Silicon Investor (SI) -- The First Internet Community

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 -- Ignore unavailable to you. Want to Upgrade?

To: Plaz who wrote (41047)	4/27/2000 2:10:00 PM
From: Bilow	Read Replies (1) \| Respond to of 93625

Hi Plaz; I just looked through the Nvidia card set, and my rule of thumb works admirably: The cards with larger memory memory bandwidths tend to have larger memory sizes, and the proportionality constant is around 120Hz/bit. The actual historical range for all display cards is around 30 to 180Hz/bit, depending on whether the memory is also used for other purposes, etc. If you have a display card that deviates far from this, give me a link. Until then, you haven't done anything as far as finding a counterexample to this rule. You wrote: "In fact, your rule is already broken. Even 1 year old products (Voodoo3 and TNT2) were memory bandwidth limited. It's much worse today." But you didn't calculate any ratios, and, in fact, when I did calculate the ratios for those products, I got numbers quite close to my constant. For example: RIVA TNT 1.8GB/sec 16MB Ratio: 0.009 RIVA TNT2 Ultra 2.9GB/sec 32MB Ratio: 0.011 So where is this exception to my rule? Note that the above chipsets could have been shipped with memory sizes in excess of the ones chosen. But the card makers instead chose to limit the memory to the amount that my rule of thumb would suggest. The reason for this is that the extra memory would have increased the cost of the card but not provided much in the way of increased performance. Sure the cards are bandwidth limited. That is the essence of my rule. It is derived from the principle that displays tend to be bandwidth limited. If they were memory size limited, you wouldn't see the rule followed as well, guys would be designing cards that had way too much bandwidth, but not enough memory. And no, my long post did not come down to "Carl's rule of thumb for display bandwidths," though my rule of thumb is quite healthy. Why don't you go look up the numbers for the Voodoo, and figure out what the ratio is? I'm not going to bother, you obviously didn't look up any numbers for the TNT series. Instead, my post came down to: (1) Display technology is one where the bandwidth increase has been proportional to the density increase (as opposed to regular memory, where bandwidth requirements have not kept up with memory size). This can be thought of as a consequence of Carl's rule for display bandwidths, or, instead, simply a funny historical coincidence. (2) Rambus provided the highest bandwidth/device ratio, and, consequently, was a natural contender for those design wins. (3) Despite this advantage, Rambus lost that market due to other costs of the technology. -- Carl