- Model:VCNA800-PB
- 40GB HBM2 memory with 1555.2 GB/s bandwidth
- 6912 CUDA cores, 432 third-generation Tensor Cores
- PCIe 4.0 x16, NVLink support with 400 GB/s
- 240W, dual-slot, active thermal solution
- 3 Years Warranty from PNY
Overview
The NVIDIA A800 40GB Active GPU accelerates data science, AI, and HPC workflows with 6912 CUDA cores and 432 third-generation Tensor Cores to maximize AI performance and ultra-fast inference capabilities. Powered by the NVIDIA Ampere architecture, it delivers 9.7 TFLOPS of FP64 performance and 19.5 TFLOPS of FP32 performance for demanding computational tasks.
With 40GB of high-speed HBM2 memory and 1555.2 GB/s of memory bandwidth, the A800 40GB Active provides the extreme bandwidth and on-package memory needed to accelerate the most compute-intensive AI models. Third-generation NVLink technology enables GPU-to-GPU data transfers at 400 GB/s bidirectional bandwidth, allowing two cards to be connected for 80GB of effective memory and doubled processing power.
The card supports Multi-Instance GPU (MIG) technology, allowing it to be partitioned into up to seven independent instances for secure, isolated multi-tenancy. Each A800 40GB Active comes with a three-year subscription to NVIDIA AI Enterprise, an end-to-end software platform with enterprise security, stability, manageability, and support including 100+ AI frameworks, libraries, and tools.
Ideal for data science and analytics, AI training and inference, HPC and engineering simulation, generative AI, and life sciences applications, the A800 40GB Active is designed to be paired with NVIDIA RTX-accelerated GPUs for display functionality.
- NVIDIA Ampere architecture with 6912 CUDA cores
- 432 third-generation Tensor Cores with up to 2x performance improvement
- 40GB HBM2 memory with 1555.2 GB/s bandwidth
- Third-generation NVLink with 400 GB/s bidirectional bandwidth
- Multi-Instance GPU (MIG) support for up to 7 isolated instances
- PCIe 4.0 x16 interface
- Active thermal solution, 240W maximum power consumption
- NVIDIA AI Enterprise software subscription included
Specifications
- PNY Part Number
- VCNA800-PB
- GPU Memory
- 40GB HBM2
- Memory Interface
- 5,120-bit
- Memory Bandwidth
- 1.5 TB/s
- CUDA® Cores
- 6,912
- Tensor Cores
- 432
- Double-Precision Performance
- 9.7 TFLOPS
- Single-Precision Performance
- 19.5 TFLOPS
- Peak Tensor Performance
- 623.8 TFLOPS
- Multi-Instance GPU
- Up to 7 MIG instances @ 5GB
- NVIDIA NVLink
- Yes
- NVLink Bandwidth
- 400GB/s
- Graphics Bus
- PCIe 4.0 x 16
- Power Consumption
- 240W
- Thermal
- Active
- Form Factor
- 4.4" H x 10.5" L, dual slot
- Display Capability1
- -
- Category
- Relative Performance
- Quadro GV100
- 1.0
- A800 40GB Active
- 1.7
- Category
- Relative Performance (GTC)
- Quadro GV100
- 1.0
- A800 40GB Active
- 1.9
- Relative Performance
- BERT - Large (Relative speedup)
- Quadro GV100
- 0.95
- A800 40GB Active
- 4.2
- Relative Performance
- BERT - Large
- Quadro GV100
- 1.0
- A800 40GB Active
- 1.9
Warranty
3 Years manufacturer warranty from PNY. Warranty service is provided by the brand's authorised service centre under the brand's own terms; keep the original invoice and the serial number for any claim, and we will guide you through the process. Details are in our Warranty Policy.