New GPU servers with a great NVIDIA® H100 – already in LeaderGPU ®!

A modern, super cool video card that many people are after is already in LeaderGPU!

The NVIDIA H100 is a high-performance graphics processor from NVIDIA designed specifically for the needs of AI and high-performance computing (HPC). The H100 incorporates 4th generation Tensor Cores and a Transformer Engine with FP8 precision to speed up the training of AI models by up to 4 times.

The H100 GPU is based on the NVIDIA Hopper™ architecture, which enables hardware-based security and isolation powered by the new built-in NVIDIA Confidential Computing feature.

Fourth generation Tensor Cores accelerate all precisions (FP64, TF32, FP32, FP16, INT8 and now FP8), helping to reduce memory consumption and improve performance in LLM (large language models).

The card uses the power of 80Gb HBM3 memory. The maximum bandwidth of the H100 PCIe is an impressive 2TB/s! Wow!

The performance of the SXM5 version reaches 33.5 TFLOPS (FP64), 66.9 TFLOPS (FP32), while the PCIe 5.0 version has 25.6 and 51.2 TFLOPS, respectively.

Servers with H100 GPUs are available with monthly, weekly, day and minute billing.

Hurry up to rent these ultra-modern servers! There is already a queue of people waiting for them!

Subscribe to our updates to stay informed about the latest GPU configurations.