture block in the shared memory; and (7) sending the shared memory pointer and the texture block offset to a Host end, and executing the step (7).
In order to further improve the memory utilization rate, in step 5, the method first acquires all the decoded compressed texture block sizes recorded in the compressed texture decompression history information, then sums all the decoded compressed texture block sizes to obtain the allocation space size, and then allocates the shared memory according to the allocation space size.
Step 6, adding the compressed texture ID, the compressed texture size, the compressed texture block offset to be decoded and the compressed texture block size to be decoded into compressed texture understanding and compressing history information, distributing shared memory record shared memory pointers according to the compressed texture block size, writing the compressed texture block to be decoded into the shared memory, and taking the compressed texture block offset to be decoded as the offset of the compressed texture block offset in the shared memory; and (7) sending the shared memory pointer and the texture block offset to a Host end, and executing the step (7).
In order to further improve the memory utilization rate, in step 6, the present invention allocates a shared memory according to the compressed texture block size obtained in step 2.
And 7, informing the GPU by the Host end to read the texture block from the shared memory according to the received shared memory pointer and the texture block offset to execute the decoding operation of the texture block, sending a decoding completion message to An Zhuoduan after finishing decoding, and then executing the rendering operation of the texture block.
And step 8, after receiving the decoding completion message, the android terminal bypasses the decoding and rendering operations of An Zhuoduan, and then executes subsequent operations of the android application.
Step 9, an Zhuoduan starts the monitoring of the ARM library loading process, if the compressed texture decoding operation occurs in the ARM library loading process, the texture parameters of the compressed texture decoding operation are obtained, and the Host end is adopted to finish the decoding process of the compressed texture by taking the texture parameters as input based on the compressed texture decoding operation realized by the CPU.
Examples
The compressed texture decoding optimization method based on heterogeneous instruction penetration provided by the embodiment of the invention realizes optimization of the compressed texture decoding process of android application in a desktop system, and specifically comprises the following steps:
s1, acquiring a compressed texture format supported by a GPU currently used by a Host terminal when starting An Zhuoduan.
S1.1, acquiring the number of compressed texture formats supported by the GPU currently used by the Host.
Taking OpenGL/OpenGL ES as an example, the number of GPU supported COMPRESSED TEXTURE FORMATS can be obtained by invoking glgetintelgerv using gl_num_compressed_text_formats or gl_num_compressed_text_formats_arb as parameters.
S1.2, enumerating compressed texture formats supported by all the GPUs.
Taking OpenGL/OpenGL ES as an example, using gl_compressed_text_format or gl_compressed_format_arb as a parameter to call glgetintelgerv can obtain the names of the COMPRESSED TEXTURE FORMATS supported by the GPU, and record all supported COMPRESSED TEXTURE FORMATS in the list supportedcompressed texturelist.
S2, when the android application in An Zhuoduan starts a rendering process of the compressed texture, acquiring an android application ID, a compressed texture size, a compressed texture block offset to be decoded and a compressed texture block size to be decoded, judging whether related information of the android application exists in compressed texture understanding pressure history information appCompressedToxoport according to the android application ID, if so, reading the compressed texture understanding pressure history information of the android application, and then executing S3, otherwise, executing S4.
The android application ID, the second compressed texture format, the compressed texture ID, the compressed texture size, the decoded compressed texture block offset and the decoded compressed texture block size are stored in the compressed texture understanding and compressing history information. The android application ID is a path of the android application, and the compressed texture ID is a path of an image used by the compressed texture.
S3, modifying a compressed texture decoding function of OpenGL, openGL ES or Vulkan, for example, a compressed texture decoding function of the glCompressedTeximage2D or the glCompressedTexSubImage2D of the OpenGL or OpenGL ES, or the compressed texture partial decoding function of the glCompressedTexSubImage2D, judging whether the currently processed compressed texture format exists in a supplementary when the android application executes compressed texture decoding, and executing S4 if the compressed texture format exists in the supplementary; otherwise, S9 is performed.
S4, an Zhuoduan, searching whether related information exists in the appCompressedToxex according to the compressed texture ID, and if so, indicating that the GPU supports the execution of processing of the compressed texture S5; otherwise, the GPU does not support S6 for compressed texture processing.
S5, for a function glCompressedToxImage 2D, creating a shared memory according to the size of the texture, storing the compressed texture into the shared memory, and transmitting the shared memory pointer to a Host end as a compressed texture buffer; for a function glCompressedTexSubImage2D, searching in an appCompressedTexHistory according to a compressed texture ID, an offset of a compressed texture block to be decoded and the size of the compressed texture block to be decoded, acquiring the sizes of all decoded compressed texture blocks related to the compressed texture ID in the appCompressedTexHistory if the compressed texture ID exists in the appCompressedTexHistory, summing the sizes of all decoded compressed texture blocks to obtain texBlockTotalSize, distributing shared memories with the same size for the compressed texture ID according to the texBlockTotalSize, writing the compressed texture block to be decoded into the shared memories, using a shared memory pointer as a compressed texture buffer, and transmitting the offset of the compressed texture block to be decoded as the shared memory offset of the compressed texture block to a Host end; s7 is performed.
Since the Host GPU is capable of rendering compressed textures directly, or the decoding of compressed textures is done inside the GPU, the present invention does not actually allocate a memory buffer for storing the decoding result at An Zhuoduan.
S6, adding the compressed texture ID, the compressed texture size, the compressed texture block offset to be decoded and the compressed texture block size to be decoded into an appCompressedToxHistory, then distributing a shared memory record shared memory pointer according to the compressed texture size, storing the compressed texture into the shared memory, and taking the compressed texture block offset to be decoded as the offset of the compressed texture block offset in the shared memory; and sending the shared memory pointer and the texture block offset to a Host end, and executing S7.
And S7, informing the GPU by the Host end to read the texture block from the shared memory according to the received shared memory pointer and the texture block offset to execute the decoding operation of the texture block, sending a decoding completion message to An Zhuoduan after decoding is completed, and executing the rendering operation of the texture block.
In the invention, the Host end does not need to actually allocate a memory buffer for storing decoding results, but calls a compressed texture processing interface of the GPU to directly transfer compressed textures to the video memory, and the GPU does not actually send the decoding results to An Zhuoduan after processing, and only informs An Zhuoduan that decoding is completed and the rendering operation of the current frame is completed at the Host end.
And S8, after receiving the decoding completion message, the android terminal bypasses An Zhuoduan decoding and rendering operations, such as glFlush or glSwapBuffer, and executing subsequent operations of the android application.
S9, modifying a Native Bridge mechanism of An Zhuoduan, wherein the Native Bridge mechanism is a heterogeneous instruction conversion loading mechanism, monitoring an ARM library loading process related to the current android application, and if the ARM library loaded by the current android application contains a compressed texture decoding function, for example: glCompressedTimeImage 2D, glCompressedTexSubImage D, etc., the above compressed texture decoding function is not called to be converted into a compressed texture decoding operation based on a CPU in the prior art, and the decoding operation of the compressed texture is completed.
The invention actually uses the compressed texture decoding realization based on the X86 CPU to replace the compressed texture decoding realization based on the ARM, thereby avoiding the huge expenditure caused by instruction conversion.
