Comparativa entre Orange Pi 5B y Raspberry Pi 5: Rendimiento en LLMs
En el mundo de la computación de placa única (SBC), el Orange Pi 5B y el Raspberry Pi 5 son dos opciones muy populares. Pero, ¿cómo se comparan realmente en términos de rendimiento cuando se trata de ejecutar modelos de lenguaje como el 53 Mini? En este artículo, exploramos las diferencias clave y descubrimos cuál de estos dispositivos se desempeña mejor en tareas específicas.
Preparación para la batalla
Antes de sumergirnos en los resultados, es importante señalar que el Orange Pi 5B tiene 16 GB de RAM, mientras que el Raspberry Pi 5 cuenta con 8 GB. Esto ya establece una diferencia en cuanto a la capacidad de procesamiento. Ambos dispositivos se preparan para el desafío, pero el Orange Pi parece tener una ventaja en cuanto a potencia. Para aquellos interesados en otras comparativas de capacidad, puedes ver la Comparativa de Capacidades de Visión AI: Jetson Nano Super vs Raspberry Pi 5.
La prueba inicial
Para esta comparación, ejecutamos el modelo 53 Mini en ambas placas. La Raspberry Pi 5 se inició primero, y aunque no pudimos grabar la pantalla debido a limitaciones de recursos, fue un proceso bastante sencillo. Se lanzó el modelo y se hizo una pregunta simple: "¿Qué es un portátil?" La respuesta fue rápida y proporcionó información general, aunque algo desactualizada.
Rendimiento del Raspberry Pi 5
- Velocidad de respuesta: A pesar de ser un modelo más pequeño, el Raspberry Pi 5 mostró un rendimiento adecuado. La generación de respuestas fue razonable, lo que sugiere que es una buena opción para quienes desean experimentar con LLMs sin complicaciones. Si te interesa conocer más sobre monitores que complementen tu experiencia con juegos, consulta nuestra guía sobre Los Mejores Monitores para PlayStation 5: Guía Completa y Consejos.
Cambiando a Orange Pi 5B
A continuación, cambiamos al Orange Pi 5B. Este dispositivo también se configuró para ejecutar Ubuntu, pero notamos que el proceso era menos intuitivo ya que requerimos SSH para interactuar con él. Esto podría ser un inconveniente para aquellos que prefieren una interfaz gráfica de usuario.
Comparativa de modelos
- Modelo utilizado: Mientras que el Raspberry Pi 5 utilizó un modelo de cuantificación de 4 bits, el Orange Pi 5B empleó uno de 8 bits. Esto significa que, aunque el Orange Pi tiene una mayor capacidad de RAM, el modelo de 8 bits podría ofrecer más detalles en las respuestas.
Rendimiento del Orange Pi 5B
- Velocidad de respuesta: Al hacer la misma pregunta, el Orange Pi 5B respondió notablemente más rápido. Sin embargo, hubo momentos en que la salida no fue tan coherente, mostrando elementos no deseados en la respuesta.
Evaluación final
Después de probar ambos dispositivos, surgen algunas conclusiones:
- Raspberry Pi 5: Ideal para quienes buscan facilidad y un rendimiento más que adecuado para explorar LLMs. Es perfecto para principiantes y aquellos que no quieren complicarse con configuraciones técnicas. Para más detalles sobre la elección de monitores, revisa Los 3 Mejores Monitores Gaming Calidad Precio para PlayStation 5.
- Orange Pi 5B: Mejor para desarrolladores o aquellos con casos de uso específicos que requieren un mayor rendimiento. Tiene el potencial de ejecutar tareas más complejas, pero puede implicar más esfuerzo en la configuración.
Conclusión
En resumen, si bien el Orange Pi 5B ofrece un rendimiento superior en tareas exigentes, el Raspberry Pi 5 es más accesible y fácil de usar, lo que lo convierte en la opción ideal para la mayoría de los usuarios. La elección entre los dos depende realmente de tus necesidades específicas y de tu nivel de comodidad con la tecnología. ¡Espero que esta comparativa te haya sido útil! Además, si estás considerando monitores para mejorar tu experiencia visual, asegúrate de leer sobre ¿Vale la pena comprar un monitor OLED? Ventajas, desventajas y recomendaciones.
so in today's video we're going to be comparing these two very well-known sbcs on the right we have an orange Pi 5B
with 16 GB of RAM and on the left we have a Raspberry Pi 5 with 8 GB of RAM so today's video is going to focus on
how these two perform running five 3 Mini in the tech jungle the orange P lies in weight circuits humming with
power near buy the Raspberry Pi 5 fidgets unaware of the challenge ahead both prepare for battle all right so the
pi is up and running now unfortunately I am not able to screen record during this because running OBS would just be more
of a constraint on system resources and I want to keep it as lightweight as possible so all I'm going to do here
simply and I will zoom in and show the screen is Ol run 53.8 b mini 53 3.8 4K instruct q40 so this is the 4bit
Quant and once I press enter we'll start going and once that's done it will kind of take a while to get everything loaded
and set up so once that happens we'll start chatting all right so it seems like the
model is loaded and you can see that it's asking me to send a message so I just going to do a very simple default
question which is is going to be what is a laptop and the response will happen now
live in real time and we can see it is going at a reasonable speed Now understand this is a smaller model
in terms of parameters but this is a large improvement over what I've noticed on the 4B so and the point of this video
is kind of just to show you the speed differential between the two and then kind of explain some of the nuances
