Hi Zeev Hed; Re: "what the bears and Carl as well are not realizing is that if indeed RDRAM is the best solution for high end systems (this year, as Barrett states), there is no place where DDR will find its entry (the high end is "occupied by RDRAM and mainstream can be well served with SDRAM) and go down its own "learning curve", even if DDR has some advantages (just as did EDO), they must find the correct market opportunity to get in. At low production runs, there is no reason that DDR will be more than 15% cheaper than RDRAM (if at all), testing of high speed parts will be as cumbersome for DDR as for RDRAM, so, unless they have assured themselves "a place under the sun", they are too late to the party and Samsung is going to go laughing all the way to the bank."
Ah, poor Zeev, you are so out of your depth when it comes to the memory market... But a sweet trader. Nice gain there.
There are some basic errors in your post:
(1) The DDR ramp up started last fall, and is doing beautifully. Nvidia, in particular, is shipping vast volumes of GeForce chipsets. Looking forward, the server market is owned by DDR, and RDRAM just lost Intel's mainstream market. Where's the Timna, for instance? The latest news from IDF has Intel producing an SDRAM workstation chipset, that's the last RDRAM stronghold left:
Among the new Intel chip sets is an 860 chip set, which is expected to debut in the first quarter of 2001 to support workstations with SDRAM memory.
semibiznews.com
The real question is how long will it be before RDRAM is squeezed completely out of the PC marketplace, not how long will it be before they take over the market. The only hope the stock has is royalty collections from SDRAM and DDR. RDRAM itself is quite dead in the PC marketplace.
(2) DDR doesn't need nearly the kind of ramp that RDRAM needed to get prices down. This is because RDRAM is revolutionary, DDR is evolutionary. In addition, IBM, Fujitsu and Samsung (and possibly others) already make memory that is configured as SDRAM or DDR SDRAM only at the test stage. RDRAM wafers have to be started 10 weeks or so before they are produced, but DDR can be chosen as an option at the last stage. Unlike RDRAM, DDR doesn't need the high cost testing equipment. Consequently, the volumes in DDR are already available from the memory makers, it is just a matter of chipsets.
(3) You wrote that " testing of high speed parts will be as cumbersome for DDR as for RDRAM". Where did you get an idea like that? PC266 DDR runs at 1/3 the frequency of PC800 RDRAM. Do you really think that tripling the frequency doesn't make it more difficult to test? How could you write something this silly? Rambus PR, perhaps? Would you like to supply a link? The industry has repeatedly been claiming exactly the opposite for several years. Now they have refused to produce RDRAM in volume, or drop the price. Intel is abandoning the technology with all possible speed, and you come up with this kind of statement. Jeez...
Re "Samsung laughing all the way to the bank". Clearly you haven't read the Samsung report on DDR:
inqst.com
You've made your trade, and you made a profit. Now you're out, but why are you spouting this nonsense? You are a great trader, (I think the best swing trader on SI) but your technical knowledge of the memory industry (or memory technology) is quite limited.
Sorry for picking on you while I ignore most of the silly posts on this thread, but I have a lot more respect for your opinions than I have for theirs.
-- Carl |
