rendering server,render farm, hpc

Pro AI calculation server

Active filters

APY AI Zx² Server

It comes standard with enough PCIe 16x slots to accommodate up to ten professional graphics cards or passive compute cards. This performance monster has the ideal profile for those who want raw power for their calculations.

This server is built around two latest Intel Xeon Scalable processors. It is today the main alternative to solutions with SXM2 calculation cards.

APY AI Zx² Ultra Server

This server is the main alternative to the NVIDIA DGX-1 server. The APY AI Zx² Ultra also benefits from the NVIDIA NVLink technology associated with the eight NVIDIA Tesla V100 32GB (SXM2) GPUs integrated on the motherboard.

The NVLink allows the 8 cards to communicate with each other at a speed of 300 Gb / s compared to 32 Gb / s on a PCIe connection. A true concentrate of power and performance.

Note that this server has as standard an Infiniband network connection of 100Gb / s. The APY AI Zx² Ultra will be your main asset for the production of your artificial intelligence solutions.

See all products 


APY AI ZX² G16 Server
- 2x Intel Xeon S4114 processors, 10/20 cores at 2.2 Ghz
- 384 GB of memory
- 1x SSD 512GB (System), 3x SSD 4TB (Data)
- 16x NVIDIA Tesla T4 16GB memory cards GDDR6 memory card 2560 Cores CUDA, 70W, ~ 8.1 TeraFlops (FP32); 320 Turing Tensor
- 2x 10Gbits ports

Voir tous les produits 


The APY One G4 2U 4GPU server is the result of a collaboration between the various APY Group research and development teams. It has been designed and thought for AI. His prime minister is in 2U format and his short depth of 620 mm.

It can be used for visual effects, post production, image capture, graphic virtualization, and artificial intelligence.

See all products 


APY AI Zx² Super Server

Cette infrastructure est-ce qui se fait de mieux aujourd’hui sur le marché de l’IA. Le Super Server est une alternative à la solution NVIDIA DGX-2.

Construit autour d’un châssis de 10U, il embarque seize cartes de calculs NVIDIA Tesla V100 32 Go (SXM2) soit une puissance brute de 2 pétaflops avec ses 81 920 cœurs CUDA et ses 10 240 cœurs Tensor.  Le tout associe 512 Go de mémoire GPU et peut accueillir jusqu’à 3 To de mémoire vive.

Le système complet est interconnecté en Infiniband pour une fluidité sans égale. Le Super Server est tout simplement un titan dans le domaine de l’IA.