distributed memory machine
- It was a fully distributed memory machine using a 3D torus topology interconnection network.
T3E after SGI
After Cray Research was acquired by Silicon Graphics in 1997, development of future Alpha-based systems was stopped and the T3E was eventually discontinued.
This isn't strictly true: the T3E got some speed bumps during the SGI years, with the -1200E coming out in 1998, and the final model (the -1350) didn't appear until 2000, after SGI divested itself of Cray. Letdorf (talk) 10:57, 24 February 2009 (UTC).
- Yes, you are right. My edit was supposed to convey Cray under SGI didn't develop any new systems - such as systems using the 21264. My understanding is that there were improvements to the existing T3E design, but they were further developments. I'll clarify my edits. Rilak (talk) 03:21, 25 February 2009 (UTC)
- It's definitely an improvement. As for a ref, how about this one? Or page 7 of this? Letdorf (talk) 13:06, 26 February 2009 (UTC).
- Sorry, that second links should have been this one (PDF page 7, numbered page 6). The SN1, of course, became the Origin 3000. Letdorf (talk) 13:08, 4 March 2009 (UTC).
- Hmm... except that the Origin 3000 was introduced in ~2000. In 1998, was not SGI still on the Origin 2000 with the "new" 250 MHz R10000s? And you previously mentioned that in 2000, SGI divested Cray, which would mean the final piece of the T3E story ends with Cray/Tera, right? Rilak (talk) 04:03, 5 March 2009 (UTC)