CN107562659A - A kind of data-moving device and method - Google Patents
A kind of data-moving device and method Download PDFInfo
- Publication number
- CN107562659A CN107562659A CN201610505894.9A CN201610505894A CN107562659A CN 107562659 A CN107562659 A CN 107562659A CN 201610505894 A CN201610505894 A CN 201610505894A CN 107562659 A CN107562659 A CN 107562659A
- Authority
- CN
- China
- Prior art keywords
- data
- dma
- moving
- management module
- instruction
- 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
Links
Landscapes
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses a kind of data-moving device, including：Direct memory access DMA administrative units, for sending data-moving instruction to DMA unit；DMA unit, for performing data-moving instruction.The invention also discloses a kind of data-moving method, including direct memory access DMA administrative units to send data-moving and instruct to DMA unit, and performs data-moving instruction by DMA unit.Compared with prior art, data-moving device and method of the invention makes DMA data-moving process depart from CPU control, allows processor resource to be released.
Description
Technical field
The present invention relates to communication technique field, more particularly to a kind of data-moving device and method.
Background technology
Traditional die internal data, which is moved, mainly two ways, CPU move with DMA (Direct Memory Access,
Direct memory access) move.The problem of CPU has been moved easy to use, the advantages that versatility is good, but maximum is to waste valuable place
Manage device resource.The benefit that DMA is moved is that after CPU completes initial configuration to DMA, follow-up data is moved process and taken over by DMA, CPU
Part liberation is obtained, other things can be done parallel, systematic function is necessarily lifted.
It is exactly that must be moved before moving every time by special signal enabling but DMA is moved there are still a problem,
So CPU still needs to stop work carries out data-moving to trigger DMA, the resource of processor still can be so wasted.
The content of the invention
The main object of the present invention is to provide a kind of data-moving device, it is intended to DMA data-moving process is departed from CPU
Control, and by the pure hardware controls of DMA administrative units, allow processor resource to be released.
To achieve the above object, the present invention proposes a kind of data-moving device, including：
Direct memory access DMA administrative units, for sending data-moving instruction to DMA unit；
The DMA unit, for performing the data-moving instruction.
Preferably, the DMA administrative units include：
Caching management module, the process of data is write toward physical memory for monitoring processor, and sent out to queue management module
Buffer status is sent to indicate.
The queue management module, task is moved according to DMA unit current state distribution；
Preferably, the caching management module includes：
Processor monitors submodule, and the progress of data is write for monitoring processor toward the physical memory；
Threshold values contrasts submodule, and whether the data volume for judging in the physical memory reaches the threshold values moved；
State instruction sending submodule, indicated for sending the buffer status to the queue management module.
Preferably, the queue management module includes：
State instruction receiving submodule, for receiving the caching shape sent at the state instruction sending submodule
State indicates；
Descriptor takes out submodule, for taking out initialized task descriptor in the queue management module；
Task sending submodule, for sending the instruction of the data-moving to the DMA unit.
Preferably, the queue management module also includes：
Task splits submodule, for the data volume for needing to move to be split；
Arbitrate submodule, according to the queue priority of the queue management module, sequencing that determination data is moved.
Present invention also offers a kind of data-moving method, including：Direct memory access DMA administrative units send data and removed
Instruction is moved to DMA unit, and the data-moving is performed by the DMA unit and instructed.
Preferably, the DMA administrative units send the data-moving and instructed to the DMA unit, in addition to：Initially
Change the DMA administrative units.
Preferably, the initialization DMA administrative units include：
Physical memory is distributed to the queue management module of the DMA administrative units；
The threshold values of the caching management module of the DMA administrative units is set；
The queue priority of the queue management module is set；
Initialize the task descriptor of the queue management module；
The task descriptor is write in the queue management module.
Preferably, the DMA administrative units send data-moving and instructed to DMA unit, including：
Monitoring processor writes the process of data toward physical memory, and sends buffer status instruction；
Order caching state instruction, and being moved according to DMA unit current state distribution for task.
Preferably, the monitoring processor writes the process of data toward physical memory, and sends buffer status instruction, including：
Monitor the progress that processor writes data toward the physical memory；
Judge whether the data volume in the physical memory reaches the threshold values moved；
If reaching threshold values, buffer status instruction is sent；If not up to threshold values, continue to monitor processor toward the physics
Internal memory writes the progress of data.
Preferably, the order caching state instruction, and being moved according to DMA unit current state distribution for task,
Including：
Receive the buffer status instruction；
Take out the task descriptor；
Split the data volume for needing to move；
According to the queue priority, sequencing that determination data is moved；
The instruction of data-moving is sent to the DMA unit.
Technical scheme has following beneficial effect：
Data-moving device in the present invention is made up of hardware completely, and the process of data-moving method of the present invention completely by
Hardware controls.Thus, unit interval inner treater can perform more instructions, and processor tasks are shunted, for wireless base
For station equipment, possesses the possibility of access more users.Meanwhile the present invention uses the process of hardware controls data-moving completely,
Can reduce because software control data is moved and caused by error situation.Compared with software control, the hardware cost that newly increases
It is lower, and the space that program takes is less.
Brief description of the drawings
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be in embodiment or description of the prior art
The required accompanying drawing used is briefly described, it should be apparent that, drawings in the following description are only some realities of the present invention
Example is applied, for those of ordinary skill in the art, on the premise of not paying creative work, can also be according to these accompanying drawings
The structure shown obtains other accompanying drawings.
Fig. 1 is a kind of optional structural representation of data-moving device in one embodiment of the invention；
Fig. 2 is a kind of optional structural representation of DMA administrative units in one embodiment of the invention；
Fig. 3 is a kind of optional structural representation of caching management module in one embodiment of the invention；
Fig. 4 is a kind of optional structural representation of queue management module in one embodiment of the invention；
Fig. 5 is another optional structural representation of queue management module in one embodiment of the invention；
Fig. 6 is a kind of optional flow chart of data-moving method in one embodiment of the invention；
A kind of optional flow chart further refined that Fig. 7 is Fig. 6；
The optional flow chart further refined of another kind that Fig. 8 is Fig. 6；
Fig. 9 is the application scenarios of the embodiment of data-moving device and method one of the present invention；
Figure 10 is structure chart of the embodiment of data-moving device one of the present invention under Fig. 6 application scenarios；
Figure 11 is flow chart of the embodiment of data-moving method one of the present invention under Fig. 6 application scenarios.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only the part of the embodiment of the present invention, rather than whole embodiments.Base
Embodiment in the present invention, those of ordinary skill in the art obtained under the premise of creative work is not made it is all its
His embodiment, belongs to the scope of protection of the invention.
The concrete structure of the data-moving device of one embodiment of the invention can refer to Fig. 1-5.
Such as Fig. 1, the data-moving device of the present embodiment includes：DMA administrative units 101 and DMA unit 102.DMA management is single
Member 101 is used to send data-moving instruction to DMA unit 102, and DMA unit 102 is used to perform data-moving instruction.
Such as Fig. 2, the DMA administrative units 101 of the present embodiment include：Cache management mould 103 and queue management module 104.It is slow
Deposit management mould 103 and be used for the process that monitoring processor writes data toward physical memory, and caching shape is sent to queue management module 104
State indicates that queue management module 104 moves task according to the distribution of the current state of DMA unit 102.
Such as Fig. 3, the caching management module 103 of the present embodiment includes：Processor monitors submodule 105, threshold values contrast submodule
Block 106 and state instruction sending submodule 107.Processor monitors submodule 105 and writes number toward physical memory for monitoring processor
According to progress, threshold values contrast submodule 106 be used for judge whether the data volume in physical memory reaches the threshold values moved, state refers to
Show that sending submodule 107 is used to send buffer status instruction to queue management module 104.
Such as Fig. 4, the queue management module of the present embodiment includes in one case：State instruction receiving submodule 108, retouch
State symbol and take out submodule 109 and task sending submodule 1010.State instruction receiving submodule 108 is used to receive to be referred to from state
Show the buffer status instruction sent at sending submodule 107, descriptor takes out submodule 109 and is used to take out queue management module
Initialized task descriptor in 104, task sending submodule 1010 are used for the finger that data-moving is sent to DMA unit 102
Order.
Such as Fig. 5, the queue management module of the present embodiment includes in another case：State instruction receiving submodule 108,
Descriptor takes out submodule 109, task sending submodule 1010, task and splits submodule 1011, arbitration submodule 1012.State
Indicate that receiving submodule 108 is used to receive the buffer status instruction sent at state instruction sending submodule 107, descriptor
Submodule 109 is taken out to be used to take out initialized task descriptor in queue management module 104, task sending submodule 1010
For sending the instruction of data-moving to DMA unit 102, task splits submodule 1011 and is used to enter the data volume for needing to move
Row is split, and arbitration submodule 1012 is according to the queue priority of queue management module 104, the sequencing that determination data is moved.
The idiographic flow of the data-moving method of one embodiment of the invention is as shown in figs 6-8.
Such as Fig. 6, the data-moving method of the present embodiment, including：
Step 201：Initialize direct memory access DMA administrative units；
Step 202：DMA administrative units send data-moving and instructed to DMA unit；
Step 203：DMA unit performs data-moving instruction.
Wherein, initialization DMA administrative units include：Physical memory is distributed to the queue management module of DMA administrative units；If
Put the threshold values of the caching management module of DMA administrative units；The queue priority of queue management module is set；Initialize queue management
The task descriptor of module；Task descriptor is write in queue management module.
Such as Fig. 7, DMA administrative units send data-moving and instructed to DMA unit in the present embodiment, including：
Step 301：Monitoring processor writes the process of data toward physical memory, and sends buffer status instruction；
Step 302：Order caching state instruction, and being moved according to the distribution of DMA unit current state for task.
Such as Fig. 8, monitoring processor writes the process of data toward physical memory in the present embodiment, and sends buffer status instruction bag
Include：
Step 401：Monitor the progress that processor writes data toward physical memory；
Step 402：Judge whether the data volume in physical memory reaches the threshold values moved；
Step 403：If reaching threshold values, buffer status instruction is sent；If not up to threshold values, it is past to continue monitoring processor
Physical memory writes the progress of data.
Such as Fig. 8, order caching state instruction in the present embodiment, and being moved according to the distribution of DMA unit current state for task
Including：
Step 404：Order caching state instruction；
Step 405：Take out initialized task descriptor；
Step 406：Split the data volume for needing to move；
Step 407：According to queue priority, sequencing that determination data is moved；
Step 408：The instruction of data-moving is sent to DMA unit.
The application scenarios of the embodiment of data-moving device and method one of the present invention are as shown in figure 9, it indicates this implementation
In example, DMA data is moved in the location of LTE down physical layers.
LTE down physical layers complete coding, rate-matched 501, modulation 502, and the conversion (FFT of frequency domain to time domain
Conversion) OFDM (OFDM) symbol can be formed after 503.Wherein, coding, speed, 501 by hardware handles, modulation 502
With FFT 503 by software processing.In order to improve processing speed, convert front and rear data and be all placed on level cache data in advance
Memory (L1D), CPU software write L1D in processing data, move 504 OFDM symbols by DMA and are moved from L1D to two level
Cache memory data (L2D), is finally sent to radio frequency 506 by antenna processing 505.Wherein, DMA move 504, antenna processing 505,
And radio frequency 506 uses hardware handles.DMA move 504 be with DMA administrative units according to being pre-configured with startup DMA unit, it is complete
Function is moved into data.
For LTE physical layer process using symbol as chronomere, a symbol lengths are 1/14ms.Traditional DMA moves mechanism,
Software will start DMA 14 times in 1ms.Start DMA unit without software using after the present invention, but started by DMA administrative units
DMA unit, 14 software start-up courses in 1ms are eliminated well.
Structure chart of the data-moving device of the present invention under Fig. 9 application scenarios is as shown in Figure 10, and it indicates this implementation
In example, the structure of data-moving device.
The data-moving device of the present embodiment includes：DMA unit 102, for by the data including OFDM symbol from
L1D moves L2D；DMA administrative units 101, for sending data-moving instruction to DMA unit 102.
DMA administrative units 101 include following two modules, are respectively：Caching management module 103, for monitoring processor
The process of data is write toward L1D；Queue management module 104, after the buffer status for being sent in order caching management module indicates,
The task of the data including OFDM symbol is moved according to the distribution of the current state of DMA unit 102.
Caching management module 103 includes following three submodules, is respectively：Processor monitors submodule 105, for monitoring
Processor writes the progress of data toward L1D；Threshold values contrast submodule 106, for judge L1D include OFDM symbol including data
Whether the threshold values moved is reached；State instruction sending submodule 107, refer to for sending buffer status to queue management module 104
Show.
Queue management module 104 includes following five submodules, is respectively：State instruction receiving submodule 108, is used for
Receive the buffer status instruction sent at state instruction sending submodule 107；Descriptor takes out submodule 109, for taking out
Initialized task descriptor in queue management module 104；Task splits submodule 1011, for the data for moving needs
Amount is split；Arbitrate submodule 1012, for the queue priority according to queue management module 104, what determination data was moved
Sequencing；Task sending submodule 1010, for sending the instruction of data-moving to DMA unit 102.
Flow chart of the data-moving method of the present invention under Fig. 9 application scenarios is as shown in figure 11, and it is indicated Fig. 9's
In scene, the idiographic flow of the present embodiment data-moving method, meanwhile, it can become apparent from flow with reference to Figure 10.
Data-moving idiographic flow in the present embodiment is divided into the following steps：
Step 601：DMA administrative units 101 initialize.
The initialization of DMA administrative units 101 is initialized by CPU, as long as and initializing once.In CPU
Carry out after once initializing, the follow-up operation of moving of DMA unit 102 fully relies on hardware completion, it is not necessary to the participation of software.Just
During beginningization, a room is divided in L1D, for caching FFT result.In caching management module 103, cache management is set
The threshold value of module 103.In queue management module 104, initialization task descriptor, including set and move source address, move mesh
Address, moving data size and the number of repetition moved.Also, the task descriptor initialized is write into queue
In management module 104.It is also desirable in queue management module 104, setting is moved end every time and need not produced in completion
It is disconnected, set queue priority (to be eated dishes without rice or wine because the result after FFT will be regularly sent to, requirement of real-time is high, therefore moves FFT
The queue priority of transformation results will be set as highest).
Step 602：CPU software carries out FFT computings, writes the result into L1D.
Step 603：Processor in caching management module 103 monitors submodule 105 and monitors CPU toward writing data in L1D
Process.
Step 604：Whether the data volume that the threshold values contrast submodule 106 in caching management module 103 enters to CPU write meets
Threshold values judges.The threshold values set when meeting initialization in this way then carries out next step flow, and such as otherwise returning to step 603 continues
Monitor.
Step 605：Buffer status instruction is sent to by the state instruction sending submodule 107 in caching management module 103
Queue management module 104.
Step 606：State instruction receiving submodule 108 in queue management module 104 is receiving buffer status instruction
Afterwards, descriptor takes out submodule 109 and takes out initialized good task descriptor in step 601, and task splits submodule
1011 are split the data volume for needing to move, and then, arbitration submodule 1012 will be set according to the priority in step 601
Put, the sequencing that determination data is moved, the instruction of data-moving is sent to DMA unit by last task sending submodule 1010
102。
Step 607：Judge whether DMA unit 102 completes the task of other moving datas, complete other in this way and move
The task of data, then carry out next step flow, and such as no step 607 that returns to continues to judge.
Step 608：After receiving data-moving instruction, FFT operation results are moved to L2D from L1D.
Step 609：After the completion of moving, the flow that step 602 continues subsequent cycle is returned to.
Compared with prior art, the beneficial effect of the present embodiment has at following 5 points：
1. unit interval inner treater can perform more instructions.Because processor tasks are shunted, need in the past in
The instruction execution flow of disconnected processor, then go to notify the operation of DMA moving datas, no longer needed after the present invention, so as to be
Performance of uniting is more excellent, for wireless base station device, possesses the possibility of access more users.
2. the design of Installed System Memory allocative decision is more simple.Multilevel memory often be present in current high performance chip system,
Level cache (L1) memory access nearest from processor is fastest, but capacity is also minimum, next to that L2 cache (L2), is finally
Double Data Rate synchronous DRAM (DDR), DDR capacity is big but memory access speed is most slow.Conventional Installed System Memory distribution is often
Need long period thinking and optimize repeatedly, which data is put into L1, which data puts DDR.There are the present apparatus and configuration side
After method, L1 and L2/DDR data interaction has reformed into transparent for software, and data are placed on DDR during system electrification, when
Program is performed when needing to use the data, then by DMA hardware subsystem automatically data exchange to L1, and existing Large Copacity is also protected
The memory access speed of processor is demonstrate,proved.
3. the unified platform, reduce error probability.Hardware logic is once solidify, and its behavior and operating result also just solidify, base
Originally will not malfunction；Perform once, provided by collocation method by platform software, no during corresponding combined arrangement method system initialization
Same business scenario its collocation method unification, business software programming personnel need not pay close attention to the process, also be equal to solidification substantially.Such as
Fruit departs from the device and method, using conventional method, by software notification DMA moving datas, joins because software programmers are horizontal
It is secondary uneven, easily there are various mistakes.
4. adding hardware cost is very low.It is a small functional module on newly-increased DMA administrative unit hardware, design is not multiple
Miscellaneous, small power consumption, after chip batch processing and production, its cost is minimum after making thinner, thus triggers DMA moving data phases with CPU
Than its cost is cheaper.
5. the space that program takes is less.Do not need software triggering DMA to move, avoid the need for realizing the generation of this part yet
Code, program scale are slightly reduced, and the system being had too many difficulties to cope with to memory headroom has certain benefit.
The preferred embodiments of the present invention are the foregoing is only, are not intended to limit the scope of the invention, it is every at this
Under the inventive concept of invention, the equivalent structure transformation made using description of the invention and accompanying drawing content, or directly/use indirectly
It is included in other related technical areas in the scope of patent protection of the present invention.
Claims (11)
- A kind of 1. data-moving device, it is characterised in that including：Direct memory access DMA administrative units, for sending data-moving instruction to DMA unit；The DMA unit, for performing the data-moving instruction.
- 2. data-moving device as claimed in claim 1, it is characterised in that the DMA administrative units include：Caching management module, the process of data is write toward physical memory for monitoring processor, and send and delay to queue management module Deposit state instruction.The queue management module, task is moved according to DMA unit current state distribution.
- 3. data-moving device as claimed in claim 2, it is characterised in that the caching management module includes：Processor monitors submodule, and the progress of data is write for monitoring processor toward the physical memory；Threshold values contrasts submodule, and whether the data volume for judging in the physical memory reaches the threshold values moved；State instruction sending submodule, indicated for sending the buffer status to the queue management module.
- 4. data-moving device as claimed in claim 3, it is characterised in that the queue management module includes：State instruction receiving submodule, refer to for receiving the buffer status sent at the state instruction sending submodule Show；Descriptor takes out submodule, for taking out initialized task descriptor in the queue management module；Task sending submodule, for sending the instruction of the data-moving to the DMA unit.
- 5. data-moving device as claimed in claim 4, it is characterised in that the queue management module also includes：Task splits submodule, for the data volume for needing to move to be split；Arbitrate submodule, according to the queue priority of the queue management module, sequencing that determination data is moved.
- A kind of 6. data-moving method, it is characterised in that including：Direct memory access DMA administrative units send data-moving and referred to Order performs the data-moving by the DMA unit and instructed to DMA unit.
- 7. data-moving method as claimed in claim 6, it is characterised in that the DMA administrative units send data-moving and referred to Before order to DMA unit, in addition to：Initialize the DMA administrative units.
- 8. data-moving method as claimed in claim 7, it is characterised in that the initialization DMA administrative units include：Physical memory is distributed to the queue management module of the DMA administrative units；The threshold values of the caching management module of the DMA administrative units is set；The queue priority of the queue management module is set；Initialize the task descriptor of the queue management module；The task descriptor is write in the queue management module.
- 9. data-moving method as claimed in claim 8, it is characterised in that the DMA administrative units send data-moving and referred to Make to DMA unit, including：Monitoring processor writes the process of data toward physical memory, and sends buffer status instruction；Order caching state instruction, and being moved according to DMA unit current state distribution for task.
- 10. data-moving method as claimed in claim 9, it is characterised in that the monitoring processor writes number toward physical memory According to process, and send buffer status instruction, including：Monitor the progress that processor writes data toward the physical memory；Judge whether the data volume in the physical memory reaches the threshold values moved；If reaching threshold values, buffer status instruction is sent；If not up to threshold values, continue to monitor processor toward the physical memory Write the progress of data.
- 11. data-moving method as claimed in claim 10, it is characterised in that the order caching state instruction, and according to The task that the DMA unit current state distribution is moved, including：Receive the buffer status instruction；Take out the task descriptor；Split the data volume for needing to move；According to the queue priority, sequencing that determination data is moved；The instruction of data-moving is sent to the DMA unit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610505894.9A CN107562659A (en) | 2016-06-30 | 2016-06-30 | A kind of data-moving device and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610505894.9A CN107562659A (en) | 2016-06-30 | 2016-06-30 | A kind of data-moving device and method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107562659A true CN107562659A (en) | 2018-01-09 |
Family
ID=60969405
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610505894.9A Pending CN107562659A (en) | 2016-06-30 | 2016-06-30 | A kind of data-moving device and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107562659A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109857686A (en) * | 2019-03-26 | 2019-06-07 | 北京简约纳电子有限公司 | A kind of method that DMA data synchronous transfer becomes asynchronous transmission |
| CN110532205A (en) * | 2019-07-17 | 2019-12-03 | 浙江大华技术股份有限公司 | Data transmission method, device, computer equipment and computer readable storage medium |
| CN111190842A (en) * | 2019-12-30 | 2020-05-22 | Oppo广东移动通信有限公司 | Direct memory access, processor, electronic device, and data transfer method |
| WO2020143237A1 (en) * | 2019-01-08 | 2020-07-16 | 郑州云海信息技术有限公司 | Dma controller and heterogeneous acceleration system |
