Hi techreports; Re: "April 1999 isn't really that recent. Over 2 years ago." in regard to Samsung's prediction that 2001 would see 50% market penetration for RDRAM.
If Samsung was wrong about 2001 in 1999, then they are likely to be wrong about 2003 in 2001, LOL! And Samsung says that RDRAM and DDR match each other in 2002. (see #reply-15905449 ).
Re: "If Samsung was wrong about predicting what RDRAM market share would be 2 years later, what makes you think you can predict which standard will win."
You're making an incorrect assumption here. You're assuming that Samsung was actively attempting to make a good faith estimate back in 1999, and is doing that now. That's not the case. You'd think that a full exposure to a lifetime of advertising would convince you to be a bit less ready to take what companies say at face value, LOL!
What Samsung was doing then, and is doing now, is trying to get people to use RDRAM by suggesting that it's going to become popular. All the memory makers do this to one degree or another, and they fool a lot of the media as well. But they don't fool the design engineers. There are some games that the memory makers always play, here's my short list:
(a) They always say that they're going to ramp faster and sell cheaper than they actually ramp and sell a new product. They do this in order to convince design engineers to use the new product, and to convince the other memory makers to not make so much of the stuff that there's a glut. Of course, being long time professionals, we've heard it all before, so we adjust their figures accordingly. One of the things they do is say that they are in "volume production" when what they really mean is that if they get any orders they could produce it in volume.
(b) They never say that they're going to close a facility or reduce production until they've gotten to the point of total desperation and are really going to cut production. I'd say Hyundai's in trouble, LOL!
This is undoubtedly an unfamiliar environment for you. You don't talk to application engineers who come around and tell you under nondisclosure about where Samsung (or Intel) is going with their new products, and so you get to hear the BS that gets put out as press releases. If you're going to base your investing off these, well, best of luck, the track record isn't too good.
Since you don't know who to believe, you really have no idea whether to believe me or to believe Samsung. You know for a fact that I don't give a fig about your investment, but do you really think Samsung cares either, LOL? Hey, people only suspect that I'm paid to bash this company, but there is absolutely no denying that Samsung will make more money if RDRAM ramps than they will if it dies. So who's getting paid to tell lies (or shade the truth) here?
Re: "Intel will support RDRAM in their next version of their IA-64 bit processor (McKinley) & EMC (biggest storage company) uses RDRAM in their products."
Sure, there are going to be products that use RDRAM. The problem for RDRAM is that in order to get cheap (and therefore become universally used), it has to be in large production (and therefore universally used). This is a chicken and egg problem. The reason DDR doesn't have this problem is that the memory makers make DDR on the same fab lines that they make SDRAM on. Unlike the cost of switching from SDRAM to RDRAM, there is almost no cost to switch the lines from SDRAM to DDR. And DDR's already almost as cheap as SDRAM, this despite the fact that SDRAM has dived in price.
Those of us who designed DDR into products are smiling easy, cause we were successful in picking the next cheap high bandwidth memory. The guys who chose RDRAM are designing DDR into the next generation products. (Just like Nintendo dropped DDR for the Dolphin, or PixelFusion dropped RDRAM from the Fuzion150, or Intel replaced the 820 with the 815, or PMC-Sierra replaced RDRAM, etc.)
As Samsung has repeatedly stated (and everybody in the industry knows), RDRAM will only become the cheapest memory, and therefore what there can only be one of, if it can achieve a higher volume than SDRAM, and then only with the 4i version:
"We think RDRAM will be very price-competitive with DDR memory." he said. "If the volumes are high enough, then it may be the same price, and if RDRAM volumes are significantly higher, then it could be even cheaper." -- Geoff Hughes, Samsung senior Vice President of Sales and Marketing.
intel.com
In other words, RDRAM will become cheap once it gets into higher volume than SDRAM and DDR. But there is only one chipset supposed to come out to support 4i. That is not going to be a high volume product. Even Samsung predicts that DDR will equal RDRAM next year, and everybody else predicts that DDR will kick butt. From there, the cost advantage is all DDR's side.
Why do you think that Samsung, IBM, Infineon and Toshiba have working prototypes for DDR-II?
Re: "Intel will support RDRAM in their next version of their IA-64 bit processor (McKinley) & EMC (biggest storage company) uses RDRAM in their products."
The problem here is that Intel is supporting DDR in their mainstream desktop, and that's probably more memory than the high end (which is mostly DDR or SDRAM) can possibly use in McKinley. DDR is fully supported by industry. Here's a list of companies supporting DDR:
Agere (AGRA) #reply-16077012
AMD (AMD) #reply-15993020 #reply-16008917
Avanti (AVNT) #reply-16004116
ATI (ATYT) #reply-16043230
Broadcom (BRCM) #reply-16051365 #reply-16004070
Cisco (CSCO) #reply-16063112
Fairchild Semiconductor (FCS) #reply-16004101
IBM (IBM) #reply-16057590 #reply-16057564 #reply-16057536 #reply-16008910
Integrated Devices Technology (IDTI) #reply-16074618
Intel (INTC) #reply-15922217
Intersil (ISIL) #reply-16077220
C-Cube (LSI-Logic) (LSI) #reply-16077207
Micron (MU) #reply-15310484 #reply-16008910
Microsoft (MSFT) #reply-16010973
National Semiconductor (NSM) #reply-15938222
NeoMagic (NMGC) #reply-16054787
Nvidia (NVDA) #reply-16067811 #reply-15893555
Phillips (PHG) #reply-16053901
PMC Sierra (PMCS) #reply-15934529 #reply-16016483
QuickLogic (QUIK) #reply-16077239
Silicon Graphics (SGI) #reply-16063510
Synopsys (SNPS) #reply-16077378
STMicroelectronics (STM) #reply-16041793 (HDTV)
Sun (SUNW) #reply-15984679
Transmeta (TMTA) #reply-16004008
Trident (TRID) #reply-15962777
Private companies
Acorn Networks (private) #reply-16077190
Actuality Systems (private) #reply-16077228
American Megatrends (private?) #reply-16053734
API Networks (private) #reply-16077298
Clearwater Networks (XStream Logic) (private) #reply-16004092
Circuit Integration Tech., Inc. (Cittek) (private?) #reply-16077099
Denali (private) #reply-16004052
Memec Design Services (Memec) (private $2B) #reply-16077081 #reply-16004032
Palmchip (private) #reply-16077378
PhaseLink Corporation, (PLL ) (private?) #reply-16077085
Zettacom (private) #reply-16063051
Foreign companies
Acer Logic Incorporated, ALi #reply-16009145 #reply-15985315 #reply-15903980
Elpida (Japan) #reply-15980428 #reply-15962808 #reply-15980428 #reply-16008910
Fujitsu (Japan) #reply-15536083
Hyundai (Korea) #reply-15310484 #reply-16008910
Infineon (Germany) #reply-16008910
Mitsubishi (Japan) #reply-16008910
MorethanIP (Germany) #reply-16077073
Mosel-Vitelic (Taiwan) #reply-15939379 #reply-16008910
Nanya (Taiwan) #reply-16008910
Samsung (Korea) #reply-15993020 #reply-15962808 #reply-15947382 #reply-16008910
SiS (Taiwan) #reply-15962777 #reply-15952125 #reply-15952123 #reply-16008917
Toshiba (Japan) #reply-15962808 #reply-15980421
VIA (Taiwan) #reply-16008917
Winbond (Taiwan) #reply-16003997
And that's just what I could find from an easy net search.
Some of those companies are kind of big. Ever hear of Cisco, IBM, or Agere (Lucent), for instance? What all this support for DDR means is that it's going to be around, and around in enough volumes that nothing that Intel can do, even if they completely screw up the 845 and 845B, can save RDRAM. RDRAM has been relegated to a niche memory status. That's why it doesn't track SDRAM pricing.
Thanks for writing, and I'm sorry to hear you're in the red. But as far as the story about RDRAM being the next mainstream memory, you can forget it. The best you can hope for is that Rambus wins some legal battles. And if Rambus did manage to collect royalties on the industry, (particularly the 3.5% on DDR) your shares would be worth hundreds of dollars.
-- Carl |
