JPEGD Performance Specifications

Performance Specifications

This section describes only the performance specifications of the AI processor of each product, which may differ based on the memory and CPU configurations of the respective product. For details about the product performance specifications, see the official descriptions at the Ascend community.

This section takes a single NPU as an example to describe the performance specifications in basic scenarios, as shown in the table below. The performance at other resolutions can be estimated in the same way.

In the following table, one channel corresponds to one thread, 720p indicates a resolution of 1280 × 720 pixels, 1080p indicates a resolution of 1920 × 1080 pixels, and 4K indicates a resolution of 3840 × 2160 pixels.

The performance specifications of JPEGD are based on hardware decoding performance. JPEGD hardware decoding does not support the image decoding with three SOS markers. For image formats not supported by hardware, software decoding is used. The software decoding performance for reference is 1080p at 15 FPS. If the output image decoded by JPEGD involves rotation, the performance is lower than the reference values of software decoding. For example, the performance is lower than 15 FPS for 1080p images.

On the Atlas A3 training products/Atlas A3 inference products, an NPU has two devices, and the maximum specifications can be reached only when the two devices work in parallel.

**Table 1** AI processor performance specifications of the Ascend 950PR/Ascend 950DT
Input Image Format	Input Image Size	Example	Total Frame Rate
YUV400/420/YUV440/422/444	≤ 500 KB	n-channel 1080p (1 ≤ n ≤ 8)	n × 512 FPS
YUV400/420/YUV440/422/444	≤ 500 KB	n-channel 1080p (n > 8)	4096 FPS
YUV400/420/YUV440/422/444	≤ 2 MB	n-channel 4K (1 ≤ n ≤ 8)	n*128fps
YUV400/420/YUV440/422/444	≤ 2 MB	n-channel 4K (n > 8)	1024fps

**Table 2** AI processor performance specifications of the Atlas A3 training products/Atlas A3 inference products
Input Image Format	Input Image Size	Example	Total Frame Rate
YUV400/420	≤ 500 KB	n-channel 1080p (1 ≤ n ≤ 56)	n × 512 FPS
YUV400/420	≤ 500 KB	n-channel 1080p (n > 56)	28672fps
YUV440/422/444	≤ 500 KB	n-channel 1080p (1 ≤ n ≤ 56)	n*256fps
YUV440/422/444	≤ 500 KB	n-channel 1080p (n > 56)	14336fps
YUV400/420	≤ 2 MB	n-channel 4k (1 ≤ n ≤ 56)	n*128fps
YUV400/420	≤ 2 MB	n-channel 4k (n > 56)	7168fps
YUV440/422/444	≤ 2 MB	n-channel 4k (1 ≤ n ≤ 56)	n*64fps
YUV440/422/444	≤ 2 MB	n-channel 4k (n > 56)	3584fps

**Table 3** AI processor performance specifications of the Atlas A2 training products/Atlas A2 inference products
Input Image Format	Input Image Size	Example	Total Frame Rate
YUV400/420	≤ 500 KB	n-channel 1080p (1 ≤ n ≤ 28)	n × 512 FPS
YUV400/420	≤ 500 KB	n-channel 1080p (n > 28)	14336fps
YUV440/422/444	≤ 500 KB	n-channel 1080p (1 ≤ n ≤ 28)	n*256fps
YUV440/422/444	≤ 500 KB	n-channel 1080p (n > 28)	7168fps
YUV400/420	≤ 2 MB	n-channel 4K (1 ≤ n ≤ 28)	n*128fps
YUV400/420	≤ 2 MB	n-channel 4K (n > 28)	3584fps
YUV440/422/444	≤ 2 MB	n-channel 4K (1 ≤ n ≤ 28)	n*64fps
YUV440/422/444	≤ 2 MB	n-channel 4K (n > 28)	1792fps

**Table 4** AI processor performance specifications of the Atlas 200I/500 A2 inference products
Example	Total Frame Rate
n-channel 1080p (1 ≤ n ≤ 2)	n*256fps
n-channel 1080p (n > 2)	512fps
n-channel 4K (1 ≤ n ≤ 2)	n*64fps
n-channel 4K (n > 2)	128fps

**Table 5** AI processor performance specifications of the Atlas inference products
Example	Total Frame Rate
n-channel 1080p (1 ≤ n ≤ 16)	n*128fps
n-channel 1080p (n > 16)	2048fps
n-channel 4K (1 ≤ n ≤ 16)	n*32fps
n-channel 4K (n > 16)	512fps

Parent topic: VDEC/JPEGD