Rambus (RMBS) - Eagle or Penguin Message Board

STOCKTALK

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

RMBS 95.26

+3.1%

Nov 14 9:30 AM EST

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next

To: richard surckla who wrote (43145)	5/29/2000 8:46:00 PM
From: jim kelley	Read Replies (3) of 93625

Richard, Re: Tom's Biased Tests QMC and the other "memory latency tests" cynically designed by Tom's boy are selected and designed to exploit the "initial latency" penalty of RDRAM. They are trying to fool "mom" and "pop" as BILOW says. RDRAM has slightly higher "intitial latency" and much lower "average latency" than SDRAM or DDR DRAM. The use of "random data accesses" to main memory as the dominant form of access penalizes RDRAM and is in fact unrealistic. It means that essentially every access is made into a different RDRAM page and that that page has to be enabled and the initial penalty is always incurred. Most applications programs have both "program and data locality". It appears that some of Tom's programs are a small enough set of instructions to be contained entirely in the cache and the data to be referenced is randomly scattered throughout memory. Further, "cache memory" would not work at all if both program and data were randomly scattered through main memory. This explains in part why the programs sometimes mysteriously work better when the cache is turned off. Typically, a instruction cache miss will produce a 64 bit read. That is four 16 bit read operations to a single channel RDRAM. Thus, ordinarily we will obtain a latency that is the average of the four memory reads from contiguous page locations. ARRAY operations can be efficiently performed with RDRAM memory in a way which will result in much higher performance than corresponding optimized SDRAM operations by arranging the array elements in memory in such a way as to take advantage of the much lower average latency. Essentially, what Tom has done here is heavily eight the benchmark tests to highlight the known initial latency penalty. He uses this "initial latency" penalty to wipe out the bandwidth advantage that RDRAM has over SDRAM with normal programs. So what Tom has done is cynically repeal the laws of "program locality" and "data locality" which have guided the develop of computer memory architectures for the last 30 years. He has no shame and a firm belief in the ignorance of the masses. It just keeps getting better all the time!

Report TOU Violation

Share This Post

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next