The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
GUNNIR demonstrated a compact workstation with two Intel Arc Pro B70 graphics cards, each carrying 32GB of ECC memory. That adds up to 64GB of graphics memory across the system, but it is not a single shared 64GB pool. The larger concept uses liquid cooling and an Intel Core Ultra 200HX processor; a smaller, single-GPU concept uses a vapor-chamber cooler. Neither concept has an announced retail price or shipping date.
What does “64GB of VRAM” mean on this workstation?
Each Arc Pro B70 has its own 32GB of GDDR6 ECC memory. In the two-card concept, the arithmetic total is 64GB, but each GPU still has a separate 32GB memory allocation. The system does not present applications with one automatically combined 64GB graphics-memory pool.
Software can distribute work across both cards when it supports multi-GPU execution. Intel’s oneAPI is relevant to that kind of development, but using two GPUs does not make every application able to load or process a model as if it had access to one 64GB card. Whether a workload can use both devices—and how it divides the work—depends on the software.
What did GUNNIR demonstrate?
At Intel Connection 2026 in China, GUNNIR showed two compact workstation concepts built around Intel Core Ultra 200HX processors and Intel Arc Pro B70 graphics. The larger concept has two B70 cards and liquid cooling; the smaller concept uses the same GPU family with a vapor-chamber cooler. Both demonstrations are Intel-only: no AMD processor or graphics configuration was reported.
#1 Best Overall
- A M D R9-9950X3D2 4.3GHz 16 core | 256GB DDR5 RAM
- N V I D I A - G e F o r c e 2X5090 64 GB | 1600W Power Supply
- 360mm Liquid Cooler | 8 TB NVMe SSD Boot Drive
- Ready to work, preloaded with Windows 11 Pro and the latest drivers
- Custom built Dual GPU AI Workstation, professional cable management, fully tested
| Concept or component | What was reported |
|---|---|
| Larger workstation concept | Two Arc Pro B70 GPUs, 64GB combined capacity across separate cards, and liquid cooling |
| Smaller workstation concept | Arc Pro B70 GPU family and vapor-chamber cooling; the demonstration report does not establish it as the dual-GPU system |
| Processor platform | Intel Core Ultra 200HX in both concepts |
| AMD option | None reported for these demonstrations |
What are the Arc Pro B70’s published specifications?
Intel lists these figures for the Arc Pro B70. They describe the GPU, not a measured result for GUNNIR’s complete workstation.
| Specification | Intel’s published figure |
|---|---|
| Graphics memory | 32GB GDDR6 ECC per GPU |
| Xe2 cores | 32 |
| XMX engines | 256 |
| INT8 performance | 367 TOPS |
| Memory bandwidth | 608GB/s |
| Interface | PCIe Gen 5 x16 |
| Partner-card power range | 160W–290W; Intel’s reference card is 230W |
Intel describes the B70 as suited to local inferencing and scalable multi-GPU deployments. Those descriptions indicate intended workloads, not proof that a particular model or application will run across both cards. Intel says the card’s 32GB of graphics memory can enable larger AI models to run at higher precision; on this two-card concept, the memory remains divided between the GPUs, so software support for distributing a model or workload matters.
Rank #2
- A M D R9-9950X 4.3GHz 16 core | 256GB DDR5 RAM
- N V I D I A - G e F o r c e 2X5090 64 GB | 1600W Power Supply
- 360mm Liquid Cooler | 8 TB NVMe SSD Boot Drive
- Ready to work, preloaded with Windows 11 Pro and the latest drivers
- Custom built Dual GPU AI Workstation, professional cable management, fully tested
Can it run local LLMs?
Potentially, if the model-serving software supports Intel Arc and can make effective use of multiple GPUs when a model or workload exceeds one card’s available memory. A model that fits on one B70 may be able to use that GPU’s 32GB without involving the second card. A workload requiring more than 32GB on a single device cannot rely on the headline 64GB total alone: it needs software that can partition or distribute the work across the two separate GPUs.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallThe available reporting does not include a tested LLM, supported model size, tokens-per-second result, or independent benchmark of the complete GUNNIR system. Intel’s TOPS and bandwidth figures are component specifications, not a guarantee of real-world model performance or multi-GPU scaling.
