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 : Rambus (RMBS) - Eagle or Penguin -- Ignore unavailable to you. Want to Upgrade?


To: REH who wrote (5838)7/22/1998 1:50:00 PM
From: mauser96  Read Replies (1) | Respond to of 93625
 
Great series of posts. << As a general rule, once you satisfy the CPU's maximum burst bandwidth rate, it is difficult to improve performance by adding more bandwidth.>> Making the assumptions that by the year 2000, Intel will be producing 1000MHz microprocessors, and that Rambus will have improved it's RDRAM speed to 1000 MHz also, then it would seem likely that latency improvements will be needed. Rambus management has said it will probably take to 2000 for more than 20% of Intel powered computers to be using the RDRAM. For longer term advancement of RDRAM use, the question arises whether latency can be decreased on RDRAM with no more difficulty or increased cost than would be expected with other types of memory. The ESDRAM solution doesn't sound too practical.
Another question is that considering the precipitous fall in regular DRAM prices, and the increased need for large amounts of memory, is RDRAM going to be cost effective for anything but very high end computers?



To: REH who wrote (5838)7/23/1998 5:05:00 PM
From: Joseph  Read Replies (1) | Respond to of 93625
 
REH, improving bandwidth can hide latency. Latency is a big problem only when all memory accesses are random. Then, the cache is of no use, and all accesses are delayed by the time it takes to get data from the memory. However, if the accesses are very regular, the hit ratio is very high. In this situation, filling the cache fast becomes the most important issue, thereby necessitating high memory bandwidth. In many compute-intensive jobs, regular access patterns are common, and an RDRAM will help significantly in this regard.

Joseph