Power Efficiency Ranking Power Efficiency Ranking    |    Performance vs. Efficiency Tradeoff    |    INT8 Results    |    FP16 Results
Processor AI Accelerator Year Lib Inference Mode INT8,
FPS × (FPS per Watt)		 FP16,
FPS × (FPS per Watt)		 Perf. – Efficiency
Balance
Snapdragon 8 EliteHexagon DSP / HTP Gen 42024qh.qhFAST SINGLE ANSWER18540 86.0
Snapdragon 8 Gen 3Hexagon DSP / HTP Gen 32023qh.qhFAST SINGLE ANSWER15632 70.8
Snapdragon 8 Gen 2Hexagon DSP / HTP Gen 22022qh.qhSUSTAINED SPEED12926.5 58.6
Dimensity 9300APU 7902023mmFAST SINGLE ANSWER93.433.2 55.7
Dimensity 9400NPU 8902025mmFAST SINGLE ANSWER10627.6 54.2
Dimensity 8400NPU 8802024mmFAST SINGLE ANSWER82.127.7 47.7
Dimensity 8300 UltraAPU 7802023mmFAST SINGLE ANSWER79.825.3 44.9
Snapdragon 8 Gen 1Hexagon DSP / HTP2021qh.qhSUSTAINED SPEED81.416.4 36.5
Dimensity 9200APU 6902022mmSUSTAINED SPEED41.716.5 26.2
Snapdragon 888DSP (Hexagon 780) + GPU (Adreno 660)2020qh.qgSUSTAINED SPEED89.76.9 24.9
Dimensity 9000APU 5902021mmSUSTAINED SPEED34.412 20.3
Snapdragon 7 Gen 1Hexagon DSP / HTP2022qh.qhFAST SINGLE ANSWER417.2 17.2
Google Tensor G2Google Tensor TPU 2.02022nnSUSTAINED SPEED23.99.6 15.1
Google TensorGoogle Tensor TPU2021nnSUSTAINED SPEED12.18 9.8
Dimensity 800APU 3.0 (4 cores)2020nnSUSTAINED SPEED12.74 7.1
Dimensity 1000+APU 3.0 (6 cores)2019nnSUSTAINED SPEED10.24.4 6.7
Dimensity 820APU 3.0 (4 cores)2020nnSUSTAINED SPEED113.3 6.0
Snapdragon 865DSP (Hexagon 698) + GPU (Adreno 650)2019hgSUSTAINED SPEED8.34.1 5.8
Kirin 9000GPU (Mali-G78 MP24)2020ggSUSTAINED SPEED5.25.5 5.3
Snapdragon 855DSP (Hexagon 690) + GPU (Adreno 640)2018hgSUSTAINED SPEED7.13.3 4.8
Exynos 2100GPU (Mali-G78 MP14)2021ggSUSTAINED SPEED4.74.9 4.8
Kirin 990 5GGPU (Mali-G76 MP10)2019ggSUSTAINED SPEED4.24.5 4.3
Exynos 9820 OctaGPU (Mali-G76 MP12)2018ggSUSTAINED SPEED3.53.7 3.6
Kirin 980GPU (Mali-G76 MP10)2018ggSUSTAINED SPEED3.23.5 3.3
Snapdragon 845DSP (Hexagon 685) + GPU (Adreno 630)2018hgSUSTAINED SPEED42.4 3.1
Snapdragon 710DSP (Hexagon 685) + GPU (Adreno 616)2018hgSUSTAINED SPEED3.61.5 2.3

xy  -  This device accelerates quantized [x] and floating-point [y] models using the:


n   -   Android NNAPI

g   -   TensorFlow Lite GPU Delegate  (OpenCL / OpenGL / Metal based)


h     -   Qualcomm Hexagon NN Direct Delegate

qh   -   Qualcomm QNN HTP Delegate

qd   -   Qualcomm QNN DSP Delegate

qg   -   Qualcomm QNN GPU Delegate


e   -   Samsung ENN Delegate

m  -   MediaTek Neuron Delegate

i    -    Apple CoreML Delegate

c   -   TensorFlow Lite / NNAPI default CPU backend


1  -  This SoC might be using unofficial / prototype hardware or drivers

2  -  These are the results of an early prototype. The results of the commercial SoC might be different

Copyright © 2018-2023 by A.I.

ETH Zurich, Switzerland