between running it on the pi or running it on an orange pie or something else with the rock chip Processing
Unit so I'll just kind of let this finish all right so you can see that the generation has completed and it was a
bit slower than you would want to actually be reading something however I would absolutely not call this like just
uselessly slow or anything like that I would say performance here especially for someone who's just kind of
interested in playing with llms on a pi is definitely definitely more than adequate and this is a rather decent
model I would say as you can see the response is lucid and gives just kind of some general information perhaps a bit
out of date like mentioning um Mac OS monery or something like that but other than
that pretty solid so now we'll pop over and try the orange P okay so we now have the orange Pi 5 loading up and you can
see that I've pushed this keyboard out of the way and it is loading Ubuntu and this is
20244 LTS however it is a version I found on GitHub that supposedly has the correct drivers and setup to actually
take advantage of the neural Processing Unit in this little chip here so that's kind of a long-winded way to say that
the way to actually use this right now is going to be to SSH into it which is why I have this laptop here so for the
remainder of the testing of the orange pi5 we're actually going to just be in a shell in SSH from this laptop
so we'll do that all right now I suppose we could kind of
immediately say that this is one negative against the orange pie is just the fact that you can't really use it
from the desktop like we did with the Raspberry Pi so just kind of a little bit of a pain and just to show here real
quick you can see the orange pie is right there off to the right side and now we'll go back in here and I will
zoom in and we'll just kind of go ahead and do the same thing we did did on Raspberry Pi which is test 53
mini so I just navigate into this directory and again this isn't a tutorial video it's just kind of a
comparison test so please keep that in mind when I do like actually go through the setup stuff here now something I
quickly want to note is there are a lot less models available for running on this specific architecture and one of
the differences in this test that that leads to is that this model here of 53 that I'm using is an 8bit Quant and the
one that was being used on the pi5 is on the Raspberry Pi 5 is a 4bit Quant so without getting super technical um I I
suppose pretend like we're playing Halo and we're on the Blood Gulch map in multiplayer the 4bit Quant is just kind
of like seeing that map with less detail but you're still on the same map and then the big Quant that the orange pie
is using is kind of being on that map and you're seeing a bit more detail but again you're still on the same map so
it's probably like a a janky explanation but just kind of something to keep note of in terms of a comparison um I want to
be as Fair as possible all right so the model's loaded in and as I did before I'm just going to go ahead and
say what is a laptop sorry I'm typing with my wrong hand and one-handed so that's why all
right and we can see here that the response time is a lot faster really like a lot faster and this
model is actually technically a bit more challenging to
run in terms of the required specific a so being that this is a 16 GB of RAM model of the orange Pi that is kind of
assisting here in actually running this and allowing a 8 bit Quant like this to run so what I'm saying is hypothetically
if this was running the 4bit Quant that the Raspberry Pi 5B was running it would actually be generating faster than this
now something I have noticed and heard is that the actual output of this can be kind of of a
little um if you will non-lucid so like you can see here it just um it's actually showing like things it
shouldn't be in here in the generation like the end and the assistant and things like that and then it also kind
of sometimes has a tendency to begin speaking to itself so definitely kind of a bit different um and I would say just
from an ease of view standpoint it is much better to just go with the pi five the Raspberry Pi five so that's going to
go ahead and conclude our quick testing of the orange Pi 5B and the Raspberry Pi 5 in running llms I think overall if
you're someone who just kind of wants to mess around with this and doesn't want to have to do a lot of extra work or
troubleshooting or anything like that the pi5 is totally the way to go if you are perhaps a developer or somebody
looking for a really Niche use case for an SBC that requires just the maximum amount of performance I would go with
the orange pie for example I am kind of thinking about taking this little arcade here and running a small local llm based
game on it and in that scenario I would totally use the orange pi5 just because I could get a little more power out of
it and obviously I would have to contend with not being able to use the desktop and stuff like that but with that said I
think for probably 95% of those of us out there if you want to do some llm stuff and you have a pi five you're
going to be more than happy with playing around with it so I hope this was helpful I figured it was a video I would
make considering I got a high five and want to kind of do some benchmarking on it so yeah
Heads up!
This summary and transcript were automatically generated using AI with the Free YouTube Transcript Summary Tool by LunaNotes.
Generate a summary for freeRelated Summaries

Alternativas al Raspberry Pi 4: Explorando los Mejores SBCs
Descubre las mejores alternativas al Raspberry Pi 4, desde el Banana Pi hasta el Orange Pi 5, y encuentra la opción perfecta para tus proyectos.

Comparativa de Capacidades de Visión AI: Jetson Nano Super vs Raspberry Pi 5
Descubre las capacidades de visión AI de Jetson Nano Super y Raspberry Pi 5 en este tutorial comparativo y práctico.

Los Mejores Monitores para PlayStation 5: Guía Completa y Consejos
Descubre los mejores monitores para PlayStation 5, desde los más económicos hasta los de alta gama. Aprende qué características buscar.

Los 3 Mejores Monitores Gaming Calidad Precio para PlayStation 5
Descubre los tres mejores monitores gaming para PS5 que combinan calidad y precio. Mejora tu experiencia de juego hoy mismo.

¿Vale la pena comprar un monitor OLED? Ventajas, desventajas y recomendaciones
Descubre si un monitor OLED es la opción ideal para ti. Aprende sobre sus ventajas, desventajas y recomendaciones de uso en este video.
Most Viewed Summaries

Mastering Inpainting with Stable Diffusion: Fix Mistakes and Enhance Your Images
Learn to fix mistakes and enhance images with Stable Diffusion's inpainting features effectively.

A Comprehensive Guide to Using Stable Diffusion Forge UI
Explore the Stable Diffusion Forge UI, customizable settings, models, and more to enhance your image generation experience.

How to Use ChatGPT to Summarize YouTube Videos Efficiently
Learn how to summarize YouTube videos with ChatGPT in just a few simple steps.

Pamaraan at Patakarang Kolonyal ng mga Espanyol sa Pilipinas
Tuklasin ang mga pamamaraan at patakarang kolonyal ng mga Espanyol sa Pilipinas at ang mga epekto nito sa mga Pilipino.

Pamamaraan at Patakarang Kolonyal ng mga Espanyol sa Pilipinas
Tuklasin ang mga pamamaraan at patakaran ng mga Espanyol sa Pilipinas, at ang epekto nito sa mga Pilipino.