Rank #3
- 【High-Performance APU】The MS-S1 MAX features an AMD Ryzen AI Max+ 395 APU, integrating a Zen 5 architecture CPU (up to 5.1GHz, 16C/32T, 64M L3 Cache), an RDNA 3.5 GPU, and an NPU (50 TOPS). The total system output is 126 TOPS. It provides powerful parallel computing capabilities for demanding AI workflows. It is ideal for running local LLMs, multimodal models, and computationally intensive tasks
- 【128GB UMA Memory】Equipped with up to 128GB of LPDDR5x-8000MT/s unified memory, it enables the CPU and GPU to access a shared, high-bandwidth memory pool with extremely low latency. Ideal for large-scale AI inference, 3D workloads, and complex timelines in video editing. It eliminates traditional VRAM bottlenecks, ensuring smoother data transfer during high-intensity computations. The UMA design maximizes performance stability under high loads
- 【Flexible Expansion】The MS-S1 MAX features USB4 V2 (up to 80Gbps), dual 10GbE LAN, HDMI 2.1 (up to 8K60), a full-length PCIe x16 expansion slot, and dual M.2 slots supporting up to 16TB RAID 0/1. Wi-Fi 7 provides stronger signal coverage and a more stable wireless experience. The slide-out design facilitates upgrades and maintenance. It easily adapts to personal, studio, or rack-mount enterprise environments
- 【High-Efficiency Cooling System】Utilizing an aerospace-grade aluminum alloy chassis, copper base plate, six heat pipes, dual turbine fans, and advanced PCM thermal conductive material, it maintains stable cooling performance even under continuous load. This system supports 130W continuous power and 160W peak power operation, with a built-in 320W power supply. It boasts multiple global certifications including CCC, FCC, UL, CE, and UKCA, ensuring stable and reliable operation in various environments
- 【Cluster Design】Two MS-S1 MAX units can be configured as a dual-unit cluster to run a large 235B Q4 model locally, achieving an output speed of 10.87 tok/s. Supporting 2U rack deployment, multiple MS-S1 MAX units can be cascaded into a distributed cluster to create a high-efficiency AI computing center. A cluster of four MS-S1 MAX units successfully ran a DeepSeek-R1 671B Q4 large model. A reserved cluster power-on interface allows for unified start-up and shutdown
Can you buy the dual-GPU workstation?
The complete dual-B70 machine was shown as a demonstration. The available coverage gives no retail price, shipping date, or release window, so it should not be treated as an announced product ready to order.
There is a more concrete GUNNIR product to investigate: VideoCardz reports that GUNNIR sells an Arc Pro B70 TF 32G workstation card with 32GB ECC memory and a blower cooler intended for continuous workstation use. That is a graphics card, not the demonstrated dual-GPU system, and a separate card purchase does not establish the cost or compatibility of a complete build.
Rank #4
- A M D R9-9900X 4.4GHz 12 core | 256GB DDR5 RAM
- N V I D I A - G e F o r c e 2X5090 64 GB | 1600W Power Supply
- 360mm Liquid Cooler | 8 TB NVMe SSD Boot Drive
- Ready to work, preloaded with Windows 11 Pro and the latest drivers
- Custom built Dual GPU AI Workstation, professional cable management, fully tested
Intel’s separately sold Intel-branded Arc Pro B70 had a suggested starting price of $949 in Intel’s 2026 pricing information. Partner-card prices vary by implementation; that figure is not the price of GUNNIR’s workstation or necessarily of GUNNIR’s card.
Quick Recap
Best Value
- AI Performance: 767 AI TOPS
- OC mode: 2632 MHz (OC mode)/ 2602 MHz (Default mode)
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Axial-tech fan design features a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- A 2.5-slot design maximizes compatibility and cooling efficiency for superior performance in small chassis
What to weigh against a conventional single-GPU workstation
- Memory capacity: Two cards provide 64GB in aggregate, but each has 32GB of its own memory. A single-GPU workstation with less total memory may be simpler for software that expects one device; the GUNNIR concept’s capacity is useful only when software can use the second device appropriately.
- Software support: Multi-GPU execution is an application and framework requirement, not an automatic property of a two-card PC. Confirm support for Intel GPUs and the specific multi-GPU workload before treating both cards as usable for a given task.
- Cooling and power: The larger concept’s liquid cooling is part of a two-GPU design, while Intel lists a 160W–290W partner power range per B70 card. The demonstration reporting does not specify the complete system’s power draw, cooler capacity, or acoustic behavior.
- Size and platform: GUNNIR presented compact concepts using Core Ultra 200HX processors, but no chassis dimensions or full platform details were reported. “Compact” does not establish that the system is smaller than a particular workstation or suitable for a specific desk or rack.
- Price and availability: There is no announced price or shipping date for the complete concept, and no published independent benchmark to compare its performance with a conventional workstation.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
