SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Silicon Graphics, Inc. (SGI)
SGI 91.87+1.8%Dec 5 9:30 AM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: THop who wrote (4963)6/10/1998 5:26:00 AM
From: Alexis Cousein   of 14451
 
First, a little comment -- TPC-D numbers reports should always be complete as per TPC rules, which means that you should also quote throughput numbers.

Secondly, TPC-D numbers, and certainly /$ numbers, do not scale linearly with the number of processors -- you wouldn't see many numbers with more than 1 CPU if they did. Instead of using the 4 CPU number you're using for extrapolating to 8 Pentiums, you could have used the large CPU numbers from e.g. Unisys with single servers that use a large number of Pentiums -- and it doesn't beat the SGI with 12 CPUs.

The purpose of this TPC-D submission is to show that:

-the TPC-D numbers reached are at the top of the heap -- the only submission that beats it is a *huge* NCR TeraData machine.
-the price/performance ratio is competitive.

And there are other sides to the story -- The Origin 2000 scalability story, that removes most of the snags you bump into when you trust other vendors' TPC-D numbers to size a system that is not exactly used to run TPC-D tests ;).

Scaling numbers of CPUs up while expecting the price to scale linearly is dangerous, too. Though the Origin 2000 can be expanded well beyond 8 CPUs without much impact on price linearity, the Sun UE450 chassis used in the TPC-D test can't grow -- replacing it with an UE3000 will have an impact on the figures.
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext