A100 SXM4 GPU Module, 80GB HBM2e Memory, 6912 CUDA Cores, 400W TDP 900-2G506-0210-320/965-2G506-0031-200

A100 SXM4 GPU Module, 80GB HBM2e Memory, 6912 CUDA Cores, 400W TDP 900-2G506-0210-320/965-2G506-0031-200

$7,949.95
Sale price  $7,949.95 Regular price  $7,949.95
Skip to product information
A100 SXM4 GPU Module, 80GB HBM2e Memory, 6912 CUDA Cores, 400W TDP 900-2G506-0210-320/965-2G506-0031-200

A100 SXM4 GPU Module, 80GB HBM2e Memory, 6912 CUDA Cores, 400W TDP 900-2G506-0210-320/965-2G506-0031-200

$7,949.95
Sale price  $7,949.95 Regular price  $7,949.95

Experience unprecedented computing power with the A100 80GB SXM4 Module, featuring NVIDIA's cutting-edge Ampere architecture. This factory-sealed unit delivers exceptional performance with 6,912 CUDA cores and 432 third-generation Tensor cores, achieving up to 624 TFLOPS in FP16 computations. The module comes equipped with 80GB of high-bandwidth HBM2e memory on a 5120-bit bus, providing an impressive 2,039 GB/s bandwidth. With base clock speeds of 1,275 MHz and boost speeds up to 1,410 MHz, this powerhouse is designed for demanding server applications. The SXM4 form factor ensures compatibility with Supermicro, NVIDIA, and Gigabyte servers. Supporting up to 7 MIG partitions and featuring a robust 40MB L2 cache, this module is ideal for data centers and high-performance computing environments. The module operates at approximately 400W TDP, reflecting its substantial processing capabilities.

  • PERFORMANCE: Features 6,912 CUDA cores and 432 third-gen Tensor cores delivering up to 19.5 TFLOPS FP32 performance for demanding AI and compute workloads
  • MEMORY SPECIFICATIONS: Equipped with 80GB of HBM2e memory on a 5120-bit bus, providing massive 2,039 GB/s bandwidth
  • ARCHITECTURE: Built on NVIDIA Ampere GA100 architecture with 40MB L2 cache and clock speeds of 1,275 MHz base to 1,410 MHz boost
  • CONNECTIVITY: Features NVLink technology with 600 GB/s bandwidth for high-speed multi-GPU communication
  • FORM FACTOR: SXM4 module design with 400W TDP, supporting up to 7 MIG partitions for workload optimization

You may also like