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.
Politics : Formerly About Advanced Micro Devices

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Ali Chen who wrote (62582)6/21/1999 3:06:00 AM
From: Tenchusatsu  Read Replies (1) of 1572295
 
<What makes you think that the "north" bridge can't send these messages simultaneously? Remember, every processor has a separate "bus", is not it?>

Doesn't matter. The chipset has to receive the request from the processor before it informs the other processors. It takes time to relay that request. On a shared multiprocessor bus, all processors simply watch the bus. No additional latency.

On the other hand, this problem can be solved using snoop filters on the north bridge. This way, the chipset knows exactly which lines are owned by which processors, so snoop traffic can be reduced tremendously. Of course, a snoop filter adds additional cost to the chipset (128K per K7 w/ 2MB L2 cache, for a total of 1MB of snoop filter for an 8-way K7 system), but it's not all that much compared to the entire cost of a 4 or 8-way K7 server.

<Second, how did you manage to do a research on a non-disclosed-yet system architecture?>

Because the P2P architecture has been done before. Alpha isn't the inventor of P2P, you know.

Tenchusatsu
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext