CN104952096A - CPU and GPU hybrid cloud rendering method, device and system - Google Patents
CPU and GPU hybrid cloud rendering method, device and system Download PDFInfo
- Publication number
- CN104952096A CN104952096A CN201410125332.2A CN201410125332A CN104952096A CN 104952096 A CN104952096 A CN 104952096A CN 201410125332 A CN201410125332 A CN 201410125332A CN 104952096 A CN104952096 A CN 104952096A
- Authority
- CN
- China
- Prior art keywords
- cpu
- gpu
- rendering
- rendering task
- primitive frame
- 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.)
- Granted
Links
Landscapes
- Processing Or Creating Images (AREA)
Abstract
The invention discloses a CPU and GPU hybrid cloud rendering method, device and system, and relates to the technical field of cloud computing. According to the method, a rendering task is decomposed into logic continuous type rendering tasks and parallel rendering tasks to be respectively transmitted to a CPU and a GPU to be processed, and original frame graphs are generated by linkage output rendering of the CPU and GPU rendering results; and the ordered original frame graphs are segmented and then allocated to coding channels in a unified coding processing resource pool in turn to perform segmented fluidization coding, and the coding results outputted by all the coding channels are integrated to form a media stream. A mode of CPU and GPU rendering capability integration, combination and focusing of labor division is adopted by the method, the device and the system, and the integrated solution scheme from picture rendering to fluidization is realized via unified segmented fluidization processing through the CPU and the GPU unified coding processing capacity resource pool. Compared with a single CPU and GPU rendering method, service quality and service efficiency can be effectively enhanced by the scheme.
Description
Technical field
The present invention relates to field of cloud computer technology, particularly a kind of CPU and GPU mixed cloud rendering intent, device and system.
Background technology
It is a kind of based on high in the clouds remote rendering service for playing up cloud, allows user with Web interface, entirely independently completes various rendering task.Playing up (Rendering) refers in computer graphics, 2D or 3D model is transformed to the process of image.In graphics pipeline, playing up is last important step, obtains model and the final display effect of animation by it.
The render time of every width picture, different according to model complexity, differ greatly, several seconds at least, several days at most; Model is more complicated, and render time is longer.For film " A Fanda ", an average frame (one deck, the 2K ~ 4K resolution) time played up by unit is 2 hours; The film of 160 minutes, 9000 width pictures per minute (60 seconds × 30 frame × 5 layer), overall render time needs 2880000 hours, is equivalent to 328 years.Add 3D three-dimensional film effect, every frame picture has two images, plays up and needs just can complete more than 600 years.
Just because of above-mentioned reason, rendering task is placed high in the clouds, calls hundreds and thousands of machines simultaneously and carry out picture and play up, the integral operation time can present the decline of the geometry order of magnitude, originally the unit rendering task that needs the centuries to complete, even can complete less than 1 year for several days.
Simultaneously, various large-scale Rendering software Blender, Maya and 3Ds MAX etc. permit that (License) price is high, and running environment requires high, be difficult to bear for small-sized animation, gaming enterprise, cloud is played up in use can solve cost of development well, is conducive to industry development and Related product exploitation.
The key playing up cloud is the treatment technology in high in the clouds, and great majority adopt CPU serial frame to carry out rendering operation, such as Ali's cloud, Shanghai net wash with watercolours, Renderbus etc. on the market at present.Because of CPU serial frame, the logic that employing process processes one by one is applicable for the static type picture possessing precedence; But for extensive continuity such as video flowing, possess the graphical application etc. of 3 dimension module, it does not possess efficient processing power.
Adopt parallel architecture relative to CPU, GPU, there is more computing unit and internal memory bit wide, make it in floating-point operation, parallel computation etc., have more outstanding advantage.Its rendering capability aspect, for continuity Real-time graphic processing, such as Film Animation etc. have the advantage that CPU hardly matches.Current Amazon has been proposed the superpower graphics calculations example of cloud computing service-Amazon cluster based on GPU, this example provides the GPU processing power in cloud, comprise the memory capacity of 22GB, 33.5 EC2 computing units provide the instant access of the GPU calculated performance of hight coordinate for developer and enterprise.Domesticly praise strange science and technology, construct similar framework too, utilize GPU advantage, the online real-time rendering of process three-dimensional model.
But, the service of playing up of current cloud is all provide service based on single CPU or GPU framework, for needing tableaux to play up large-scale CG, the Film Animation of playing up demand with real-time three-dimensional model online simultaneously, can not meet very well.
Summary of the invention
The present inventor finds to have problems in above-mentioned prior art, and therefore proposes a kind of new technical scheme at least one problem in described problem.
An object of the present invention is to provide a kind of technical scheme played up for cloud.
According to a first aspect of the invention, provide a kind of CPU and GPU mixed cloud rendering intent, comprising:
Rendering task is decomposed into logic continuous type rendering task and parallel rendering task, logic continuous type rendering task is sent to CPU serial processing, send to GPU to carry out concurrent operation process parallel rendering task, and export by the interlock of CPU and GPU rendering result the primitive frame figure playing up generation;
The primitive frame figure playing up generation integrated according to fluidisation and sorts, exporting the primitive frame figure of sequence;
Carry out segmentation to the primitive frame figure of sequence, in order primitive frame figure segment assignments is carried out sectional type fluidisation coding to the coding pass in Unified coding treatment source pool, wherein, Unified coding treatment source pool is built by CPU, GPU code capacity and forms;
The coding result exported by each coding pass carries out sorting and integrates and forms Media Stream.
Alternatively, the method also comprises: in the Media Stream completing fluidization treatment, increase personage or cartoon character.
Alternatively, export by the interlock of CPU and GPU rendering result the primitive frame figure playing up generation to comprise:
Before CPU and GPU carries out rendering task, CPU in resource coordination and hardware assigning process for rendering task carries out order registration, distribute according to relevant order in render process and play up, in Output rusults integration process, graphic result is integrated by corresponding corresponding order numbering.
Alternatively, CPU serial processing is sent to comprise logic continuous type rendering task: CPU employing multinuclear simultaneously treatment mechanism completes serial processing according to priority logic to rendering task, and exports relevant rendering result;
Parallel rendering task sent to GPU to carry out concurrent operation process to comprise: adopt the local internal memory of GPU and bus memory mutual, and rely on GPU to share delivery of content respectively to process core to GPU and carry out parallel processing and calculating to rendering task, and final output plays up result accordingly.
Alternatively, logic continuous type rendering task comprises that actor model is played up, action is played up; Parallel rendering task comprises large scene and plays up.
According to a further aspect in the invention, a kind of CPU and GPU mixed cloud rendering device is provided, comprises:
Interlock hybrid rending module, for rendering task being decomposed into logic continuous type rendering task and parallel rendering task, logic continuous type rendering task is sent to CPU serial processing, parallel rendering task is sent to GPU concurrent operation process, and export by the real-time linkage of CPU and GPU rendering result the primitive frame figure playing up generation, the primitive frame figure playing up generation integrated according to fluidisation and sorts, exporting the primitive frame figure of sequence;
Dynamic fluidized processing module, for the primitive frame figure of the sequence from interlock hybrid rending module is carried out segmentation, in order primitive frame figure segment assignments is carried out sectional type fluidisation coding to the coding pass in Unified coding treatment source pool, wherein, Unified coding treatment source pool is built by CPU, GPU code capacity and forms; The coding result exported by each coding pass carries out sorting and integrates and forms Media Stream.
Alternatively, this device also comprises: character embedded module, for increasing personage or cartoon character in the Media Stream completing fluidization treatment.
Alternatively, the hybrid rending module that links comprises:
Rendering task classification scheduling unit, for rendering task is decomposed into logic continuous type rendering task and parallel rendering task by rendering task, sends to CPU serial processing by logic continuous type rendering task, parallel rendering task is sent to GPU concurrent operation process;
The former figure render processing unit of frame, plays up work accordingly for completing based on the arithmetic element of CPU and GPU, and CPU focuses on single serial macrooperation amount rendering task; GPU focuses on the macrooperation amount rendering task of parallel synchronous; Play up rear respective output primitive frame figure;
Graphic result integrates sequencing unit, for the primitive frame figure exported by CPU and GPU hybrid rending, exports the primitive frame figure of sequence according to fluidisation sequence integration and sequence.
Alternatively, dynamic fluidized processing module comprises:
Primitive frame figure segmenting unit, for resolving into each segmentation in order by the primitive frame figure playing up output;
Segmentation fluidization treatment unit, unify video coded treatment resource pool is built for utilizing CPU, GPU fluidisation ability, the coded treatment passage that the primitive frame figure of different fragments correspondence distributes to unified resource pond is carried out fluidization treatment, and after having processed, coded treatment passage exports segmentation Media Stream separately;
Fluidisation TB control unit, for integrating the segmentation fluidization treatment result that each coded treatment passage exports, carries out integrating the final Media Stream of formation.
Alternatively, logic continuous type rendering task comprises that actor model is played up, action is played up; Parallel rendering task comprises large scene and plays up.
According to another aspect of the invention, provide a kind of cloud rendering system, comprise above-mentioned CPU and GPU mixed cloud rendering device.
An advantage of the present invention is, CPU and GPU rendering capability is adopted to integrate, associating has again the pattern stressing to divide the work, segmentation fluidization treatment is unified by CPU, GPU Unified coding processing power resource pool, realize the integrative solution that picture is rendered into fluidisation, can effectively promote service quality and service efficiency.
By referring to the detailed description of accompanying drawing to exemplary embodiment of the present invention, further feature of the present invention and advantage thereof will become clear.
Accompanying drawing explanation
What form a part for instructions drawings describes embodiments of the invention, and together with the description for explaining principle of the present invention.
With reference to accompanying drawing, according to detailed description below, clearly the present invention can be understood, wherein:
Fig. 1 illustrates the process flow diagram of an embodiment according to CPU and GPU mixed cloud rendering intent of the present invention;
Fig. 2 illustrates the overall operation frame diagram of an embodiment according to CPU and GPU mixed cloud rendering system of the present invention;
Fig. 3 is the process flow diagram of another embodiment illustrated according to CPU and GPU mixed cloud rendering intent of the present invention;
Fig. 4 illustrates CPU, GPU hybrid rending procedure chart according to an embodiment of the invention;
Fig. 5 illustrates CPU, GPU interlock fluid mapper process figure according to an embodiment of the invention;
Fig. 6 illustrates the structural drawing of an embodiment according to CPU and GPU mixed cloud rendering device of the present invention;
Fig. 7 illustrates the structural drawing of another embodiment according to CPU and GPU mixed cloud rendering device of the present invention.
Embodiment
Various exemplary embodiment of the present invention is described in detail now with reference to accompanying drawing.It should be noted that: unless specifically stated otherwise, otherwise positioned opposite, the numerical expression of the parts of setting forth in these embodiments and step and numerical value do not limit the scope of the invention.
Meanwhile, it should be understood that for convenience of description, the size of the various piece shown in accompanying drawing is not draw according to the proportionate relationship of reality.
Illustrative to the description only actually of at least one exemplary embodiment below, never as any restriction to the present invention and application or use.
May not discuss in detail for the known technology of person of ordinary skill in the relevant, method and apparatus, but in the appropriate case, described technology, method and apparatus should be regarded as a part of authorizing instructions.
In all examples with discussing shown here, any occurrence should be construed as merely exemplary, instead of as restriction.Therefore, other example of exemplary embodiment can have different values.
It should be noted that: represent similar terms in similar label and letter accompanying drawing below, therefore, once be defined in an a certain Xiang Yi accompanying drawing, then do not need to be further discussed it in accompanying drawing subsequently.
GPU computing unit is many, for playing up to calculate and process, there is outstanding ability equally, but, owing to more easily disposing based on the solution of CPU architecture and mainboard inbuilt display, monokaryon processing power is stronger, and in actual application, the cloud rendering solution of CPU architecture becomes main flow.
In order to better meet CG and large batch of Film Animation, mass data play up demand, the present inventor notices: need CPU serial and GPU concurrent operation ability unified planning and scheduling, realize GPU, CPU mixing interlock to play up, just can accomplish that hardware resource Optimized Operation, graph data efficiently process and real-time online is played up, the seamless connection of fluidisation.
Fig. 1 illustrates the process flow diagram of an embodiment according to CPU and GPU mixed cloud rendering intent of the present invention.
As shown in Figure 1, step 101, rendering task is decomposed into logic continuous type rendering task and parallel rendering task, logic continuous type rendering task is sent to CPU serial processing, send to GPU to carry out concurrent operation process parallel rendering task, and export by the real-time linkage of CPU and GPU rendering result the primitive frame figure playing up generation.
Step 102, to integrate according to fluidisation the primitive frame figure playing up generation and sorts, exporting the primitive frame figure of sequence.
Step 103, carries out segmentation to the primitive frame figure of sequence, in order primitive frame figure segment assignments is carried out sectional type fluidisation coding to the coding pass in Unified coding treatment source pool, and wherein, Unified coding treatment source pool is built by CPU, GPU code capacity and forms.
Step 104, the coding result exported by each coding pass carries out sorting and integrates and forms Media Stream.
Fig. 2 illustrates the overall operation frame diagram of an embodiment according to CPU and GPU mixed cloud rendering system of the present invention.Below main composition graphs 2 is explained and how to realize comprising CPU, GPU hardware integration capability and basic Rendering software ability on bottom software and hardware basis, realize CPU, GPU interlock hybrid rending function.
Wherein, bottom software and hardware resources is the basis realizing interlock hybrid rending, and hardware resource mainly comprises CPU and GPU etc., and software refers to provide basis to play up function tool software, such as, comprise Maya, 3DS Max etc.
1) link hybrid rending management.Mainly comprise and build the distribution of bottom hardware (CPU, GPU) resource and competence, task scheduling function, realize making full use of hardware capabilities, realize CPU, GPU hybrid rending function.This part can be further subdivided into: Task-decomposing and scheduling of resource, graphic result integration and sequence, fluidisation result merge and export.Task-decomposing and scheduling of resource, decompose rendering task, and comprehensively transfer CPU, GPU resource, jointly taken and complete relevant rendering task by CPU, GPU resource; Graphic result is integrated and sequence, completes result, integrates and sort to primitive frame figure, for follow-up fluidization treatment lays the foundation for graphically playing up process; Fluidisation result merges and exports, and refers to that the Media Stream fragment exported of CPU, GPU being encoded respectively combines, and editor exports the animation that needs or CG etc.
2) processing section is played up in interlock.Mainly comprise: primitive frame figure plays up, dynamic fluidized process and embedded three parts of character.It is the core of CPU, GPU hybrid rending that primitive frame figure plays up, main finger is by various raw data, such as person model, pixel and scene original painting etc., totally dispatched by CPU, apply in a flexible way CPU serial and GPU walks abreast advantage, the rendering task that corresponding distribution is relevant or step, realize the hybrid rending merging CPU, GPU ability, final output can be directly used in the primitive frame figure of fluidization treatment.Dynamic fluidized process, refers to according to designing requirement, the primitive frame figure playing up acquisition is converted into the process of animation or CG fluidisation.Adopt segmentation fluidization treatment mechanism in this process, give CPU and GPU respectively by fluidisation task, the two video encoding capability comprehensive, relatively single CPU or GPU pattern, more fast can complete fluid mapper process.Character is embedded, refers in the Media Stream completing fluidization treatment, increases corresponding personage or cartoon character, finally complete the manufacturing process of film CG or animation according to story of a play or opera design.
GPU+CPU mixed cloud rendering intent in above-described embodiment, break through single CPU, GPU rendering efficiency is low, be difficult to the task that satisfied compatible Static and dynamic in enormous quantities, extensive is played up, such as 3D film, CG(Computer Graphics, computer graphical) etc., adopt CPU and GPU rendering capability to integrate, associating has again the pattern stressing to divide the work, the macrooperation amount serial task that the process of CPU emphasis is simple; The process of GPU emphasis walks abreast macrooperation task, and finally, by CPU, GPU Unified coding processing power resource pool, unified segmentation fluidization treatment, realizes the integrative solution that picture is rendered into fluidisation.Relatively single CPU, GPU rendering intent, the program can effectively promote service quality and service efficiency.
Fig. 3 is the process flow diagram of another embodiment illustrated according to CPU and GPU mixed cloud rendering intent of the present invention.
As shown in Figure 3, step 301, high in the clouds rendering server receives the rendering server request of user, and response user plays up services request.Playing up services request is such as remote rendering services request from long-distance user.
Step 302, plays up services request in response to user, and high in the clouds rendering server carries out playing up hardware environment monitoring, judge that hardware environment condition meets service request? if so, continue step 303, otherwise return service cannot provide (step 310).
Step 303, high in the clouds rendering server configure hardware environment, calls bottom hardware resource, starts Rendering software, builds and plays up hardware environment.
Step 304, user prepares and original plays up material, arranges and plays up demand.
Step 305, comprehensive CPU, GPU ability of high in the clouds rendering server, mixing is called CPU, GPU and is carried out primitive frame figure and merge and play up, and exports rendering result primitive frame figure.In one embodiment, high in the clouds arranges requirement according to playing up, and unified co-allocation CPU, GPU rendering task, by resource consolidations such as master pattern, scene and pixels, mainly bear serial, parallel rendering task respectively by CPU, GPU, associating carries out hybrid playing up together; To play up and export various picture, and according to fluidisation frame number order, export respectively and preserve.
Step 306, to play up the primitive frame figure of generation for material, calls CPU, GPU and carries out segment encoding fluidisation.By the primitive frame figure played up, call CPU, GPU hybrid coding resource pool, adopt segmentation fluidization treatment.
Step 307, integrates each segment encoding stream and gathers, and forms standard cartoon medium stream and exports.
Step 308, in standard media stream, increases corresponding personage/cartoon character image, completes Film Animation or CG result.In large-scale production of film and TV or game, after corresponding scene first completes by needs, just personage or cartoon character are embedded in animated video stream, form final film effect.
The disclosure is based on GPU, CPU mixed architecture of innovation, and comprehensive GPU, CPU processing power, builds and be applicable to playing up from tableaux, exports and the integrated solution of CPU, GPU hybrid rending of fluidization treatment to real time 3-D image.
Fig. 4 illustrates CPU, GPU hybrid rending procedure chart according to an embodiment of the invention.As shown in Figure 4, in the present embodiment, CPU, GPU hybrid rending that links comprises: scheduling of resource and hardware coordination, primitive frame figure play up and process and output, graphic result are integrated and sequence etc.Wherein make full use of CPU serial advantage, first carry out scheduling of resource and hardware coordination by CPU.Then, CPU, GPU, according to coordinating of task, carry out primitive frame figure respectively and play up process and export.Wherein CPU utilizes bus memory to call multinuclear ability and realizes linear logic process process; GPU utilizes shared video memory and bus memory alternately, and each process core is by the parallel processing of GPU Sharing Memory Realization; Finally by cpu logic computing and GPU graph rendering, give CPU to carry out integrating, sorting by unified for graphic result, realize the overall logicization series connection of rendering result, totally complete CPU, GPU interlock hybrid rending link.Specifically comprise:
(1) scheduling of resource and hardware coordination
Make full use of CPU operational efficiency and core processing ability, play up by scheduling of resource before starting, namely rendering task distribution is dominated by it, for original render picture and task arrangement.By bus memory, carry out related data storage and transmission, will need serialization calculation process, such as actor model is played up, action is played up etc. distributes to CPU; To need parallelization calculation process, such as large scene is played up, distribute to GPU.
(2) primitive frame figure plays up process and exports
According to unified task matching, CPU because of its process core relatively less, but the feature that single arithmetic element processing power is strong, emphasis process individual task needs the work such as a large amount of computing such as actor model is played up, action/special efficacy is played up, adopt multinuclear treatment mechanism simultaneously, according to priority logic to having played up serial processing, and export relevant rendering result; GPU is many according to its process core, the feature of the far super CPU of comprehensive arithmetic capability, emphasis processing example is as large scene, need the situation elements calculating such as a large amount of pixel, object that walked abreast, adopt the local internal memory of GPU and bus memory mutual, and rely on GPU to share delivery of content respectively to process core to GPU and carry out parallel processing and calculating to rendering task, and final output plays up result accordingly.In processing procedure, CPU with GPU handling principle is similar, is all to complete based on respective arithmetic element to play up work accordingly, and it is few that difference is just that CPU calculates core, but core computing power is strong; It is many that GPU calculates core, but core ability is more weak, and comprehensive overlay capability is superpower.Just occurred that CPU focuses on a large amount of operand task of single serial based on above-mentioned basis just, GPU focuses on the macrooperation amount task of parallel synchronous.In actual mechanical process, also according to CPU, GPU idle degrees each other, respective task can be distributed in order, and need not be confined to mentioned above principle.
(3) graphic result is integrated, is sorted
Before CPU and GPU plays up, in resource coordination and hardware assigning process, for rendering task carries out order registration, distribute according to relevant order in render process and play up, in final Output rusults integration process, graphic result carries out integrating and sorts by corresponding corresponding order numbering, completes and exports according to fluidisation order primitive frame figure.
Fig. 5 illustrates CPU, GPU interlock fluid mapper process figure according to an embodiment of the invention.As shown in Figure 5, in the present embodiment, CPU, GPU interlock fluidization treatment comprises: primitive frame figure segmentation, segmentation fluidization treatment, fluidisation TB control etc.
(1) primitive frame figure segmentation
Utilize cpu logic processing power, the primitive frame figure playing up output is resolved into different section in order, unified CPU, GPU merging becomes unified fluidization treatment resource pool, and overall scheduling task by different fragments mixed allocation to the Video coding treatment channel in CPU, GPU unified resource pond.
(2) segmentation fluidization treatment
Corresponding CPU, GPU hybrid processing resource pool, the primitive frame figure of each processing unit alignment processing segmentation separately, carries out segmentation fluidization treatment.
(3) fluidisation TB control
The result that segmentation fluidization treatment completes is integrated, forms final Media Stream result.
Complete as above after processing procedure, corresponding personage or cartoon character can be added, become 3D film or CG result.
Fig. 6 illustrates the structural drawing of an embodiment according to CPU and GPU mixed cloud rendering device of the present invention.As shown in Figure 6, this device comprises: interlock hybrid rending module 61, for rendering task being decomposed into logic continuous type rendering task and parallel rendering task, logic continuous type rendering task is sent to CPU serial processing, parallel rendering task is sent to GPU concurrent operation process, and the primitive frame figure playing up generation is exported by the real-time linkage of CPU and GPU rendering result, the primitive frame figure playing up generation integrated according to fluidisation and sorts, exporting the primitive frame figure of sequence; Dynamic fluidized processing module 62, for the primitive frame figure of the sequence from interlock hybrid rending module is carried out segmentation, in order primitive frame figure segment assignments is carried out sectional type fluidisation coding to the coding pass in Unified coding treatment source pool, wherein, Unified coding treatment source pool is built by CPU, GPU code capacity and forms; The coding result exported by each coding pass carries out sorting and integrates and forms Media Stream.Logic continuous type rendering task such as comprises that actor model is played up, action is played up; Parallel rendering task such as comprises large scene and plays up.
Fig. 7 illustrates the structural drawing of another embodiment according to CPU and GPU mixed cloud rendering device of the present invention.As shown in Figure 7, in one embodiment, cloud rendering device comprises: interlock hybrid rending module 71, dynamic fluidized processing module 72 and character embedded module 73.Wherein, character embedded module 73 for increasing personage or cartoon character in the Media Stream completing fluidization treatment.
