NVIDIA DGX Spark™ – Superordenador de escritorio personal con IA – Modelo de escritorio GB10 con chip Grace Blackwell
$180.59
- El rendimiento de un superordenador directamente en tu escritorio, con un diseño compacto y energéticamente eficiente, lo que permite utilizar IA a escala empresarial y computación de alto rendimiento justo donde lo necesitas.
- La potencia de la arquitectura Grace Blackwell, que ofrece hasta 1 petaFLOP de rendimiento de IA para el ajuste fino de modelos locales, la inferencia y el análisis, acelerando el tiempo de obtención de soluciones.
- Diseñado desde cero para crear y ejecutar IA, ofrece una integración perfecta de toda la pila de software de IA de NVIDIA, para que puedas desarrollar localmente e implementar en cualquier lugar.
- NVIDIA DGX Spark te ofrece la libertad de experimentar, crear prototipos e innovar más rápido al ampliar los recursos de portátiles, ordenadores de sobremesa, la nube o los centros de datos. Con más potencia para aprender, crear prototipos, probar e innovar, NVIDIA DGX Spark ofrece un retorno de la inversión excepcional para aumentar la productividad.
- Utiliza NVIDIA DGX Spark para dar rienda suelta a nuevas ideas y experimentar con modelos de gran tamaño (hasta 200 000 millones de parámetros en FP4) directamente en tu ordenador de sobremesa con 128 GB de memoria unificada. Potencia las pruebas, la validación y la iteración rápidas, impulsando la innovación en un entorno seguro y de alto rendimiento.
Acerca de este producto
El rendimiento de un superordenador directamente en tu escritorio, con un diseño compacto y energéticamente eficiente, lo que permite aplicar la IA a escala empresarial y realizar cálculos de alto rendimiento justo donde lo necesitas.
La potencia de la arquitectura Grace Blackwell, que ofrece hasta 1 petaFLOP de rendimiento de IA para el ajuste fino de modelos locales, la inferencia y el análisis, acelerando así el tiempo de obtención de soluciones.
Diseñado desde cero para crear y ejecutar IA, ofrece una integración perfecta de toda la pila de software de IA de NVIDIA, para que puedas desarrollar localmente e implementar en cualquier lugar.
NVIDIA DGX Spark te ofrece la libertad de experimentar, crear prototipos e innovar más rápido al ampliar los recursos de portátiles, ordenadores de sobremesa, la nube o los centros de datos. Con más potencia para aprender, crear prototipos, probar e innovar, NVIDIA DGX Spark ofrece un retorno de la inversión excepcional para aumentar la productividad.
Utiliza NVIDIA DGX Spark para dar rienda suelta a nuevas ideas y experimentar con modelos de gran tamaño (hasta 200 000 millones de parámetros en FP4) directamente en tu ordenador de sobremesa con 128 GB de memoria unificada. Potencia las pruebas, la validación y la iteración rápidas, impulsando la innovación en un entorno seguro y de alto rendimiento.
› Ver más detalles del producto
Información adicional
| Dimensiones | 24.13 × 24.13 × 15.24 cm |
|---|
por Victor Williams
Actually, I’m very impressed with this item. It works as deign and I plan to re-buy this in the future when they fix the thermal issue soon as I got everything set up and it loaded and working just the way I wanted it. I ran into this thermal problem where it kept cutting off without notice Crashing without notice that I was running it broke my heart to send it back because I really wanted it. I will buy this again in the future when they fix the thermal issue.
I have an issue with the seller. The item kept overheating and I was forced to return it, but the seller charged me $1000 restocking fee so that tells me that they’re gonna put a broken item back on the shelf to resell it to somebody else and charge them a restocking fee if you really want the item buy it if Amazon is selling it or buy it from Nvidia do not under no circumstances buy from Micro Center because they won’t treat you right
por behnam
The item is buggy, mine was restarting every 30 mins due to Operating system issue.
It runs only on ARM operating system and you cannot much change to their OS.
Most of their drivers are not open source
por ooTo
I am pretty knowledgeable with Linux terminal/command prompts which makes it very user friendly for my skill set. I usually SSH in from any of my other computers. For what I don’t know, I use different LLM for helping solve the task that I work on.
I am currently using at as a hobby dev project on my home lab. Its definitely been a great asset with helping make other agents for other task.
I have very large LLM running locally. One of them is helping with discover more about my business and better ways I can run it.
Its very quiet very fast.
If you are considering ordering one, if you don’t have another computer you can login via phone to setup your connection to wifi. It doesn’t come with usb-c keyboard or mouse. If you have old USB keyboard and mouse, you will need USB to usb-c converter, if you don’t know how to ssh from another computer to the DGX Spark
por Scott M Lawrence
This uses the Blackwell GB10 (SM 121) architecture. Currently, mainstream PyTorch stable binaries lack native support for this specific instruction set, so you may need to use NVIDIA’s NGC Docker containers or manually compile from source to enable GPU acceleration. Keep this in mind if your workflow requires a bare-metal installation.
por P. Martin
Excellent for local LLM research.
It gave me a shock when it appeared not to boot up after first setting it up. It turned out it just needed some time and started up after a while.
That was not at all obvious though. It’s basically silent and it doesn’t have any lights to show that it’s even on. I feel like nvidia could at least have forked out for an off/on light on the case.
por Daniel Connor
The chip is fast but the memory bandwidth kills performance. It’s about 2-3x slower than my M3 Ultra for PyTorch AI workloads. Basically the memory bandwidth difference. That’s really disappointing.
Out-of-the-box PyTorch or CUDA libraries may:
– fall back to older kernels (e.g., Hopper SM 8.9)
– fail to recognize new tensor core formats (FP8/FP4)
– miss optimized paths for sparsity or unified memory
– mis-handle Arm64 optimizations on Grace CPU
por Kirill Keker
Depending on the purpose for which you buy the device. But it works well.
por Ahiro
To the embarrassment of NVIDIA engineers, the WiFi drivers weren’t working, and the system wouldn’t even finish the initial boot process. I had to boot from a USB stick, set a root password, and fix everything myself. It’s a total disgrace.
por D
It is speedy, convenient and easy to install