In summary, the above embodiments are only preferred embodiments of the present invention, and are not intended to limit the scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (5)
1. The compressed texture decoding optimization method based on heterogeneous instruction penetration is characterized by comprising the following steps of:
step 1, acquiring a compressed texture format supported by a Host GPU when starting An Zhuoduan, and taking the compressed texture format as a first compressed texture format;
step 2, when the android application in An Zhuoduan starts a rendering process of compressed textures, obtaining an android application ID, a compressed texture size, a compressed texture block offset to be decoded and a compressed texture block size to be decoded, if the android application ID exists in compressed texture understanding pressure history information, reading compressed texture understanding pressure history information of the android application, and then executing step 3, otherwise, executing step 4; the compressed texture decompression history information stores an android application ID, a second compressed texture format, a compressed texture ID, a compressed texture size, a decoded compressed texture block offset and a decoded compressed texture block size; the second compressed texture format is a compressed texture format of a compressed texture;
step 3, if the second compressed texture format recorded in the compressed texture understanding pressure history information exists in the first compressed texture format, executing step 4, otherwise executing step 9;
step 4, an Zhuoduan searches whether related information exists in the compressed texture understanding pressure history information according to the compressed texture ID, and if so, step 5 is executed; otherwise, executing the step 6;
step 5, distributing a shared memory according to the compressed texture size recorded in the compressed texture decompression history information, recording a shared memory pointer, and writing the compressed texture block to be decoded into the shared memory; taking the decoded compressed texture block offset corresponding to the compressed texture block to be decoded recorded in the compressed texture decompression history information as the texture block offset in the shared memory; transmitting the shared memory pointer and the texture block offset to a Host end, and executing the step 7;
step 6, adding the compressed texture ID, the compressed texture size, the compressed texture block offset to be decoded and the compressed texture block size to be decoded into compressed texture understanding and compressing history information, distributing shared memory record shared memory pointers according to the compressed texture block size, writing the compressed texture block to be decoded into the shared memory, and taking the compressed texture block offset to be decoded as the texture block offset in the shared memory; transmitting the shared memory pointer and the texture block offset to a Host end, and executing the step 7;
step 7, the Host informs the GPU to read the texture block from the shared memory according to the received shared memory pointer and the texture block offset to execute decoding operation, after decoding is completed, a decoding completion message is sent to An Zhuoduan, and then rendering operation of the texture block is executed;
step 8, after receiving the decoding completion message, the android receives the decoding and rendering operation of the bypass An Zhuoduan, and then executes the subsequent operation of the android application;
step 9, an Zhuoduan starts the monitoring of the ARM library loading process, if the compressed texture decoding operation occurs in the ARM library loading process, the texture parameters of the compressed texture decoding operation are obtained, and the Host end is adopted to finish the decoding process of the compressed texture by taking the texture parameters as input based on the compressed texture decoding operation realized by the CPU.
2. The compressed texture decoding optimization method according to claim 1, wherein the android application ID in the step 2 is a path of an android application.
3. The compressed texture decoding optimization method of claim 1, wherein the compressed texture ID is a path of an image used by the compressed texture.
4. The compressed texture decoding optimization method according to claim 1, wherein the step 5 further comprises: and obtaining the sizes of all decoded compressed texture blocks recorded in the compressed texture decompression history information, summing the sizes of all decoded compressed texture blocks to obtain the size of an allocation space, and allocating the shared memory according to the size of the allocation space.
