The supercomputer Cray/SGI business was the only surviving bit and basically DoD and other TLAs are the only buyers.

I suspect it's mostly the Cray networking topologies (T3D/T3E double-taurus with 2048+ cpus) that kept them relevant. The processors were just plain alphas and later Xeons I think.

MIPS then Itanium then Xeons. A lot of companies rode the Itanic because it looked so high-class. One, good result is you can get an Itanium server on ebay for about $120 if it's a small SGI. I kind of smiled and winced when I noticed that trend.

MIPS then Xeons.

Those were likely former Rackable Systems clusters, which bought and renamed itself SGI. Sadly, I think that all that's left of the old SGI is the name. Even the HPC magic they had I believe originally came from Cray -- another name from a bygone era.

Cray still builds fairly impressive single-image Limux machines. Not as pretty as their forerunners, but impressive on their own.

You're right... I was thinking back to 2000's SGI. It's hard to track of when / where these companies are. SGI buys Cray, SGI spins Cray out to Tera, Rackable buys SGI, HP splits, HPE buys SGI...

It's all crazy.

But yes, Cray still has some impressive gear independently.

They are for sure not single image machines. Each node is running its own kernel, sees only its own resources and is scheduled by the system-wide batch scheduler node. Running across nodes requires MPI to tie things together (tenuously) at the application level.

If they are impressive, it's only because of physical size and aggregate resources available, not because there's the least bit of magic involved. It's the usual Intel+-NVidia CPUs/chipsets/etc. At the current time, it's mostly their legacy proprietary interconnect.

Could I have confused those with, oh the irony, some SGI machine? I remember someone had single-image boxes with up to a thousand CPUs. I assumed it was something derived from Cray's T3E.