sysGen devCUBE™ Deep Learning Workstation - AMD Threadripper™ PRO / RTX A4000 Edition - consisting of:
Big-Tower Chassis with 4x NVIDIA RTX A4000 GPUs, 2000W 80+ Platinum Power Supply and an AMD Ryzen™ Threadripper™ PRO compatible Mainboard
- CPU, RAM and Storage medium are Configurable!
- AMD Ryzen™ Threadripper™ PRO 5000WX/3000WX
Up to 64 Cores and 128 Threads
- 8x DIMM Slots with 8-channel memory interface
up to 2TB ECC Reg. DDR4-3200 MHz Memory
- 4x RTX A4000 GPUs
- 6x PCI-E 4.0 x16 slots (4x occupied by GPUs)
- 4x M.2 Interface: PCI-E 4.0 x4, RAID 0, 1, 5 & 10
M.2 Form Factor: 2280, 22110; M.2 Key: M-Key
- 2x U.2 Interface (Software RAID 0,1)
- 4x fixed internal 3.5"/2.5" SATA drive bays, 2x hot swap 2.5" SATA drive bays
- 1x 10GBase-T LAN port, 1x 1GbE LAN port (shared with IPMI)
- 1x VGA port (dedicated for IPMI), 7.1 HD Audio
- 2000W Platinum Level Power Supply
- High-Airflow for GPUs or other PCIe Cards
Ready to Go
Complete System incl. all required configured Software:
- Pre-installed Ubuntu LTS
- Pre-installed PyTorch and Tensorflow (based on latest versions available through Anaconda)
- CUDA including corresponding cuDNN
Pre-requisite: Before sysGen can install cuDNN, you have to successfully register on the NVIDIA website as a system developer for "Deep Learning" and sign the NVIDIA "CUDA Software Distribution Agreement for Toolkit cuDNN"
- Tensorflow - Open Source Software Library for Machine Intelligence, latest version Pre-installation and Compilation with Cuda
- CPU AIO Liquid-Cooler
PS: Rack-Conversion-Kit incompatible to AIO!
For the included GPU cards NVIDIA will grant full access to the NVIDIA NGC cloud software. In additon you have full access to our preinstalled sysGen Learning Image (OS Ubuntu 22.04),
all software containers are tuned, tested, certified, and maintained for the top frameworks.