£500 less than elsewhere – might be useful for someone who does AI stuff. I suppose it is technically a deal. I have put the heating on as a precaution.
Join JoyPlus here for a free trial to get back £45.99 as Joybuy Points which should be able to get you 48 Mars bars here.
Review
Asus Ascent GX10 review IT Pro
Key Features
The ASUS Ascent GX10 is a compact desktop AI supercomputer built for local AI development, model experimentation, inference, fine-tuning and data-science workloads. Powered by the AMD Radeon Graphics (Desktop Integrated), it combines a 20-core Arm CPU with an integrated NVIDIA Blackwell GPU, 128GB LPDDR5X coherent unified memory and a 1TB NVMe PCIe 4.0 SSD.
The integrated AMD Radeon Graphics (Desktop Integrated) includes fifth-generation Tensor Cores and delivers up to 1 petaFLOP of AI performance using FP4 precision. Its unified-memory architecture enables compatible AI workflows to use the shared 128GB memory pool across CPU and GPU resources, supporting local work with large language models, generative-AI tools and accelerated data-processing applications.
NVIDIA DGX OS provides an AI-focused Linux software environment and access to the NVIDIA AI software stack. The compact 150 150 51mm desktop design is suitable for developers, researchers, data scientists and technical teams who need a local AI-computing system without deploying a conventional rack server.
For high-speed connectivity, the ASUS Ascent GX10 includes 10G Ethernet, NVIDIA ConnectX-7 SmartNIC networking, WiFi 7, Bluetooth 5.4, HDMI 2.1 and four USB-C ports. Three USB-C ports support up to 20Gbps data transfer and DisplayPort 2.1 Alt Mode, while a dedicated USB-C port supports 180W EPR Power Delivery input.
Manufacturer: ASUSTeK Computer Inc., 1F, No. 15, Lide Road, Beitou District, Taipei City 112, Taiwan. Email: emailasus.com.
EU Responsible Person: ASUS Europe B.V., Paasheuvelweg 25, 1105 BP Amsterdam, The Netherlands. Email: ProductSafetyasus.com.
Key Features
NVIDIA GB10 Grace Blackwell Superchip for local AI development.
20-core Arm CPU with 10 Cortex-X925 and 10 Cortex-A725 cores.
Integrated NVIDIA Blackwell GPU with fifth-generation Tensor Cores.
Up to 1 petaFLOP FP4 AI performance.
128GB LPDDR5X coherent unified system memory.
1TB M.2 NVMe PCIe 4.0 SSD.
NVIDIA DGX OS and NVIDIA AI software stack.
10G Ethernet and NVIDIA ConnectX-7 SmartNIC.
WiFi 7 and Bluetooth 5.4 connectivity.
Three USB-C 20Gbps ports with DisplayPort 2.1 Alt Mode.
Dedicated USB-C 180W EPR Power Delivery input.
HDMI 2.1 display output.
Compact 150 150 51mm desktop format.
Supports compatible multi-node AI workflows through high-speed networking.
Important Information
This product is an AI-development computer running NVIDIA DGX OS, not a Windows desktop PC.
It is intended for AI development, inference, model experimentation, technical computing and compatible Linux-based workflows.
The 128GB LPDDR5X unified memory is integrated and is not user-upgradeable.
The system uses one M.2 2242 storage slot compatible storage upgrades are limited to this physical format.
Compatible external displays, USB-C devices, network equipment, cables and software are required to use the corresponding connectivity functions.
AI performance depends on model size, precision, software framework, workload, system configuration and operating conditions.
A compatible USB-C Power Delivery power source is required for operation.
Safety Warning
Use only a compatible USB-C power source that meets the 180W EPR PD 3.1 power-input requirement.
Do not use damaged, modified, loose or overheating power cables, chargers or plugs.
Keep ventilation areas unobstructed during operation do not cover the device or use it on soft surfaces that retain heat.
Keep the product away from liquids, excessive moisture, direct sunlight, naked flames and heat sources.
Disconnect the power supply before cleaning use a soft, dry cloth only.
Do not disassemble, repair or modify the device. Repairs must be carried out by qualified personnel.
Route cables safely to avoid tripping, crushing, pinching or connector damage.
Dispose of the device and electronic accessories through authorised WEEE recycling channels do not dispose of them with household waste.