In one embodiment; interlock hybrid rending module 71 comprises: rendering task classification scheduling unit 711; for rendering task is decomposed into logic continuous type rendering task and parallel rendering task by rendering task; logic continuous type rendering task is sent to CPU serial processing, parallel rendering task is sent to GPU concurrent operation process; The former figure render processing unit 712 of frame, plays up work accordingly for completing based on the arithmetic element of CPU and GPU, and CPU focuses on single serial macrooperation amount rendering task; GPU focuses on the macrooperation amount rendering task of parallel synchronous; Play up rear respective output primitive frame figure; Graphic result integrates sequencing unit 713, for the primitive frame figure exported by CPU and GPU hybrid rending, exports the primitive frame figure of sequence according to fluidisation sequence integration and sequence.
In one embodiment, dynamic fluidized processing module 72 comprises: primitive frame figure segmenting unit 721, for the primitive frame figure playing up output is resolved into each segmentation in order; Segmentation fluidization treatment unit 722, unify video coded treatment resource pool is built for utilizing CPU, GPU fluidisation ability, the coded treatment passage that the primitive frame figure of different fragments correspondence distributes to unified resource pond is carried out fluidization treatment, and after having processed, coded treatment passage exports segmentation Media Stream separately; Fluidisation TB control unit 723, for integrating the segmentation fluidization treatment result that each coded treatment passage exports, carries out integrating the final Media Stream of formation.
The treatment mechanism of disclosure technical scheme is using CPU as high in the clouds arithmetic core, and playing up processing procedure at whole cloud increases GPU concurrent operation advantage, sets up and is more applicable to the new technique scheme playing up development.Wherein CPU and GPU joint-action mechanism, it is one of technological core that integration is played up in establishment fusion, is also the basis realizing CPU and GPU hybrid rending.Play up for 3D Film Animation, it is as follows that GPU+CPU mixed cloud plays up integrity service main process:
(1) software and hardware resources response: hardware resource is called in high in the clouds, start and large-scalely play up basic software, CPU overall coordination resource, such as play up quality, play up duration (start frame--ending frame) according to what set, play up dimension of picture etc. and play up demand, being responsible for overall instruction scheduling, getting out software and hardware resources basis for playing up.
(2) CPU, GPU interlock hybrid rending: be one of core of the technical program, based on the scene picture of art designing's modeling, pixel and person model, according to the rendering requirements set, export and produce the process of serial picture, these pictures are as the primitive frame of final animated video stream.Its key is based on CPU overall tasks distribution mechanism, and logic continuous type rendering task is given CPU serial processing; The task that a large amount of logicality is not strong is given GPU concurrent operation process.Being formed is overall coordination with CPU, gives CPU, a large amount of parallel rendering task is given GPU according to the rendering task division of labor by serial order task, and by the real-time linkage of CPU and GPU rendering result, more efficiently exports corresponding rendering result;
(3) dynamic fluidized process: hybrid rending that CPU, GPU are linked play up the result of generation, the primitive frame playing up generation is transformed into the process of animation streams, the method for realization is gathered and edited according to corresponding format by frame figure to become animated video stream.Because up-to-date CPU and GPU has comparatively powerful Video coding function, by segmenting video stream editor, and the final mode integrated can effective raising efficiency, reaches hardware performance maximization.
(4) character is embedded: in the final 3D Film Animation formed, according to story of a play or opera designing requirement, increases personage leading role's image or cartoon character etc., forms final mute.This process, processes fairly simple comparatively speaking, calls without the need to carrying out complicated CPU, GPU mixing, only need according to special software call-by mechanism, mainly to add according to actual requirement based on CPU and embedded.
Above-mentioned (2), (3) step all needs CPU, GPU to coordinate, and utilizes respective advantage to complete and plays up and fluid mapper process, is to build CPU, GPU mixed cloud to play up main process.
The disclosure abandons existing single CPU, GPU high in the clouds render mode, makes full use of concentrating type CPU serial and GPU concurrent operation advantage, with CPU overall coordination resource, will have the rendering task of logical relation, and give CPU serial processing; By the rendering task on a large scale without logical relation, give GPU parallel processing.Overall building is applicable to playing up from tableaux, and real-time three-dimensional model rendering is to CPU, GPU hybrid rending integration solution of dynamic fluidized output.
So far, described in detail according to CPU and GPU mixed cloud rendering device of the present invention and method and system.In order to avoid covering design of the present invention, details more known in the field are not described.Those skilled in the art, according to description above, can understand how to implement technical scheme disclosed herein completely.
Method and system of the present invention may be realized in many ways.Such as, any combination by software, hardware, firmware or software, hardware, firmware realizes method and system of the present invention.Said sequence for the step of described method is only to be described, and the step of method of the present invention is not limited to above specifically described order, unless specifically stated otherwise.In addition, in certain embodiments, can be also record program in the recording medium by the invention process, these programs comprise the machine readable instructions for realizing according to method of the present invention.Thus, the present invention also covers the recording medium stored for performing the program according to method of the present invention.
Although be described in detail specific embodiments more of the present invention by example, it should be appreciated by those skilled in the art, above example is only to be described, instead of in order to limit the scope of the invention.It should be appreciated by those skilled in the art, can without departing from the scope and spirit of the present invention, above embodiment be modified.Scope of the present invention is limited by claims.
Claims (11)
1. a CPU and GPU mixed cloud rendering intent, is characterized in that, comprising:
Rendering task is decomposed into logic continuous type rendering task and parallel rendering task, described logic continuous type rendering task is sent to CPU serial processing, send to GPU to carry out concurrent operation process described parallel rendering task, and export by the interlock of CPU and GPU rendering result the primitive frame figure playing up generation;
The described primitive frame figure playing up generation integrated according to fluidisation and sorts, exporting the primitive frame figure of sequence;
Carry out segmentation to the primitive frame figure of described sequence, in order primitive frame figure segment assignments is carried out sectional type fluidisation coding to the coding pass in Unified coding treatment source pool, wherein, described Unified coding treatment source pool is built by CPU, GPU code capacity and forms;
The coding result exported by each coding pass carries out sorting and integrates and forms Media Stream.
2. method according to claim 1, is characterized in that, also comprises:
Personage or cartoon character is increased in the Media Stream completing fluidization treatment.
3. method according to claim 1, is characterized in that, the described interlock by CPU and GPU rendering result exports the primitive frame figure playing up generation and comprises:
Before CPU and GPU carries out rendering task, CPU in resource coordination and hardware assigning process for rendering task carries out order registration, distribute according to relevant order in render process and play up, in Output rusults integration process, graphic result is integrated by corresponding corresponding order numbering.
4. method according to claim 1, is characterized in that, describedly sends to CPU serial processing to comprise described logic continuous type rendering task:
CPU employing multinuclear simultaneously treatment mechanism completes serial processing according to priority logic to rendering task, and exports relevant rendering result;
Send to GPU to carry out concurrent operation process described parallel rendering task to comprise:
Adopt the local internal memory of GPU and bus memory mutual, and rely on GPU to share delivery of content respectively to process core to GPU and carry out parallel processing and calculating to rendering task, and final output plays up result accordingly.
5. method as claimed in any of claims 1 to 4, is characterized in that, described logic continuous type rendering task comprises that actor model is played up, action is played up; Described parallel rendering task comprises large scene and plays up.
6. a CPU and GPU mixed cloud rendering device, is characterized in that, comprising:
Interlock hybrid rending module, for rendering task being decomposed into logic continuous type rendering task and parallel rendering task, described logic continuous type rendering task is sent to CPU serial processing, described parallel rendering task is sent to GPU concurrent operation process, and export by the real-time linkage of CPU and GPU rendering result the primitive frame figure playing up generation, the described primitive frame figure playing up generation integrated according to fluidisation and sorts, exporting the primitive frame figure of sequence;
Dynamic fluidized processing module, for the primitive frame figure of the sequence from described interlock hybrid rending module is carried out segmentation, in order primitive frame figure segment assignments is carried out sectional type fluidisation coding to the coding pass in Unified coding treatment source pool, wherein, described Unified coding treatment source pool is built by CPU, GPU code capacity and forms; The coding result exported by each coding pass carries out sorting and integrates and forms Media Stream.
7. device according to claim 6, is characterized in that, also comprises:
Character embedded module, for increasing personage or cartoon character in the Media Stream completing fluidization treatment.
8. device according to claim 6, is characterized in that, described interlock hybrid rending module comprises:
Rendering task classification scheduling unit; for rendering task is decomposed into logic continuous type rendering task and parallel rendering task by rendering task; described logic continuous type rendering task is sent to CPU serial processing, described parallel rendering task is sent to GPU concurrent operation process;
The former figure render processing unit of frame, plays up work accordingly for completing based on the arithmetic element of CPU and GPU, and CPU focuses on single serial macrooperation amount rendering task; GPU focuses on the macrooperation amount rendering task of parallel synchronous; Play up rear respective output primitive frame figure;
Graphic result integrates sequencing unit, for the primitive frame figure exported by CPU and GPU hybrid rending, exports the primitive frame figure of sequence according to fluidisation sequence integration and sequence.
9. device according to claim 6, is characterized in that, described dynamic fluidized processing module comprises:
Primitive frame figure segmenting unit, for resolving into each segmentation in order by the primitive frame figure playing up output;
Segmentation fluidization treatment unit, unify video coded treatment resource pool is built for utilizing CPU, GPU fluidisation ability, the coded treatment passage that the primitive frame figure of different fragments correspondence distributes to unified resource pond is carried out fluidization treatment, and after having processed, coded treatment passage exports segmentation Media Stream separately;
Fluidisation TB control unit, for integrating the segmentation fluidization treatment result that each coded treatment passage exports, carries out integrating the final Media Stream of formation.
10. according to the device in claim 6 to 9 described in any one, it is characterized in that, described logic continuous type rendering task comprises that actor model is played up, action is played up; Described parallel rendering task comprises large scene and plays up.
11. 1 kinds of cloud rendering systems, is characterized in that, comprise CPU and the GPU mixed cloud rendering device in claim 6 to 10 described in any one.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410125332.2A CN104952096B (en) | 2014-03-31 | 2014-03-31 | CPU and GPU mixed clouds rendering intent, device and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410125332.2A CN104952096B (en) | 2014-03-31 | 2014-03-31 | CPU and GPU mixed clouds rendering intent, device and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104952096A true CN104952096A (en) | 2015-09-30 |
| CN104952096B CN104952096B (en) | 2018-06-08 |
Family
ID=54166722
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410125332.2A Active CN104952096B (en) | 2014-03-31 | 2014-03-31 | CPU and GPU mixed clouds rendering intent, device and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104952096B (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105427366A (en) * | 2015-11-11 | 2016-03-23 | 广州华多网络科技有限公司 | Image rendering method and image rendering system |
| CN105916023A (en) * | 2016-05-16 | 2016-08-31 | 武汉斗鱼网络科技有限公司 | GPU-based video rendering method and system |
| CN106447595A (en) * | 2016-09-23 | 2017-02-22 | 努比亚技术有限公司 | Distributed image and graph rendering method and terminal |
| CN106534667A (en) * | 2016-10-31 | 2017-03-22 | 努比亚技术有限公司 | Distributed cooperative rendering method and terminal |
| CN106600521A (en) * | 2016-11-30 | 2017-04-26 | 宇龙计算机通信科技(深圳)有限公司 | Image processing method and terminal device |
| WO2017157062A1 (en) * | 2016-03-15 | 2017-09-21 | 北京金山安全软件有限公司 | 一种动态文件的传输方法、装置及电子设备 method and apparatus for transmitting dynamic document, and electronic device |
| CN108170526A (en) * | 2017-12-06 | 2018-06-15 | 北京像素软件科技股份有限公司 | Load capacity optimization method, device, server and readable storage medium storing program for executing |
| CN108289185A (en) * | 2017-01-09 | 2018-07-17 | 腾讯科技（深圳）有限公司 | A kind of video communication method, device and terminal device |
| CN108961380A (en) * | 2017-05-26 | 2018-12-07 | 阿里巴巴集团控股有限公司 | Method for rendering graph and device |
| CN111190735A (en) * | 2019-12-30 | 2020-05-22 | 湖南大学 | Linux-based on-chip CPU/GPU (Central processing Unit/graphics processing Unit) pipelined computing method and computer system |
| EP3635954A4 (en) * | 2017-04-21 | 2020-07-15 | Zenimax Media Inc. | Systems and methods for rendering & pre-encoded load estimation based encoder hinting |
| CN112153411A (en) * | 2020-08-18 | 2020-12-29 | 南斗六星系统集成有限公司 | Method and system for improving performance of chat room streaming media server |
| CN112163986A (en) * | 2020-09-28 | 2021-01-01 | 山东黄金矿业(莱州)有限公司三山岛金矿 | Distributed processing method for underground logging and mining three-dimensional data of metal mine |
| CN112218117A (en) * | 2020-09-29 | 2021-01-12 | 北京字跳网络技术有限公司 | Video processing method and device |
| WO2021036784A1 (en) * | 2019-08-30 | 2021-03-04 | 中兴通讯股份有限公司 | Media data processing method and apparatus, media server and computer-readable storage medium |
| CN113244614A (en) * | 2021-06-07 | 2021-08-13 | 腾讯科技（深圳）有限公司 | Image picture display method, device, equipment and storage medium |
| CN114529444A (en) * | 2022-04-22 | 2022-05-24 | 南京砺算科技有限公司 | Graphics processing module, graphics processor and graphics processing method |
| CN115375530A (en) * | 2022-07-13 | 2022-11-22 | 北京松应科技有限公司 | Multi-GPU collaborative rendering method, system, device and storage medium |
| CN116433818A (en) * | 2023-03-22 | 2023-07-14 | 宝钢工程技术集团有限公司 | Cloud CPU and GPU parallel rendering method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101587583A (en) * | 2009-06-23 | 2009-11-25 | 长春理工大学 | The farm of playing up based on the GPU cluster |
| CN102609971A (en) * | 2012-01-11 | 2012-07-25 | 南京大学 | Quick rendering system using embedded GPU (Graphics Processing Unit) for realizing 3D-GIS (Three Dimensional-Geographic Information System) |
| CN102663208A (en) * | 2012-04-28 | 2012-09-12 | 浪潮电子信息产业股份有限公司 | Method for carrying out large-eddy simulation real-time rendering by utilizing graphic processing unit (GPU) |
| CN102959517A (en) * | 2010-06-10 | 2013-03-06 | Otoy公司 | Allocation of gpu resources accross multiple clients |
| CN103268220A (en) * | 2012-02-24 | 2013-08-28 | 苏州蓝海彤翔系统科技有限公司 | Software architecture suitable for large-scale animation rendering service cloud platform |
| CN103631634A (en) * | 2012-08-24 | 2014-03-12 | 中国电信股份有限公司 | Graphics processor virtualization achieving method and device |
-
2014
- 2014-03-31 CN CN201410125332.2A patent/CN104952096B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101587583A (en) * | 2009-06-23 | 2009-11-25 | 长春理工大学 | The farm of playing up based on the GPU cluster |
| CN102959517A (en) * | 2010-06-10 | 2013-03-06 | Otoy公司 | Allocation of gpu resources accross multiple clients |
| CN102609971A (en) * | 2012-01-11 | 2012-07-25 | 南京大学 | Quick rendering system using embedded GPU (Graphics Processing Unit) for realizing 3D-GIS (Three Dimensional-Geographic Information System) |
| CN103268220A (en) * | 2012-02-24 | 2013-08-28 | 苏州蓝海彤翔系统科技有限公司 | Software architecture suitable for large-scale animation rendering service cloud platform |
| CN102663208A (en) * | 2012-04-28 | 2012-09-12 | 浪潮电子信息产业股份有限公司 | Method for carrying out large-eddy simulation real-time rendering by utilizing graphic processing unit (GPU) |
| CN103631634A (en) * | 2012-08-24 | 2014-03-12 | 中国电信股份有限公司 | Graphics processor virtualization achieving method and device |
Non-Patent Citations (3)
| Title |
|---|
| 张首斌等: "GPU异构平台下的深度包检测技术", 《天津理工大学学报》 * |
| 薄海光等: "基于GPU加速的SDOG并行可视化实验", 《地理与地理信息科技》 * |
| 许帧: "关于CPU+GPU异构计算的研究与分析", 《科技信息》 * |
Cited By (28)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105427366B (en) * | 2015-11-11 | 2018-07-27 | 广州华多网络科技有限公司 | A kind of image rendering method and image rendering system |
| CN105427366A (en) * | 2015-11-11 | 2016-03-23 | 广州华多网络科技有限公司 | Image rendering method and image rendering system |
| WO2017157062A1 (en) * | 2016-03-15 | 2017-09-21 | 北京金山安全软件有限公司 | 一种动态文件的传输方法、装置及电子设备 method and apparatus for transmitting dynamic document, and electronic device |
