SPARKLE Intel Arc Pro B70 32GB GDDR6 Passive Graphic Card

SPARKLE Intel Arc Pro B70 32GB GDDR6 Passive Graphics Card is an enterprise-class workstation GPU powered by Intel Xe2 (Battlemage) architecture. Featuring 32GB ECC-capable GDDR6 memory, 256 Intel XMX AI engines, 367 INT8 TOPS AI performance, PCIe Gen5 x16, and a fanless passive cooling design, it is ideal for AI inference, local LLM deployment, CAD, rendering, virtualization, HPC, and rack-mounted server environments.

Compare

SPARKLE Intel Arc Pro B70 32GB GDDR6 Passive Graphic Card

Features

  • Professional Intel Arc Pro B70 workstation and server GPU.
  • Massive 32GB GDDR6 ECC-capable memory for AI and professional workloads.
  • Powered by the latest Intel Xe2 (Battlemage) architecture.
  • 256 Intel XMX AI Engines accelerate AI inference and machine learning.
  • Delivers up to 367 INT8 TOPS AI performance.
  • High-speed PCIe Gen5 x16 interface.
  • Fanless passive cooling optimized for high-airflow rack servers.
  • 275W Total Board Power for enterprise deployments.
  • Supports SR-IOV virtualization with up to 7 virtual GPUs.
  • Hardware Ray Tracing and DirectX 12 Ultimate support.
  • Hardware AV1 encode/decode for professional media workflows.
  • 4 × DisplayPort 2.1 outputs supporting ultra-high-resolution displays.
  • Enterprise-grade drivers optimized for AI, rendering, CAD, engineering, and virtualization.

Specifications

Product Name SPARKLE Intel Arc Pro B70 32GB GDDR6 Passive Graphic Card
Manufacturer SPARKLE
GPU Intel Arc Pro B70
GPU Architecture Intel Xe2 (Battlemage)
Graphics Memory 32GB GDDR6 (ECC Supported)
Memory Interface 256-bit
Memory Bandwidth Up to 608 GB/s
AI Engines 256 Intel XMX AI Engines
Xe Cores 32
Ray Tracing Units 32
AI Performance Up to 367 INT8 TOPS
PCI Express Interface PCIe 5.0 x16
Cooling Fanless Passive Cooling
Form Factor Dual-Slot
Total Board Power (TBP) 275W
Display Outputs 4 × DisplayPort 2.1
DirectX Support DirectX 12 Ultimate
OpenGL Support OpenGL 4.6
AV1 Hardware Encode/Decode Yes
Virtualization SR-IOV Support (Up to 7 vGPUs)
Target Workloads AI Inference, LLMs, HPC, CAD, Rendering, Virtualization, Enterprise Servers