Product Details:
NVIDIA A100 PCIE AMPERE GPU ACCELERATOR 40GB HBM2 1555GB/S MEMORY BANDWIDTH PCI-E 4.0 X16 GENERAL PURPOSE GRAPHICS PROCESSING UNIT GPGPU
MANUFACTURER: NVIDIA
PART NUMBER: 900-21001-0000-000
ENGINE SPECS:
ENGINE ARCHITECTURE: AMPERE
CUDA CORES: 6912
TENSOR CORES: 432 ( 3RD GEN )
GPU CLOCK ( BASE ): 765 MHZ
BOOST CLOCK: 1410 MHZ
DOUBLE PRECISION PERFORMANCE: FP32 19.5 TFLOPS ( TERAFLOPS ) / TENSOR FLOAT 32 ( TF32 ) 156 TFLOPS ( TERAFLOPS ) / 312 TFLOPS ( TERAFLOPS )
SINGLE PRECISION PERFORMANCE: FP64 9.7 TFLOPS ( TERAFLOPS ) / FP64 TENSORE CORE 19.5 TFLOPS ( TERAFLOPS )
HALF PRECISION PERFORMANCE: 312 TFLOPS ( TERAFLOPS ) / 624 TFLOPS ( TERAFLOPS )
BFLOAT16: 312 TFLOPS ( TERAFLOPS ) / 624 TFLOPS ( TERAFLOPS )
INTEGER PERFORMANCE: INT8 624 TOPS / 1248 TOP; INT4 1248 TOPS / 2496 TOPS
NVLINK TRANSFER SPEEDS: 600GB/S
MEMORY:
MEMORY CONFIGURATION: 40GB
MEMORY INTERFACE: HBM2
MEMORY INTERFACE WIDTH: 5120 BIT
MEMORY BANDWIDTH: 1.6TB/S ( 1555GB/S )
MEMORY CLOCK: 1215 MHZ
ERROR CORRECTION CODE ( ECC ): YES ( ENABLED BY DEFAULT / CAN BE DISABLED VIA SOFTWARE )
SUPPORT:
BUS SUPPORT: PCI-E 4.0
PHYSICAL BUS INTERFACE: PCI-E 4.0 X16
SUPPORTED TECHNOLOGIES: NVIDIA CUDA CORE TECHNOLOGY, DIRECTCOMPUTE, OPENCL, OPENACC
OPERATING SYSTEMS: MICROSOFT WINDOWS 7, MICROSOFT WINDOWS 8, MICROSOFT WINDOWS 8.1, MICROSOFT WINDOWS 10, MICROSOFT WINDOWS SERVER 2008 R2, MICROSOFT WINDOWS 2016, LINUX ( ENGLISH US / UK )
CONNECTIONS:
POWER CONNECTORS: ( 1 ) ONE 8 PIN AUXILIARY POWER CONNECTOR
ADDITIONAL: ( 1 ) ONE NVLINK INTERFACE ( 3RD GEN )
THERMAL AND POWER SPECS:
MAX POWER CONSUMPTION ( W ): 250W
COOLING SYSTEM:
PASSIVE HEATSINK ( BIDIRECTIONAL AIRFLOW )
FORM FACTOR:
DUAL SLOT, FULL HEIGHT / HIGH PROFILE
DIMENSIONS:
4.375 INCH ( H ) ( MEASUREMENT INCLUDES PCIE INTERFACE ) X 10.5 INCH ( L )
