Nvidia H100 8*80GB HGX GDX SXM5 GPU Baseboard HPC Deep Learning Ai Accelerator

C $487,202.85
ApproximatelyRM 1,445,857.17
or Best Offer
Condition:
New
10 available
People want this. 24 people are watching this.
Shipping:
Free Canada Post Tracked Packet - International (Non-US).
Located in: Ottawa, Ontario, Canada
Due to US customs policies, the buyer of this item will need to pay import fees to the shipping carrier prior to delivery. Learn more
Import fees:
Import fees due prior to delivery
Delivery:
Estimated between Tue, 6 Jan and Mon, 12 Jan to 94104
Estimated delivery dates - opens in a new window or tab include seller's handling time, origin ZIP Code, destination ZIP Code and time of acceptance and will depend on shipping service selected and receipt of cleared paymentcleared payment - opens in a new window or tab. Delivery times may vary, especially during peak periods.
Seller ships within 10 days after receiving cleared payment.
Returns:
No returns accepted.
Coverage:
Read item description or contact seller for details. See all detailsSee all details on coverage
(Not eligible for eBay purchase protection programmes)
Seller assumes all responsibility for this listing.
eBay item number:167577727449
Last updated on Dec 04, 2025 07:45:23 MYTView all revisionsView all revisions

Item specifics

Condition
New: A brand-new, unused, unopened, undamaged item in its original packaging (where packaging is ...
Brand
NVIDIA
Connectors
PCIe connectivity, NVLink Switch System, 4 AirMax power connectors up to 6,860 watts of power
Chipset Manufacturer
NVIDIA
Memory Size
80GB per GPU (Total: 640GB across 8 GPUs)
APIs
CUDA, TensorRT, cuDNN, DirectX, Vulkan, OpenCL, NVLink/NVSwitch API, NCCL (NVIDIA Collective Communications Library), OpenACC, MPI (Message Passing Interface)
Features
Tensor Cores (4th Gen), Up to 60 TFLOPS FP64, 1,979 TFLOPS FP8, Optimized for LLMs, Deep Learning & Scientific Computing, 80GB HBM3 per GPU, 640GB total memory, Up to 3.35TB/s memory bandwidth per GP, NVLink 4.0 & NVSwitch Fabric, GPU-to-GPU communication NVLink 4.0 (900GB/s per GPU), full interconnect bandwidth between all 8 GPUs, PCIe Gen5 & High-Speed Interconnects, NVIDIA Magnum IO improved storage & network acceleration, 8x NVIDIA H100 SXM5 GPUs unprecedented AI & HPC acceleration, Optimized for AI Workloads, Transformer Engine for accelerating LLMs like ChatGPT., FP8, FP16, TF32, INT8 & FP64 for various AI & HPC workloads, Accelerates TensorFlow, PyTorch, JAX, Data Center & Cloud Ready, Designed for hyperscale AI training, inference & supercomputing, NVIDIA AI Enterprise and DGX system, Energy Efficiency & Cooling, Passive cooling advanced data center cooling solution, Optimized power consumption for AI workloads
MPN
N/A
Cooling Component Included
Passive Cooling
Memory Type
HBM3
Chipset/GPU Model
H100 8*80GB HGX SXM5 GPU Baseboard
Country of Origin
China
Category

Item description from the seller

About this seller

yudr3m

100% positive feedback323 items sold

Joined Sep 2017
Usually responds within 24 hours
Welcome to Yudr3m, a versatile Canadian based e-commerce superstore that accommodates all your needs with a dedicated team to answer all your questions.Bienvenue chez Yudr3m, un magasin e-commerce ...
See more

Seller feedback (69)

All ratingsselected
Positive
Neutral
Negative
  • e***3 (12)- Feedback left by buyer.
    Past 6 months
    Verified purchase
    Quality & Reliable parachute system. This seller offers very fast shipping way before the timeline shown on eBay on top of an exceptional customer service and technical know how. I 100% Recommend this seller.
  • a***l (24)- Feedback left by buyer.
    Past 6 months
    Verified purchase
    Seller sourced the right item for me and shipped it in record time. Super responsive, works perfectly!
  • e***d (20)- Feedback left by buyer.
    Past month
    Verified purchase
    Everything worked as expected. Boxes arrived in the original package as new stuff. I'm very happy. Price was good and appearance excellent. I highly recommend yudr3m.