5. The method according to claim 1, wherein in step 6, the shared memory is allocated according to the compressed texture block size obtained in step 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310442918.0A CN116597025B (en) | 2023-04-24 | 2023-04-24 | Compressed texture decoding optimization method based on heterogeneous instruction penetration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310442918.0A CN116597025B (en) | 2023-04-24 | 2023-04-24 | Compressed texture decoding optimization method based on heterogeneous instruction penetration |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116597025A CN116597025A (en) | 2023-08-15 |
| CN116597025B true CN116597025B (en) | 2023-09-26 |
Family
ID=87598258
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310442918.0A Active CN116597025B (en) | 2023-04-24 | 2023-04-24 | Compressed texture decoding optimization method based on heterogeneous instruction penetration |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116597025B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107003892A (en) * | 2016-12-29 | 2017-08-01 | 深圳前海达闼云端智能科技有限公司 | GPU vitualization method, device, system and electronic equipment, computer program product |
| CN114025238A (en) * | 2022-01-10 | 2022-02-08 | 北京蔚领时代科技有限公司 | Native android application cloud virtualization method based on Linux server |
| CN114035967A (en) * | 2022-01-10 | 2022-02-11 | 北京蔚领时代科技有限公司 | Resource sharing rendering method and device based on server |
| CN114945096A (en) * | 2022-05-26 | 2022-08-26 | 合肥埃科光电科技股份有限公司 | Lossless decompression algorithm based on CPU + GPU heterogeneous platform and storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9159114B2 (en) * | 2013-11-08 | 2015-10-13 | Qualcomm Incorporated | Texture decompression for graphics processors |
| KR102245137B1 (en) * | 2014-09-17 | 2021-04-28 | 삼성전자 주식회사 | Apparatus and method for decompressing rendering data and recording medium thereof |
-
2023
- 2023-04-24 CN CN202310442918.0A patent/CN116597025B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107003892A (en) * | 2016-12-29 | 2017-08-01 | 深圳前海达闼云端智能科技有限公司 | GPU vitualization method, device, system and electronic equipment, computer program product |
| CN114025238A (en) * | 2022-01-10 | 2022-02-08 | 北京蔚领时代科技有限公司 | Native android application cloud virtualization method based on Linux server |
| CN114035967A (en) * | 2022-01-10 | 2022-02-11 | 北京蔚领时代科技有限公司 | Resource sharing rendering method and device based on server |
| CN114945096A (en) * | 2022-05-26 | 2022-08-26 | 合肥埃科光电科技股份有限公司 | Lossless decompression algorithm based on CPU + GPU heterogeneous platform and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 沈春江.面向OpenGL图形流水线的纹理载入单元设计与验证.《中国优秀硕士学位论文全文数据库 信息科技辑》.2019,(第2019年第02期),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116597025A (en) | 2023-08-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7533237B1 (en) | Off-chip memory allocation for a unified shader | |
| CN113515396B (en) | Graphics rendering method, graphics rendering device, electronic equipment and storage medium | |
| US8407443B1 (en) | Off-chip out of order memory allocation for a unified shader | |
| EP3594807A1 (en) | Virtual disk file format conversion method and device | |
| JP2010509688A (en) | Pixel cache for 3D graphics circuits | |
| JPWO2012081099A1 (en) | Data migration program, computer and data migration method | |
| US10558496B2 (en) | Techniques for accessing a graphical processing unit memory by an application | |
| US20220004405A1 (en) | 3D API Redirection for Virtual Desktop Infrastructure | |
| US11645145B2 (en) | Methods and apparatus to facilitate speculative page fault handling in a graphics processing unit | |
| WO2005086096A2 (en) | Embedded system with 3d graphics core and local pixel buffer | |
| CN106415655B (en) | Virtual memory supported compression control surface | |
| CN115827506A (en) | Data writing method, data reading method, device, processing core and processor | |
| CN111881104A (en) | NFS server, data writing method and device thereof, and storage medium | |
| CN116597025B (en) | Compressed texture decoding optimization method based on heterogeneous instruction penetration | |
| KR100809293B1 (en) | Apparatus and method for managing stacks in virtual machine | |
| US9471584B2 (en) | Demand paging method for mobile terminal, controller and mobile terminal | |
| CN111240765B (en) | LINUX compression application program loading method | |
| CN109727183B (en) | Scheduling method and device for compression table of graphics rendering buffer | |
| US20100115141A1 (en) | Processor and method for controlling memory | |
| CN114911525A (en) | Binary instruction translation execution method and device | |
| CN113704588A (en) | File reading method and system based on mapping technology | |
| CN117435532B (en) | Copying method, device and storage medium based on video hardware acceleration interface | |
| CN117312201B (en) | Data transmission method and device, accelerator equipment, host and storage medium | |
| EP4235562A1 (en) | Image processing method and electronic device | |
| CN117830073A (en) | Graphics processor, data processing method thereof and electronic device comprising the same |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |