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: Tenchusatsu who wrote (39762)	4/13/2000 3:52:00 AM
From: Bilow	Read Replies (3) \| Respond to of 93625

Hi Tenchusatsu; SDRAM power calc... A brief history of the power consumption issue with RDRAM... A good long time ago, the wizards at Rambus concluded, quite incorrectly, that their RDRAM memory solution consumed less power than the alternatives. The myth persists even today, despite such inconvenient facts as the fact that an RDRAM solution needs a heat sink: Aavid Thermal Products Inc. has launched cooling solutions for the Rambus in-line memory module (RIMM). Initially configured for desktop computer applications, the Cool Covers family of heat sinks will expand techweb.com By the way, the Rambusites, in a flurry of PR, decided that "heat sink" was too descriptive of what the module does. Consequently, they renamed it to be a "heat spreader". The logic behind this was that since only one RDRAM could be addressed at a time, there could only be one chip that was fully on at a time. Thus while that chip could overheat, the other chips could underheat. So a "heat spreader" evens out the thermal differences between the packages. We should disect this logic a little. It is well known that its granularity is supposed to be a great advantage of RDRAM. You could build a RIMM module with as little as a single RDRAM. This is opposed to the case of SDRAM, where it would require at least 4 SDRAMs to build a DIMM, supposing that x16 SDRAMs were selected. But if you only put one RDRAM on a RIMM, where does the heat spread to? I guess that makes the "heat spreader" a "heat sink". Why doesn't SDRAM require a heat sink, then? Because the heat is already spread among the four (or more) chips of the DIMM. Each gets heated equally, and not by much, so no heat spreader, or heat sink, is required. This was the state of the argument from back this past summer. Now, there are x64 SDRAMs, and these could be used to stuff a single DIMM module. Since the chip is only 22mm long, it will leave a lot of space around it, but you could do it, just like you could make a RIMM module with only one RDRAM. And the question. Would the x64 SDRAM chip require a heat sink? Funny thing, you wait around a little, and technology advances, and voila, the experiment is made. It is a 256Mb sized part, that is, 32MBytes in size: semiconductor.hitachi.com Operating currents are on page 9. Burst access current is 270mA. Even refresh current is only 370mA, though that current is not a steady current, and is, consequently, not significant in power consumption calculations. (More explanation on this subtle point later, it is one that dramreview is ignoring.) So a 64-bit wide SDRAM memory using this particular chip would burn a dainty 0.675W, ignoring the I/O consumption. (We could argue forever about the I/O consumption, but first lets argue forever about the core consumption.) Putting two of the chips on a DIMM brings power consumption up to 1.350W. From the Rambus web site: Idd3: One RDRAM in Read, balance in Active mode: 256MB: 2.787A x 2.63V = 7.33W 64MB: 1.011A x 2.63V = 2.66W rambus.com From the site run by the guy who used to be a big shot at Rambus, and is therefore in a position to give an unbiased view: Module Power Consumption, Max/Typical: SDRAM 11.6W / 6.5W RDRAM 4.6W / 4.1W dramreview.com Now I'll admit that I've been a little unfair in these calculations. But the other side was quite a bit more unfair in the other direction. The best way to test this is to get two machines, one with RDRAM, the other with SDRAM, and with equivalent motherboards and processors (if you could do that), and measure the power consumption. Under that test, I have no doubt that RDRAM gets toasted. -- Carl By the way, a good proportion of the bulls on this thread would take issue with your statement that it is a good trading stock. They prefer it as a buy and hold stock, I believe.