Cisco UCSC-GPU-A100-80 | NVIDIA A100 80GB PCIe GPU, 300W Passive, MIG
SKU:UCSC-GPU-A100-80
Stock Status: Enquire
Trade supply only. Server & GPU quote requests need a company name, ABN and work email address. We do not quote Gmail, Outlook or Hotmail addresses, or supply for home use.
Need a quote or bulk pricing? Answered within 20 minutes, Sydney business hours.
Custom Built Servers
Need a different spec, quantity or configuration?
We build configured-to-order refurbished servers from Cisco, Dell, HPE, NVIDIA Lenovo and Supermicro. Tell us what you need and we'll come back with options within 24 hours.
Configure your server now →
Description
Description
The NVIDIA A100 80GB Tensor Core GPU accelerator is a dual-slot PCIe Gen4 data-centre card built on the NVIDIA Ampere architecture. It is designed for AI training and inference, data analytics and high-performance computing, and it suits Cisco UCS C-Series and X-Series deployments that need dense GPU compute.
The card carries 80GB of HBM2e memory with 1,935 GB/s of memory bandwidth and a maximum TDP of 300W. It is a passive, air-cooled design that relies on server chassis airflow. Peak throughput includes 9.7 TFLOPS FP64, 19.5 TFLOPS FP64 Tensor Core, 156 TFLOPS TF32, 312 TFLOPS FP16 and BF16 Tensor Core, and 624 TOPS INT8, with figures doubling when structural sparsity is used. The host interface is PCIe Gen4 x16 at 64 GB/s.
Multi-Instance GPU (MIG) partitions the card into up to seven hardware-isolated instances of 10GB each, so several workloads can share one GPU with guaranteed resources. An NVLink Bridge can pair two PCIe GPUs at 600 GB/s for larger models.
Enquire with XS Network Tech for new or used pricing and availability.
Features
Features
- Third-generation Tensor Cores for AI training and inference
- Multi-Instance GPU partitions the card into up to 7 hardware-isolated instances
- HBM2e memory with over 1.9 TB/s bandwidth
- NVLink Bridge links up to 2 PCIe GPUs at 600 GB/s
- Structural sparsity delivers up to 2X performance for sparse models
- PCIe Gen4 host interface for high-bandwidth server connectivity
Warranty
Warranty
All products sold by XS Network Tech include a 12-month warranty on both new and used items. Our in-house technical team thoroughly tests used hardware prior to sale to ensure enterprise-grade reliability.
All technical data should be verified on the manufacturer data sheets.
View full details
What is the difference between add to cart and request a quote?
Add to Cart lets you purchase in-stock products immediately at listed pricing.
Request a Quote is for customised pricing, volume-based deals, or orders that need a tailored response.
specs-tabs
Collapsible content
Technical Specifications
FAQs
Technical Specifications
- GPU architecture: NVIDIA Ampere
- GPU memory: 80GB HBM2e
- Memory bandwidth: 1,935 GB/s
- Maximum TDP: 300W
- Cooling: Passive, air cooled
- Form factor: PCIe dual-slot
- Host interface: PCIe Gen4 x16 (64 GB/s)
- NVLink Bridge for 2 GPUs: 600 GB/s
- Multi-Instance GPU: Up to 7 MIGs at 10GB
- FP64: 9.7 TFLOPS
- FP64 Tensor Core: 19.5 TFLOPS
- FP32: 19.5 TFLOPS
- TF32: 156 TFLOPS (312 TFLOPS with sparsity)
- BF16 Tensor Core: 312 TFLOPS (624 TFLOPS with sparsity)
- FP16 Tensor Core: 312 TFLOPS (624 TFLOPS with sparsity)
- INT8 Tensor Core: 624 TOPS (1248 TOPS with sparsity)
FAQs
Q: How many workloads can share the GPU with MIG?
A: Multi-Instance GPU splits the A100 80GB into up to seven hardware-isolated instances of 10GB each, so several jobs can run at once with dedicated memory and compute.
Q: How is this card cooled and what power does it draw?
A: It is a passive, air-cooled card with a maximum TDP of 300W. It depends on airflow from the server chassis, so it must be installed in a server designed to cool passive GPUs.
Q: Can two A100 cards be linked together?
A: Yes. An NVLink Bridge connects up to two PCIe GPUs at 600 GB/s, which is useful for larger models that benefit from a faster GPU-to-GPU link than PCIe Gen4.
Q: What host interface does it use?
A: It uses PCIe Gen4 x16, providing 64 GB/s of bandwidth to the host server.
Related Products
-
Vendor:NVIDIANVIDIA 900-54935-2500-000 | Quadro RTX NVLink Bridge, 3-Slot, RTX 5000, 50 GB/s
900-54935-2500-000
Stock Status: Enquire
$106.00 USDVendor:NVIDIAThis NVLink bridge links two identical NVIDIA Quadro RTX 5000 GPUs in a dual-GPU professional workst...$106.00 USDStock Status: Enquire
-
Vendor:NVIDIANVIDIA 900-54934-2500-000 | Quadro RTX NVLink Bridge, 2-Slot, 50GB/s, RTX 5000
900-54934-2500-000
Stock Status: Enquire
$98.00 USDVendor:NVIDIAThe NVIDIA Quadro RTX NVLink 2-Slot Bridge is a GPU-to-GPU interconnect for professional workstation...$98.00 USDStock Status: Enquire
-
Vendor:NVIDIANVIDIA 900-53669-2500-000 | NVLink Bridge for Ampere GPUs, 3-Slot, 112GB/s
900-53669-2500-000
Stock Status: Enquire
$195.00 USDVendor:NVIDIAThis NVIDIA NVLink bridge joins two identical Ampere-generation professional RTX GPUs in a 2-way con...$195.00 USDStock Status: Enquire
-
Vendor:NVIDIANVIDIA 900-53669-0000-000 | RTX A-Series NVLink Bridge, 3-Slot, 2-Way
900-53669-0000-000
Stock Status: Enquire
$199.00 USDVendor:NVIDIAThe NVIDIA RTX A-Series NVLink Bridge links two compatible NVIDIA RTX A-series professional GPUs, su...$199.00 USDStock Status: Enquire
-
Vendor:NVIDIANVIDIA 900-5G153-2220-000 | RTX PRO 5500 Blackwell 84GB GDDR7 GPU, Bulk Pack
900-5G153-2220-000
Stock Status: Enquire
$15,806.00 USDVendor:NVIDIAThe RTX PRO 5500 Blackwell is a professional GPU for rack-mounted workstation deployments, built for...$15,806.00 USDStock Status: Enquire
-
Vendor:CiscoCisco UCSXE-GPU-L4 | NVIDIA L4 24GB GPU, 70W, HHHL, PCIe Gen4 x16
UCSXE-GPU-L4
Stock Status: Enquire
Request QuoteVendor:CiscoThe NVIDIA L4 Tensor Core GPU is a low-profile, single-slot accelerator based on the Ada Lovelace ar...Request QuoteStock Status: Enquire
-
Vendor:CiscoCisco UCSX-NVL2-H200 | UCS X-Series 2-Way NVLink Bridge, 900GB/s, H200 NVL
UCSX-NVL2-H200
Stock Status: Enquire
Request QuoteVendor:CiscoThis 2-way NVIDIA NVLink bridge is built for the Cisco UCS X-Series and links two NVIDIA H200 NVL GP...Request QuoteStock Status: Enquire
-
Vendor:CiscoCisco UCSX-GPU-RTXP6000 | UCS X-Series RTX Pro 6000 96GB GPU, 600W, 2-Slot FHFL
UCSX-GPU-RTXP6000
Stock Status: Enquire
Request QuoteVendor:CiscoThe UCS X-Series RTX Pro 6000 is an NVIDIA RTX PRO 6000 Blackwell Server Edition GPU for the Cisco U...Request QuoteStock Status: Enquire
-
Request a Quote
Looking for competitive pricing? Submit a request, and our team will provide a tailored quote that fits your needs.
-
Contact Us Directly
Call us NowHave a question or need immediate assistance? Call us for expert advice and real-time support.
-
Contact Us Directly
Contact usHave a question or need immediate assistance? Call us for expert advice and real-time support.