On Wednesday 28 May 2008, Ashok Gautham J. wrote: > On Sun, May 25, 2008 at 11:12 PM, Arun Khan <[EMAIL PROTECTED]> wrote: > > On the topic of "mammoth rams", circa 2003 SGI came out with a > > system, which in full config had 384GB RAM. > > When you have such huge RAMS, will the cache transfers be efficient > since a single miss will mean searching the entire RAM again... Al > this while, We had P*(Speed of cache retrieval) + (1-p)*(Speed of RAM > retrieval). When hit rate -> 1, the entire thing speeds up. But here, > the time to locate on the RAM might be significant too right?
IIRC, these systems were enterprise class in the same league as IBM's high end. SGI would be the best people to answer the question. Technology wise, their reputation is solid. -- Arun Khan _______________________________________________ To unsubscribe, email [EMAIL PROTECTED] with "unsubscribe <password> <address>" in the subject or body of the message. http://www.ae.iitm.ac.in/mailman/listinfo/ilugc
