view all server options


With plenty of RAM and cores, plus a Nvidia P4 GPU, this box is perfect for executing models, neural nets and matching engines.

One day we were looking at our "king of the hill" server (the m2.xlarge) and we suddenly knew how to make it even more badass: turn it into an inference monster by adding an Nvidia P4 GPU.  Bingo!

So take those trained models, and run them lightning fast on some accelerated bare metal.

Hardware Specs

  • 28 Physical Cores @ 2.2 GHz
    (2 x Xeon Gold 5120)
  • 384 GB of DDR4 ECC RAM
  • 2 × 120 GB SSD boot
  • 3.2 TB of NVMe Flash
  • 1x Nvidia P4
  • 20Gbps Bonded Network
    (2 × 10Gbps w/ LACP)

OS Support

  • All
  • Popular
  • Licensed
  • VMWare/ESXi

RedHat Enterprise Linux 7

Windows 2019 Standard

VMware ESXi 6.5

FreeBSD 12.1

NixOS 19.03

“With Packet’s developer-friendly bare metal, we’re able to take our Curiosity IoT platform to any city in the United States in 90 days or less.  This is simply unheard of.”

Ivo Rook | SVP of IoT @ Sprint