| CN111615692A (en) * | 2019-05-23 | 2020-09-01 | 深圳市大疆创新科技有限公司 | Data transfer method, calculation processing device, and storage medium |
| CN112506437A (en) * | 2020-12-10 | 2021-03-16 | 上海阵量智能科技有限公司 | Chip, data moving method and electronic equipment |
| CN114399034A (en) * | 2021-12-30 | 2022-04-26 | 北京奕斯伟计算技术有限公司 | Data handling method for direct memory access device |
| WO2022193108A1 (en) * | 2021-03-16 | 2022-09-22 | 华为技术有限公司 | Integrated chip and data transfer method |
-
2016
- 2016-06-30 CN CN201610505894.9A patent/CN107562659A/en active Pending
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020143237A1 (en) * | 2019-01-08 | 2020-07-16 | 郑州云海信息技术有限公司 | Dma controller and heterogeneous acceleration system |
| CN109857686A (en) * | 2019-03-26 | 2019-06-07 | 北京简约纳电子有限公司 | A kind of method that DMA data synchronous transfer becomes asynchronous transmission |
| CN109857686B (en) * | 2019-03-26 | 2020-12-29 | 北京简约纳电子有限公司 | Method for converting synchronous transmission of DMA data into asynchronous transmission |
| CN111615692A (en) * | 2019-05-23 | 2020-09-01 | 深圳市大疆创新科技有限公司 | Data transfer method, calculation processing device, and storage medium |
| CN110532205A (en) * | 2019-07-17 | 2019-12-03 | 浙江大华技术股份有限公司 | Data transmission method, device, computer equipment and computer readable storage medium |
| CN111190842A (en) * | 2019-12-30 | 2020-05-22 | Oppo广东移动通信有限公司 | Direct memory access, processor, electronic device, and data transfer method |
| CN111190842B (en) * | 2019-12-30 | 2021-07-20 | Oppo广东移动通信有限公司 | Direct memory access, processor, electronic device, and data transfer method |
| CN112506437A (en) * | 2020-12-10 | 2021-03-16 | 上海阵量智能科技有限公司 | Chip, data moving method and electronic equipment |
| WO2022193108A1 (en) * | 2021-03-16 | 2022-09-22 | 华为技术有限公司 | Integrated chip and data transfer method |
| CN114399034A (en) * | 2021-12-30 | 2022-04-26 | 北京奕斯伟计算技术有限公司 | Data handling method for direct memory access device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107562659A (en) | A kind of data-moving device and method | |
| CN106844017A (en) | The method and apparatus that event is processed for Website server | |
| CN103377157B (en) | A kind of double-core data communications method for built-in digital control system | |
| CN103473141A (en) | Method for out-of-band check and modification of BIOS (basic input/output system) setting options | |
| US20180129939A1 (en) | Method and system of managing computing paths in an artificial neural network | |
| CN104657308A (en) | Method for realizing server hardware acceleration by using FPGA | |
| CN104714785A (en) | Task scheduling device, task scheduling method and data parallel processing device | |
| CN109074281B (en) | Method and device for distributing graphics processor tasks | |
| CN102253860A (en) | Asynchronous operation method and asynchronous operation management device | |
| US20210004685A1 (en) | Techniques to manage training or trained models for deep learning applications | |
| CN106775491B (en) | Data processing method and storage equipment | |
| WO2021135574A1 (en) | Data storage method and apparatus, and terminal device | |
| US20200351155A1 (en) | Base station configuration method and apparatus | |
| CN108089920A (en) | A kind of methods, devices and systems of data processing | |
| TWI566087B (en) | System and method for controlling cpu | |
| CN104133671B (en) | The hardware interface screening arrangement and method of a kind of digital processing device | |
| CN111858440A (en) | Distributed data parallel training system, method, terminal and storage medium | |
| CN111294377B (en) | Dependency network request sending method, terminal device and storage medium | |
| CN104850516A (en) | DDR frequency conversion design method and device | |
| US20130238871A1 (en) | Data processing method and apparatus, pci-e bus system, and server | |
| CN110535759A (en) | A kind of communication gate, correlation technique and relevant apparatus | |
| CN105677606B (en) | The hot-plug method and bus unit of bus apparatus | |
| CN102542525B (en) | Information processing equipment and information processing method | |
| CN103986588B (en) | Computer system and far-end remote control method of computer device | |
| CN109189705B (en) | USB extension method, device, equipment, storage medium and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180109 |