NVIDIA A800 Enterprise 80GB PCIe

NVIDIA A800 Enterprise 80GB PCIe
AI Hardware, Enterprise GPUs
Instock

NVIDIA A800 Enterprise 80GB PCIe

0.0 out of 5 (0)

Two months after it was banned by US government from selling high-performance AI chips to China, NVIDIA has introduced a new A800 GPU designed to bypass those restrictions.

The new NVIDIA A800 is based on the same Ampere microarchitecture as the A100, which was used as the performance baseline by the US government.

Despite its numerically larger model number (the lucky number 8 was probably picked to appeal to the Chinese), this is a detuned part, with slightly reduced performance to meet export control limitations and the 33% reduced NVLINK speed is hardly noticeable. It can only be stacked to 8-GPUs per system, not 16 like the A100. This card is the industry standard and premier value choice for ever demanding high-end AI and CGI workloads.

vipera
Will be delivered to your location via DHL or UPS. Ask an agent if import tariffs apply.

Ships in 10 days from payment. Limited allocations available. 3-year warranty included. All sales final. No returns or cancellations. For volume pricing, consult a live chat agent or call our toll-free number.

Update 04.10.2023: 200 units have just arrived and are now available for immediate purchase. Any further quantities require a 4-6 week lead time.

vipera
Will be delivered to your location via DHL or UPS. Ask an agent if import tariffs apply.

Ships in 10 days from payment. Limited allocations available. 3-year warranty included. All sales final. No returns or cancellations. For volume pricing, consult a live chat agent or call our toll-free number.

Update 04.10.2023: 200 units have just arrived and are now available for immediate purchase. Any further quantities require a 4-6 week lead time.

⭐ Trusted by Customers
What Our Customers Say

Verified reviews from real customers on Trustpilot

The new NVIDIA A800

Despite its numerically larger model number (the lucky number 8 was probably picked to appeal to the Chinese), this is a detuned part, with slightly reduced performance to meet export control limitations and the 33% reduced NVLINK speed is hardly noticeable. It can only be stacked to 8-GPUs per system, not 16 like the A100. This card is the industry standard and premier value choice for ever demanding high-end AI and CGI workloads.

 

    SpecificationsA100
    80GB PCIe    		A800
    80GB PCIe
    FP649.7 TFLOPS
    FP64 Tensor Core19.5 TFLOPS
    FP3219.5 TFLOPS
    Tensor Float 32156 TFLOPS
    BFLOAT 16 Tensor Core312 TFLOPS
    FP16 Tensor Core312 TFLOPS
    INT8 Tensor Core624 TOPS
    GPU Memory80 GB HBM2
    GPU Memory Bandwidth1,935 GB/s
    TDP300 W
    Multi-Instance GPUUp to 7 MIGs @ 10 GB
    InterconnectNVLink : 600 GB/s
    PCIe Gen4 : 64 GB/s
    		NVLink : 400 GB/s
    PCIe Gen4 : 64 GB/s
    Server Options1-8 GPUs

     

    SpecificationsA100
    40GB PCIe    		A800
    40GB PCIe
    FP649.7 TFLOPS
    FP64 Tensor Core19.5 TFLOPS
    FP3219.5 TFLOPS
    Tensor Float 32156 TFLOPS
    BFLOAT 16 Tensor Core312 TFLOPS
    FP16 Tensor Core312 TFLOPS
    INT8 Tensor Core624 TOPS
    GPU Memory40 GB HBM2
    GPU Memory Bandwifth1,555 GB/s
    TDP250 W
    Multi-Instance GPUUp to 7 MIGs @ 10 GB
    InterconnectNVLink : 600 GB/s
    PCIe Gen4 : 64 GB/s    		NVLink : 400 GB/s
    PCIe Gen4 : 64 GB/s
    Server Options1-8 GPUs

No file available

Review this product
Select Rating
Choose File

No reviews available.

No related products