| CN105916023A (en) * | 2016-05-16 | 2016-08-31 | 武汉斗鱼网络科技有限公司 | GPU-based video rendering method and system |
| CN105916023B (en) * | 2016-05-16 | 2019-04-23 | 武汉斗鱼网络科技有限公司 | A kind of Video Rendering method and system based on GPU |
| CN106447595A (en) * | 2016-09-23 | 2017-02-22 | 努比亚技术有限公司 | Distributed image and graph rendering method and terminal |
| CN106534667A (en) * | 2016-10-31 | 2017-03-22 | 努比亚技术有限公司 | Distributed cooperative rendering method and terminal |
| CN106534667B (en) * | 2016-10-31 | 2020-02-11 | 努比亚技术有限公司 | Distributed collaborative rendering method and terminal |
| CN106600521A (en) * | 2016-11-30 | 2017-04-26 | 宇龙计算机通信科技(深圳)有限公司 | Image processing method and terminal device |
| CN108289185A (en) * | 2017-01-09 | 2018-07-17 | 腾讯科技（深圳）有限公司 | A kind of video communication method, device and terminal device |
| EP3635954A4 (en) * | 2017-04-21 | 2020-07-15 | Zenimax Media Inc. | Systems and methods for rendering & pre-encoded load estimation based encoder hinting |
| US11503313B2 (en) | 2017-04-21 | 2022-11-15 | Zenimax Media Inc. | Systems and methods for rendering and pre-encoded load estimation based encoder hinting |
| CN108961380B (en) * | 2017-05-26 | 2022-06-14 | 创新先进技术有限公司 | Graph rendering method and device |
| CN108961380A (en) * | 2017-05-26 | 2018-12-07 | 阿里巴巴集团控股有限公司 | Method for rendering graph and device |
| CN108170526A (en) * | 2017-12-06 | 2018-06-15 | 北京像素软件科技股份有限公司 | Load capacity optimization method, device, server and readable storage medium storing program for executing |
| WO2021036784A1 (en) * | 2019-08-30 | 2021-03-04 | 中兴通讯股份有限公司 | Media data processing method and apparatus, media server and computer-readable storage medium |
| CN111190735A (en) * | 2019-12-30 | 2020-05-22 | 湖南大学 | Linux-based on-chip CPU/GPU (Central processing Unit/graphics processing Unit) pipelined computing method and computer system |
| CN111190735B (en) * | 2019-12-30 | 2024-02-23 | 湖南大学 | On-chip CPU/GPU pipelining calculation method based on Linux and computer system |
| CN112153411A (en) * | 2020-08-18 | 2020-12-29 | 南斗六星系统集成有限公司 | Method and system for improving performance of chat room streaming media server |
| CN112163986A (en) * | 2020-09-28 | 2021-01-01 | 山东黄金矿业(莱州)有限公司三山岛金矿 | Distributed processing method for underground logging and mining three-dimensional data of metal mine |
| CN112218117A (en) * | 2020-09-29 | 2021-01-12 | 北京字跳网络技术有限公司 | Video processing method and device |
| CN113244614B (en) * | 2021-06-07 | 2021-10-26 | 腾讯科技（深圳）有限公司 | Image picture display method, device, equipment and storage medium |
| CN113244614A (en) * | 2021-06-07 | 2021-08-13 | 腾讯科技（深圳）有限公司 | Image picture display method, device, equipment and storage medium |
| CN114529444A (en) * | 2022-04-22 | 2022-05-24 | 南京砺算科技有限公司 | Graphics processing module, graphics processor and graphics processing method |
| CN114529444B (en) * | 2022-04-22 | 2023-08-11 | 南京砺算科技有限公司 | Graphics processing module, graphics processor, and graphics processing method |
| CN115375530A (en) * | 2022-07-13 | 2022-11-22 | 北京松应科技有限公司 | Multi-GPU collaborative rendering method, system, device and storage medium |
| CN116433818A (en) * | 2023-03-22 | 2023-07-14 | 宝钢工程技术集团有限公司 | Cloud CPU and GPU parallel rendering method |
| CN116433818B (en) * | 2023-03-22 | 2024-04-16 | 宝钢工程技术集团有限公司 | Cloud CPU and GPU parallel rendering method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104952096B (en) | 2018-06-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104952096A (en) | CPU and GPU hybrid cloud rendering method, device and system | |
| US20200045285A1 (en) | Adaptive resolution of point cloud and viewpoint prediction for video streaming in computing environments | |
| US20210266515A1 (en) | Selective packing of patches for immersive video | |
| CN102592315B (en) | 3D rendering platform based on GPU cloud cluster | |
| US10169084B2 (en) | Deep learning via dynamic root solvers | |
| DE102019120554A1 (en) | VIDEO TRANSCODING MECHANISM WITH SIX DEGREE OF FREEDOM | |
| JP5154551B2 (en) | Fast reconstruction of graphics pipeline state | |
| DE102019120661A1 (en) | Video refinement mechanism | |
| US20200043122A1 (en) | Point-based rendering and removal of projection noise | |
| CN111754614A (en) | Video rendering method and device based on VR (virtual reality), electronic equipment and storage medium | |
| JP2021149941A (en) | Motion blur using dynamic quantization grid | |
| CN109936745A (en) | For improving the method and system of the decompression of original video data | |
| CN108694376A (en) | Including static scene determination, occlusion detection, frame rate conversion and the video motion processing for adjusting compression ratio | |
| Zhang et al. | Multi-GPU Parallel Pipeline Rendering with Splitting Frame | |
| DE102021123500A1 (en) | Unified memory compression mechanism | |
| CN102123137B (en) | Multi-person collaborative virtual-real mixed disassembly and assembly system and multi-person collaborative virtual-real mixed disassembly and assembly method | |
| Zhang et al. | F-CAD: A framework to explore hardware accelerators for codec avatar decoding | |
| DE102020114630A1 (en) | Grid shader output management technology | |
| DE102020106728A1 (en) | Background estimation for object segmentation using coarse level tracking | |
| WO2023160041A1 (en) | Image rendering method and apparatus, computer device, computer-readable storage medium and computer program product | |
| Liu et al. | Design and implementation of distributed rendering system | |
| CN113177263A (en) | Method and system for establishing digital twin body of ship equipment | |
| CN112070868B (en) | Animation playing method based on iOS system, electronic equipment and medium | |
| KR20180008168A (en) | Holographic Content Creation System and Method using a Parallel Processing Environment | |
| Streit et al. | Understanding next-generation VR: classifying commodity clusters for immersive virtual reality |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |