Hariom Computer

Compact Desktop AI Supercomputer, powered by the NVIDIA® GB10 Grace Blackwell Superchip and NVIDIA DGX™ Spark. Seamless integration and deployment, revolutionizing AI development and research.

• Extreme AI Performance: Powered by NVIDIA® GB10 Grace Blackwell Superchip delivering 1 petaFLOP of AI performance and 128GB memory for 200B model fine-tuning.
• Developer-Optimized Platform: Tailored for AI researchers and developers with full-stack compatibility and streamlined workflows.
• Scalable Architecture: Featuring NVIDIA® NVLink™-C2C for ultra-fast CPU-GPU memory communication and NVIDIA® ConnectX-7 networking to support dual GX10 system stacking, unlocking superior scalability and performance.
• Advanced Thermal Design: Engineered cooling ensures sustained high performance and reliability in an ultra small form factor.
• Minimal Footprint, Maximum Efficiency: High-density compute power in a compact system designed for labs, offices, and edge environments.
• Full Stack AI Solution: The GB10 and NVIDIA AI software stack provide a full stack solution for AI development and deployment.
• ModelASUS Ascent GX10Operating SystemUbuntu LinuxCPUARM v9.2-A CPU (GB10)ChipsetIntegratedGraphicsNVIDIA Blackwell GPU
  (GB10, integrated)Memory128 GB LPDDR5x, unified system memoryStorageIncluded: 2TB M.2 NVMe™ PCIe® 4.0 SSD storage
  Included: 1TB M.2 NVMe™ PCIe® 4.0 SSD storage
  Included: 4TB M.2 NVMe™ PCIe® 5.0 SSD storageWireless Data NetworkAW-EM637 Wi-Fi 7 (Gig+) , Bluetooth 5.4, 2*2LAN10G Gigabit EthernetFront(Side) I/O Ports1 x Power buttonBack I/O Ports3 x USB 3.2 Gen 2x2 Type-C, 20Gbps, alternate mode (DisplayPort 2.1)
  1 x USB 3.2 Gen 2x2 Type-C,with PD in(180W EPR PD3.1 SPEC)
  1 x HDMI 2.1
  1 x NVIDIA® ConnectX®-7 SmartNIC
  1 x 10G LAN
  1 x Kensington LockPower SupplyPower cord
  240W AdapterDimensions (W x D x H)150 x 150 x 51 mm (5.91 x 5.91 x 2.01 inch)Weight1.48kgColorStellar GreyAccessoriesAC Adapter
  Power Cord
  User Manual
  Warranty CardCertificatesBSMI/CB/CE/FCC/UL/CCC/C-Tick/WiFi/RF/VCCI

NVIDIA DGX Spark provides organizations and developers with a powerful, economical experimentation ground for prototype models, freeing up valuable compute resources in their cluster environments better suited for training and deploying production models. Leveraging the NVIDIA AI platform software architecture makes it possible for NVIDIA DGX Spark users to easily move their work from their desktop to DGX Cloud or any accelerated cloud or data center infrastructure, making it easier than ever to prototype, fine-tune, and iterate.

NVIDIA DGX H200

• 8x NVIDIA H200 GPUs with 1,128GBs of Total GPU Memory
  18x NVIDIA NVLink® connections per GPU, 900GB/s of bidirectional GPU-to-GPU bandwidth
  
  
• 4x NVIDIA NVSwitches™
  7.2TB/s of bidirectional GPU-to-GPU bandwidth, 1.5X more than previous generation
  
  
• 10x NVIDIA ConnectX®-7 400Gb/s Network Interface
  1TB/s of peak bidirectional network bandwidth
  
  
• Dual Intel Xeon Platinum 8480C processors, 112 cores total, and 2TB System Memory
  Powerful CPUs for the most intensive AI jobs
  
  
• 30TB NVMe SSD
  High-speed storage for maximum performance

Experience next-level gaming and content creation with the Aorus RTX 5060 Ti. Built on NVIDIA’s latest architecture, this graphics card delivers blazing-fast performance, stunning visuals, and advanced AI-powered features. Its factory-overclocked cores and premium cooling system ensure smooth gameplay even under intense workloads.

Overview

Seven New Chips, One AI Supercomputer

NVIDIA Vera Rubin NVL72 unifies leading-edge technologies from NVIDIA—72 Rubin GPUs, 36 Vera CPUs, ConnectX®-9 SuperNIC™s, and BlueField®-4 DPUs. It scales up intelligence in a rack-scale platform with the NVIDIA NVLink™ 6 switch and scales out with NVIDIA Quantum-X800 InfiniBand and Spectrum-X™ Ethernet to power the AI industrial revolution at scale. When deployed with NVIDIA Groq 3 LPX racks, Vera Rubin NVL72 delivers a new class of inference performance for trillion-parameter models and million-token context.

Vera Rubin NVL72 is built on the third-generation NVIDIA MGX™ NVL72 rack design, offering a seamless transition from prior generations. It delivers AI training with one-fourth the GPUs and AI inference at one-tenth the cost per million tokens versus NVIDIA Blackwell. Featuring cable‑free modular tray designs and support from over 80 MGX ecosystem partners, the rack-scale AI supercomputer delivers world‑class performance with rapid deployment.

Performance

Massive Efficiency Gains in AI Training and InferenceBoosting Training Efficiency

NVIDIA Rubin trains mixture-of-expert (MoE) models with one-fourth the number of GPUs over the NVIDIA Blackwell architecture.

Projected performance subject to change. Number of GPUs based on a 10T MoE model trained on 100T tokens in a fixed timeframe of 1 month.

 

LLM inference performance subject to change. Cost per 1 million tokens based on Kimi-K2-Thinking model using 32K/8K ISL/OSL comparing Blackwell NVL72 and Rubin NVL72.