Cray XMT

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search

The Cray XMT (codenamed Eldorado) is the third generation of the Cray MTA supercomputer architecture originally developed by Tera. The earlier generations were called the Cray MTA and the Cray MTA-2.[1] The XMT makes the MTA's multithreaded processors, now dubbed Threadstorm, compatible with the 1207-pin Socket F used by AMD Opteron processors.[2] The Threadstorm processors are plugged into systems otherwise identical to the Cray XT4.

The primary advantage of these processors is that their performance is not constrained by the memory access time. In a simplified model, at each clock cycle an instruction from one of the threads is executed and another memory request is queued with the understanding that by the time the next round of execution is ready the requested data has arrived.[3] This is contrast to many conventional architectures which stall on memory access. The architecture excels in data walking schemes where subsequent memory access cannot be easily predicted and thus wouldn't be well suited to a conventional cache model.[4]

The Threadstorm processors only execute user code, on top of a simple BSD Unix-based microkernel called MTX; system I/O is performed by Opteron processors running Linux.[4] This third generation MTA system improves clock speed from 220 MHz to 500 MHz, the maximal processor count from 256 to 8192, and maximum memory to 512 TB.

The architecture has been reworked in the Threadstorm4 CPU,[5] which is used by the XMTs technological successor: the Cray Urika-GD. Currently, the future of Threadstorm is unclear due to competition from commodity processors such as Intel's Xeon.[6]

See also[edit]


  1. ^ "Cray History". Cray Inc. Archived from the original on July 12, 2014. Retrieved August 19, 2014.
  2. ^ Timothy Prickett Morgan (February 20, 2011). "Swiss boffins go nuts for Cray supers". The Register. Retrieved February 14, 2015.
  3. ^ Nieplocha J, Marquez A, Petrini F, Chavarria-Miranda D (2007). "Unconventional Architectures for High-Throughput Sciences" (PDF). SciDAC Review. Pacific Northwest National Laboratory (5, Fall 2007): 46–50. Archived from the original (PDF) on February 14, 2015. Retrieved February 14, 2015.
  4. ^ a b "Why is uRiKA So Fast on Graph-Oriented Queries?". Cray, Inc. November 14, 2012. Archived from the original on February 14, 2015. Retrieved February 14, 2015.
  5. ^ Kopser A, Vollrath D (May 2011). Overview of the Next Generation Cray XMT (PDF). 53rd Cray User Group meeting, CUG 2011. Fairbanks, Alaska. Retrieved February 14, 2015.
  6. ^ "Cray CTO Connects The Dots On Future Interconnects". The Next Platform. 8 January 2016. Retrieved 2 May 2016. Steve Scott: You can do it just great with a Xeon. We are not planning on doing another ThreadStorm processor. But it does take some software technology that comes out of the ThreadStorm legacy.

External links[edit]