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 95.57

+0.7%

Nov 28 9:30 AM EST

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next

To: KM who wrote (30631)	9/25/1999 11:36:00 AM
From: John Walliker	Read Replies (3) of 93625

KM, I've been reading all these engineering posts with great interest, even though I don't know a circuit board from a Ouija board. Carl has really made some compelling arguments and I would love to see some refutations if there are any. Zeev? Carl is convinced that he has proved that Rambus is a power hog compared with DDR DRAM. However, his argument was based on several starting assumptions that not everyone agrees with. 1) That Rambus will be 1 generation behind DDR because of larger die size. Therefore he compared the results from a system using 256Mbit DDR chips with 128Mbit Rambus chips. This will have biased the analysis in favour of DDR. 2) He used an obsolete Samsung data sheet for Rambus power consumption data and an advance IBM datasheet for DDR power consumption. The newer Samsung data sheet showed a significantly lower power consumption for the Intel qualified revision A silicon than the older version that Carl used. 3) The Samsung data sheet quoted maximum current consumption, while the IBM data sheet did not. When a value is not explicitly defined as being "maximum" it normally means that it is "typical", or sometimes just a guess based on a few samples. 4) Carl assumed that DDR will work reliably without bus termination. Then he revised his opinion, assuming that the clocks might need termination but not the data, address, device select and read/write lines. 5) The comparison ignored the fact that the input/output circuits of DRDRAM can be operated from a 1.8V supply while the core runs at 2.5V. This difference can reduce the power dissipation of the bus drivers by around 30% without altering the signalling voltage on the bus. 6) In applications where the memory is on the motherboard, such as games, telecommunications devices and many notebook PCs the Rambus specification allows the designer to use a much higher bus characteristic impedance than the 28 ohms specified for RIMMs. This can translate into an additional approximate halving of the Rambus signalling power consumption. Carl suggests that the Rambus designers don't know about worst-case design. A look at the design guides that can be downloaded from the Rambus web site shows quite clearly that tolerances are defined for just about everything. Carl suggests that a peak temperature of 50C might cause timing problems, while acknowledging that it is not really that high for semiconductors. The normal temperature range for commercial grade semiconductors is 0-70C. Industrial grade devices are usually rated up to 85C while military devices are rated to 125C. Some automotive parts are specified for operation up to 150C. Each Rambus chip is regularly temperature calibrated to ensure that the bus drive currrent is correct. There is an over-temperature sensor so that if the die temperature reaches a level at which the manufacturer expects the reliability to be reduced the duty cycle of the chip can be reduced, thus cooling it down. This would only happen under extreme conditions. Carl's arguments may seem compelling but they are built on foundations of sand. He devalues them by adding liberal insults and the suggestion that he alone has a clear understanding of the issues involved. John

Report TOU Violation

Share This Post

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next