A method of increasing the number of real-time layers in large-format nonlinear editing based on multiple GPUs
A non-linear editing and large-format technology, applied in the field of video editing, can solve the problem of limited real-time performance of non-linear editing, achieve the effect of improving GPU task division, increasing the number of real-time layers of ultra-high-definition editing, and improving rendering performance
Active Publication Date: 2020-12-25
CHENGDU SOBEY DIGITAL TECH CO LTD
View PDF11 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
We tested this method and found that in the application scenario of NLE software, the way of SLI firefighting with dual GPUs is not ideal, and the overall real-time improvement of NLE is very limited, which can only bring about 10%-20 % performance improvement
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0026] All features disclosed in all embodiments in this specification (including any appended claims, abstract and drawings), or steps in all methods or processes implicitly disclosed, except for mutually exclusive features and / or steps, can be used as Combine or replace in any way.
[0027] like figure 1 , shown in 2, a method for increasing the number of real-time layers of large-format nonlinear editing based on multiple GPUs, including steps:
[0028] S1, multiple GPUs are connected to the PCIE bus through independent PCIE slots, and the PCIE bus is connected to the CPU; the frames decoded by the CPU are alternately uploaded to multiple GPUs for rendering according to the frame number;
[0029] S2, reordering the rendered frames;
[0030] S3, the non-linear programming software sequentially fetches frames from the sorted frames, and judges which GPU the frame is on, calls the downlink of the corresponding GPU, and sends it to the IO card for output.
[0031] Further, i...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a method for increasing the number of large-format nonlinear editing real-time layers based on multiple GPUs, and the method comprises the steps: S1, enabling the GPUs to be connected with a PCIE bus through independent PCIE slots, and enabling the PCIE bus to be connected with a CPU; alternately uploading the frames decoded by the CPU to a plurality of GPUs for rendering according to frame numbers; S2, reordering the rendered frames; S3, enabling the non-editing software to take frames from the sorted frames in sequence, judges which GPU the frames are located on, calling the downlink of the corresponding GPU, and sending the downlink to an IO card for output and the like. According to the method, on one hand, the overall rendering capability of the non-editing software is enhanced, and on the other hand, the ultra-high-definition editing real-time layer number of the non-editing software is remarkably increased; moreover, dynamic subtitles and special effectscan be rendered by using a plurality of GPUs (Graphics Processing Units), so that the balance of the compute-intensive threads among a plurality of CPU groups is realized, and all cores and the like of the CPUs are fully utilized.
Description
technical field [0001] The invention relates to the technical field of video editing, and more specifically, to a method for increasing the number of real-time layers of large-format nonlinear editing based on multiple GPUs. Background technique [0002] Non-linear editing software In the process of video editing, video decoding and video rendering are the two most important links. Currently, the mainstream architecture of NLE software is CPU+single GPU. Generally speaking, video decoding is done on the CPU, and video rendering is done on the GPU. The CPU passes the decoded data to the GPU through the PCIE bus (data uplink), and the GPU passes the rendered data to the CPU through the PCIE bus. (data downlink). [0003] The architecture of CPU+single GPU worked well in the era of SD and HD, but after entering the era of UHD, this architecture gradually exposed performance problems. When the number of real-time layers for ultra-high-definition 4K 50p editing reaches 6 layer...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06F9/50G06T1/20G06F13/40H04N5/262H04N19/42A63F13/60
CPCA63F13/60G06F9/5027G06F13/4068G06F2209/508G06F2213/0026G06T1/20H04N5/262H04N19/42
Inventor 罗天张金沙郑建宏张旭
Owner CHENGDU SOBEY DIGITAL TECH CO LTD