A US supercomputer with 8,000 Intel Xeon CPUs and 300TB of RAM is being auctioned — 160th most powerful computer in the world has some maintenance issues though and will cost thousands per day to run

[[{“value”:”

The Cheyenne supercomputer, based at the NCAR-Wyoming Supercomputing Center (NWSC) in Cheyenne, Wyoming, was ranked as the 20th most powerful computer in the world in 2016 – but now it’s up for sale through the US General Services Administration (GSA).

By November 2023, the 5.34-petaflops system’s ranking had slipped to 160th in the world, but it’s still a monster, able to carry out 5.34 quadrillion calculations per second. It has been put to a number of noteworthy purposes in the past, including studying weather phenomena and predicting natural disasters.

The Cheyenne Supercomputer is a monster installation made up of SGI ICE XA modules which comprises of 14 E-Cells, weighing 1,500 lbs each, and 28 water-cooled E-Racks. There are 8,064 Intel “Broadwell” Xeon processors (18-core 2.3GHz E5-2697v4) with a total count of 145,152 cores. In terms of memory, it has 313,344GB of DDR4-2400 ECC single-rank memory and 224 IB Switches. The supercomputer also comes with two air-cooled management racks, each featuring 26 1U servers.

“Repairable”

Howeer, potential buyers do need to be aware of a few issues. Firstly, the unit doesn’t come with fiber optic and CAT5/6 cabling, although the internal DAC cables within each cell are provided and will be “meticulously labeled and packaged in boxes”, and it comes with previously used PGW coolant fluid (around10 gallons per E-cell). It will need to be collected by a professional moving company and the purchaser “assumes responsibility for transferring the racks from the facility onto trucks using their equipment.”

A major red flag is that the supercomputer is worryingly listed as “repairable”. The auction page states “the system is currently experiencing maintenance limitations due to faulty quick disconnects causing water spray. Given the expense and downtime associated with rectifying this issue in the last six months of operation, it’s deemed more detrimental than the anticipated failure rate of compute nodes.

Approximately 1% of nodes experienced failure during this period, primarily attributed to DIMMs with ECC errors, which will remain unrepaired. Additionally, the system will undergo coolant drainage.”

With a couple of days to go before the auction ends, bidding currently stands at $50,085 with the reserve not yet met. Should you wish to buy a piece of supercomputer history, and have the deep pockets required to get it up and running, and the space to house it, you can put in a bid here.