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 : Advanced Micro Devices - Moderated (AMD)
AMD 230.23-4.3%Nov 18 3:59 PM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Ali Chen who wrote (4208)8/10/2000 6:14:21 AM
From: BilowRead Replies (1) of 275872
 
Hi Ali Chen; Of course I agree with you that the 1GB/sec bw limit of the FSB is not a bottleneck for real applications.

But the context was a discussion of why there wasn't even much in the way of hyped up synthetic Intel benchmarks that showed 1.6GB/sec RDRAM doing much good. For someone writing those benchmarks, the FSB is a nasty bottleneck. (So they did goofy things with the AGP & PCI buses, if I remember correctly.)

I think there will be a (barely) measurable difference with single processors, about what you are saying, but when they put together 4 processors on the same single width memory channel, DDR should provide a significant difference.

-- Carl

P.S. I put a post here showing (more or less) how to estimate performance improvements assuming linearity in execution time. If you want to assume non linearity, it helps to know that the nonlinear results should be "convex". That is, a system that includes two improvements will run at least as well as the linear approximations based on adding the two improvements suggests. The reason for this is that the nonlinearities usually show up as maximum functions. That is, the actual time to complete some task is the longer of two subtasks rather than the sum.
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext