SXM2 V100 16GB HBM2 Accelerator Card — Volta Tensor Cores, 125 TFLOPS Inference, 5,120 Cores (7.5 TFLOPS FP64)

Built on the Volta architecture, the V100 integrates dedicated tensor cores tuned for matrix-heavy workloads and delivers up to 125 TFLOPS of deep-learning inference performance. Memory and bandwidth: 16GB of HBM2 high-bandwidth memory provides stacked-die throughput, offering significantly greater memory bandwidth and bus width than GDDR5/GDDR6. SXM2 form factor and power delivery: SXM2 is a direct board-to-board slot packaging that connects via a dedicated socket on the motherboard. This design enables higher power delivery and a more stable power supply, allowing the V100 to sustain operation at its Boosting frequency and unlock its full performance potential. General-purpose compute: In addition to its Tensor Cores, the card includes 5,120 traditional cores that provide up to 7.5 TFLOPS of double-precision floating-point performance. Enterprise grade: The V100 is engineered as a professional, data centre-grade GPU built to deliver reliable, sustained high-load computing within server environments.