
AI Benchmark
Er din telefon klar til kunstig intelligens? Kør AI Benchmark for at kontrollere dens ydeevne!
Appinfo
Advertisement
Appbeskrivelse
Android App -Analyse Og Gennemgang: AI Benchmark, Udviklet Af Ignatov Andrey. Listet I Værktøjer -Kategori. Nuværende Version Er 6.0.2, Opdateret På 16/03/2025 . Ifølge Brugere Anmeldelser På Google Play: AI Benchmark. Opnået Over 252 Tusind Installationer. AI Benchmark Har I Øjeblikket 2 Tusind Anmeldelser, Gennemsnitlig Vurdering 4.4 Stjerner
Neural billedgenerering, ansigtsgenkendelse, billedklassificering, besvarelse af spørgsmål...Er din smartphone i stand til at køre de nyeste Deep Neural Networks til at udføre disse og mange andre AI-baserede opgaver? Har den en dedikeret AI-chip? Er det hurtigt nok? Kør AI Benchmark for professionelt at evaluere dens AI-ydelse!
Nuværende telefonrangering: http://ai-benchmark.com/ranking
AI Benchmark måler hastigheden, nøjagtigheden, strømforbruget og hukommelseskravene for flere vigtige AI-, Computer Vision- og NLP-modeller. Blandt de testede løsninger er metoder til billedklassificering og ansigtsgenkendelse, AI-modeller, der udfører neurale billeder og tekstgenerering, neurale netværk, der bruges til billed-/video-superopløsning og fotoforbedring, samt AI-løsninger, der bruges i autonome køresystemer og smartphones til real- tidsdybdeestimering og semantisk billedsegmentering. Visualiseringen af algoritmernes output gør det muligt at vurdere deres resultater grafisk og at lære den aktuelle state-of-the-art at kende inden for forskellige AI-områder.
I alt består AI Benchmark af 83 tests og 30 sektioner anført nedenfor:
Afsnit 1. Klassifikation, MobileNet-V3
Afsnit 2. Klassifikation, Inception-V3
Afsnit 3. Ansigtsgenkendelse, Swin Transformer
Afsnit 4. Klassifikation, EfficientNet-B4
Afsnit 5. Klassifikation, MobileViT-V2
Afsnit 6/7. Parallel modeludførelse, 8 x Inception-V3
Afsnit 8. Objektsporing, YOLO-V8
Afsnit 9. Optisk tegngenkendelse, ViT Transformer
Afsnit 10. Semantisk segmentering, DeepLabV3+
Afsnit 11. Parallel Segmentering, 2 x DeepLabV3+
Afsnit 12. Semantisk segmentering, segmentér hvad som helst
Afsnit 13. Fotosløring, IMDN
Afsnit 14. Image Super-Resolution, ESRGAN
Afsnit 15. Image Super-Resolution, SRGAN
Afsnit 16. Image Denoising, U-Net
Afsnit 17. Dybdevurdering, MV3-Dybde
Afsnit 18. Dybdevurdering, MiDaS 3.1
§ 19/20. Billedforbedring, DPED
Afsnit 21. Lært kamera ISP, MicroISP
Afsnit 22. Bokeh Effect Rendering, PyNET-V2 Mobile
Afsnit 23. FullHD Video Super-Resolution, XLSR
§ 24/25. 4K Video Super-opløsning, VideoSR
Afsnit 26. Besvarelse af spørgsmål, MobileBERT
Afsnit 27. Neural tekstgenerering, Lama2
Afsnit 28. Generering af neural tekst, GPT2
Afsnit 29. Neural billedgenerering, stabil diffusion V1.5
Afsnit 30. Hukommelsesgrænser, ResNet
Udover det kan man indlæse og teste deres egne TensorFlow Lite deep learning-modeller i PRO Mode.
En detaljeret beskrivelse af testene kan findes her: http://ai-benchmark.com/tests.html
Bemærk: Hardwareacceleration understøttes på alle mobile SoC'er med dedikerede NPU'er og AI-acceleratorer, inklusive Qualcomm Snapdragon, MediaTek Dimensity / Helio, Google Tensor, HiSilicon Kirin, Samsung Exynos og UNISOC Tiger-chipsæt. Fra AI Benchmark v4 kan man også aktivere GPU-baseret AI-acceleration på ældre enheder i indstillingerne ("Accelerate" -> "Enable GPU Acceleration" / "Arm NN", OpenGL ES-3.0+ er påkrævet).
Vi Tilbyder I Øjeblikket Version 6.0.2. Dette Er Vores Seneste, Mest Optimerede Version. Det Er Velegnet Til Mange Forskellige Enheder. Gratis Download Direkte Apk Fra Google Play Store Eller Andre Versioner, Vi Er Vært For. Desuden Kan Du Downloade Uden Registrering Og Der Ikke Kræves Noget Login.
Vi Har Mere End 2000+ Tilgængelige Enheder Til Samsung, Xiaomi, Huawei, Oppo, Vivo, Motorola, LG, Google, OnePlus, Sony, Tablet ... Med Så Mange Muligheder, Det Er Let For Dig At Vælge Spil Eller Software, Der Passer Til Din Enhed.
Det Kan Komme Godt Med, Hvis Der Er Nogen Landsbegrænsninger Eller Begrænsninger Fra Siden Af Din Enhed I Google App Store.
Hvad Er Nyt
1. New tasks and models: Vision Transformer (ViT) architectures, Large Language Models (LLMs), Stable Diffusion network, etc.
2. Added tests checking the performance of quantized INT16 inference.
3. LiteRT (TFLite) runtime updated to version 2.17.
4. Updated Qualcomm QNN, MediaTek Neuron, TFLite NNAPI, GPU and Hexagon NN delegates.
5. Added Arm NN delegate for AI inference acceleration on Mali GPUs.
6. The total number of tests increased to 83.
2. Added tests checking the performance of quantized INT16 inference.
3. LiteRT (TFLite) runtime updated to version 2.17.
4. Updated Qualcomm QNN, MediaTek Neuron, TFLite NNAPI, GPU and Hexagon NN delegates.
5. Added Arm NN delegate for AI inference acceleration on Mali GPUs.
6. The total number of tests increased to 83.
Seneste Kommentarer
Abraham Braun
This is a very interesting app that showcases the different tasks the NPU has to deal with. Very intuitive. One gripe though, my 2 year old S21 Ultra scored 734.9k while my brand-new S23 Ultra scored a measly 132.6k. Both were running the latest of their respective softwares (Android 13 One UI 5.0/5.1). Anybody rocking an S23 with similar results? I should point out that the S23 did score higher in the accuracy and CPU. NN-Speed was WAY higher on the S21 than that of the S23 (48-337 vs 16-16).
Noland C
Stops working on Samsung... Trying to compare with OnePlus but it bugged out... Useless.
A Google user
Seems like it'd work if I had a better phone, but half way through the test it said "Unfortunately, AI Benchmark has stopped". I don't know if it's my phone, or the app it self. I'll be willing to give a higher rating if it gets fixed. My phone is a Samsung J7 Sky pro. edit: I just realized it's supposed to crash, sorry! I'll give it a 5 star now lol. it's a good app.
Edward Dolan
Very fun benchmark to run and watch! I wasn't aware phone AI could do some of those things. My phone didn't crash during the last test (Galaxy S10 Plus running Android 10). I'm just surprised by the low score (67.4k). I might try to run it again after clearing my background apps and RAM.
Luiz Brito
This app crashed my phone so hard it couldn't do anything for 20 minutes. Black screen, my charging led indicator not working and all. I got scared. Tried force restarting 4 times before it finally woke up. Aftee that it went into thqt booting cycle for three times before it went back into my home page. There's something really shady with this app, uninstalled immediately. (Snapdragon gen 2, rog phone 7).
A Google user
Works on some of my phones, crashes on others, keeps crashing on test 5/11, deblurring, right after Atari test on one of my devices, can't get a score because I get an exception error every time. So on nice phones, it works, on budget soc's it's hit and miss.
Kevin Holmes
Seems like app is only showing first 3 digits of final score?? OnePlus Nord3 with dimensity 9000 is showing score 109, which is half my old phone with dimensity 1200, but text below 109 score says "excellent" Vs “good“ on old phone, and ranking list shows score as 1092, so it seems like a bug???
Phil Mac
Really interesting app. I still have to read alot about how the scoring and acceleration works. When I get my new Samsung S22 Ultra, which is claimed to have 73% more powerful NPU than last years model...I'm keen to run it along side my current Samsung Note10 plus at the same time. which is 2.5 yrs old now and see the difference.