Skip to main content
Message Font: Serif | Sans-Serif
 
No. of Recommendations: 20
I am taking this Rule-Breaker seminar, so I thought I would run Rambus through the gauntlet. Just for fun.

I actually didn't even consider this stuff in my decision to buy it. I think it fits the RB profile nicely, though. Anyway, here it is, for your reading enjoyment.


The company should be a top dog and a first mover in an important, emerging field. In other words, being top dog in the left-handed scissors industry isn't enough. The left-handed scissors industry just isn't really "emerging" -- ya know? It's pretty mature -- and it ain't going anywhere in the near or distant future. Electronic commerce, though, now there's something that's emerging, and Amazon.com is the top dog and first mover in that category. Similarly, in direct retailing of computers, Dell Computer was a Rule Breaker. Starbucks has been the first-mover and top dog in the gourmet coffee field for a while.

Silicon has been around for a while. The thing that makes this both "important" and "emerging" is the speed. Only the processor makers have been working on things that go above 1GHz. The memory makers and peripheral makers have kept the speeds lower. They haven't really needed to go extremely fast until now. The benefits are fewer chips and fewer signals, but the down side is tougher standards in board layout and signal levels. Rambus has solved most of the problems that provide a barrier to entry into the microwave realm, and they took the trouble to patent as well.

Top dog? Rambus is the only dog. First mover? How does 1990 sound? They have been planning to profit from this era of technology for 10 years.

Right now, the DRAM thing is first and foremost on everyone's minds. The real technology here, though, is the part that lets you send data from chip-to-chip at speeds well in excess of 1Gb/S on each signal line. They are at 800MHz now, and they are working on quadrupling that.

The kicker is that any future design for high-speed transfer of digital data between chips will almost certainly involve one or more of the Rambus patents. And since they have shown that they will take you to court if you infringe, people are likely to pay 1% or 2% to use their technology.

Let me help you with the pronounciation: it's gor-IH-lah. Say it slow. gor-IH-lah.


The company needs to demonstrate sustainable advantage gained through business momentum, patent protection, visionary leadership, or inept competitors. Examples of these include Wal-Mart (with business momentum that featured net income gains of 25% during much of the 1980s), Amgen (enjoying patent protection of its drug formulas for many years), and Microsoft (with visionary leadership that benefited from Apple Computer's regrettable decision not to license its technology).

Just starting the business momentum, but it's looking good. They have patents (almost 100, over 100 more in the queue). Is there anything that is not visionary about this company? Inept competitors - it takes two hands to count them. Why didn't the companies manufacturing RAM chips figure out that one day they would need to completely change their architecture and invent it themselves? They make really good chips, all of them. But visionaries they are not.

Recently, their sustainable advantage became more concrete when they decided to sue Hitachi for patent infringement. The suit covers every competing high-speed DRAM technology on the market. If they win in a big way, they rule the DRAM research roost. Even if they lose, they may still come out pretty good (unless they lose big, which no one thinks will happen at this point).


The market should have recognized a Rule Breaker's promise by rewarding it with strong price appreciation. A good indication of this is a relative strength rating of 90 or above. (You can check up on company relative strength ratings in the Investor's Business Daily newspaper.)

If it isn't above 90, it is darned close. I am not even going to bother checking up on this one.


Look for good management and strong backing. Like the steel company Nucor (yes, steel!), led by Ken Iverson, which became a world-class powerhouse by revolutionizing steel production processes. Or Scott Cook, whose singular focus on serving customers drove the success of personal finance software giant Intuit. Also consider the "backing," or supporters of a company. eBay was backed by Starbucks and Sun Microsystems executives.

Backed by Intel and Dell. Backed by Sony and Nintendo. Almost all of the DRAM manufacturers as well.

AMD is holding out, but I suspect this is really because they don't want to tip us off that they are scrambling to bring RDRAM controller chips to market. They don't want to admit that Intel beat them to the punch. And don't try to tell me that Athalon would not get a speed boost from RDRAM!

Their management seems to be pretty good. You don't get a good technology to succeed without management and marketing, and they seem to have both in addition to a really cool technology.


Also important is having a strong consumer brand. Again consider Starbucks, and how its name recognition is so much stronger than competitors such as um, like (get the point?).

Not much until now. Rambus is getting a lot of free advertising from the PC makers who offer it. This is good for the general perceptions of consumers, who will soon equate RDRAM with "SPEED".

Maybe more important right now is that the chipmakers and PC makers understand what it is. They do. They do.


We also consider it a good sign when the financial media, not seeing the big picture, calls a company overvalued. (Perhaps the greatest single contrary indicator is Barron's lead editorialist. When Barron's is asking about America Online: "Short on Value?" Good. When Barron's leads with "Sell now!" -- excellent.)

It was overpriced at $70, remember? The short-sellers do (may God rest their portfolios). We have had a few Wise folk take a tentative interest, but thankfully they still don't have a clue. There are only 5 analysts covering this company even though it has been blessed by Intel for a very long time!

Besides, we don't need the Wise to say it is overvalued. Just look at that P/E! Look at that PSR! It is obviously overvalued. Duh!

<RANT>The wise will probably take more of an interest when the stock hits $400 or $500. Maybe $800, and certainly $1000. $2000? If they haven't figured it out by now, the situation for them may be hopeless. After all, they totally missed the buying opportunity at $66.

When all the wise are recommending, then it is time to sell! :-) </RANT>
Print the post Back To Top
No. of Recommendations: 3
Take a look at this link. I can give you some perspective over what RDRAM really is and how it stacks up against other things coming up. As I see it it really is too expensive for the results you would get (unless you are accessing lots of data in a sequential manner which is not usually the case in multitasking environments). Also Intel is pushing hard for it but it seems to be messing up the whole thing. If the chipsets to handle this memory is lousy, who do you think will take the fall? The memory maker or the chipset maker?

http://www6.tomshardware.com/mainboard/98q3/980710/
Print the post Back To Top
No. of Recommendations: 3
Take a look at this link. I can give you some perspective over what RDRAM really is and how it stacks up against other things coming up.

This is an excellent article and one I've read in the past.

The truth is, while RDRAM can provide more bandwidth than traditional SDRAM or DDR SDRAM, systems can't take advantage of it. The bottleneck is in the front side bus which currently runs at 133MHz at best. SDRAM can already provide that much bandwidth, and it's a whole lot cheaper. And, SDRAM has lower latency.

Some people suggest RDRAM prices are going to fall to comparable levels with SDRAM. I tend to disagree. Currently if you add an additional 128MB RDRAM to a system (even over at the Dell site) you'll end up paying about $700 more than if you were adding an additional 128MB SDRAM, which has lower latency and already fulfills the 133MHz maximum front side bus speed. It's going to take a long time for that high price to come down.

So, while RAMBUS may be the first and only mover with RDRAM, RDRAM is not necesarily the memory technology of the future. Others are working on better solutions all the time, including Micron.

I wouldn't invest in RAMBUS.

-Corey
Print the post Back To Top
No. of Recommendations: 4
As I see it it really is too expensive for the results you would get (unless you are accessing lots of data in a sequential manner which is not usually the case in multitasking environments).

The article is about PC133 mostly. Yes, RDRAM is more expensive than other kinds by a large margin - for now. The news of the past few weeks indicates that it won't be for much longer.

If you have a computer that is using an 820 chipset, it will perform just about as well as one equipped with the faster SDRAMs. That is for one channel moving 8 bits at a time. The 840 chipset results in much better performance.

All this is beside the point. This is the first generation of RDRAM, and the processors it is tied to are being retrofitted with control chips. Those processors are designed to work with SDRAM, so the fact that you see any performance improvement at all going to a completely different architecture is interesting.

You start to get better, more tangible performance gains (theoretically) when the processor is designed and optimized to the Rambus memory. Right now, we are limited by 133MHz and 200MHz external busses on the current generation of processors. Already you have a 3 or 4 to 1 disparity in the speed of the processor and the maximum speed of the memory, and they need a design change to fix it!

The other big issue is the future. You can't scale the SDRAM technology from 133MHz to 266MHz to 532MHz to 1064MHz... forever. The technology just isn't designed for that. You can, however, parallel 8 RDRAM channels to get 64 bits at 800MHz today. In a year or so, you will be able to get 4 times that. Direct to the processor.

Of course, the processors won't be able to handle those kinds of data rates for a while, but having a 1GHz processor running on a 200MHz bus is a real shame. What will we do at 10GHz?

Latency will continue to be a bigger and bigger problem for all technologies. At 10GHz, the electrical impulse can only travel about 6/10 of an inch. This is discounting completely line capacitance and signal rise/fall time. RDRAM has excellent latency characteristics when compared with other technologies. If you want links to a white-paper that goes deeper, mail me. It will only get better when it does not have to go through a controller chip - i.e., the controller is built in to the processor and RDRAM talks directly to the L1 cache.

So what about now? Go to http://www.rambus.com and look at the performance benchmarks. Wait! Before you think that they are somehow rigged, you should know that Tom's has done the same set of benchmarks and they look remarkably similar to what you see on the Rambus site.

The one thing about Tom's is that they compare it to an early brassboard-motherboard with DDR (no pictures, alas). The benchmarks for DDR and RDRAM with 840i chipset are virtually identical. However, Rambus is currently in court challenging DDR - apparently they think it infringes on a number of their patents (4 that I know of). If they win the case, they get to collect royalties on DDR as well.

Why do RDRAM and DDR have the same performance specs? Because DDR is designed to maximize performance on a 200MHz front-side bus (different on different processors, but just assume 200, okay?). The 200MHz bus is a bottleneck, but these processors were designed for slow memory. No one thought it would be a good idea to have a 32-bit 800MHz bus on an 800MHz processor, because at the time the memory did not exist. Give the processor designs a little time to catch up, and give RDRAM a little time to come down in price.

Why RDRAM now? Why not some other technology? Good question. I own the stock, but I will not rush out to buy an 820i board when I could get comparable performance out of fast PC-133 memory. I might buy an 840i board after the prices settle a bit. Remember, this stuff is still not even in full production yet. Supply goes up, meets demand, prices drop. I expect RDRAM to eventually reduce the cost of a system while still providing performance as good as or better than anything else on the market.

However, the Japanese are rushing to buy an RDRAM device - the new Sony PlayStation II. Why RDRAM? Two chips instead of 8 or 16. Rambus estimates this saves Sony anywhere from $30 to $50 per unit shipped. I don't currently own a gaming system, but I will be in line this summer to get one of these. PlayStation II is going to be huge, and Rambus gets a couple of bucks for every unit sold.

In the next year, look for RDRAM to drop in price to the point where it is within 10% of PC-133. Also look for it to be used in cheaper computers and desktops. I am still trying to find hard numbers on these issues, but I will be posting them to the RMBS board as soon as I find them.

JasonX
Print the post Back To Top
No. of Recommendations: 0
jasonxsmith,

Excellent post. In fact, this post addresses many of the concerns I've had with RDRAM. Again, thanks!

-Corey
Print the post Back To Top
No. of Recommendations: 0
Hello,

I just want to point out that RDRAM is pretty darn expensive compared to SDRAM. RDRAM is a great product, very fast but unless you want pure performance it might just be too costly for the very price sensitive mainstream PC market. Peole might not be willing to spend hundreds of $$$ more to have RDRAM instead of SDRAM in their systems.

I'm not expert, but i think the main reason for the high price is the fabrication process. It is not efficient compared to the SDRAM process.

You also want to compare RDRAM with the upcoming DDR-SDRAM (Double Data Rate SDRAM) which transfer twice as much information as normal SDRAM and is probably not much more expensive to produce. I think RDRAM would still be faster than DDR-SDRAM but i'm not so sure it will be worth the extra $$$.

I know Rambus is backed by migthy Intel but it will be interesting to see how the story unfold since it appears that Intel switching to SDRAM for most of it's future maintream products leaving RDRAM support only for high-end systems.

The question is can SPEED overcome $$$?

Smile and have fun!

Stephan
Print the post Back To Top
No. of Recommendations: 0
Rambus is a much faster memory system than we have currently, but the DDR-DRAM will give the same speed benefits of the Rambus. DDR has the advantage, though, in price. Manufacturing plants that currently produce standard SDRAM chips today can be refitted to produce DDR-DRAM chips fairly easily and inexpensively. It is far more expensive to switch the plants over to produce Rambus, in addition to the licensing costs for Rambus. When Rambus and DDR-DRAM compete against each other later this year, comsumers will find that they can get twice as much memory for the same cost if they buy DDR-DRAM.

What is interesting is that Intel is tying itself to the Rambus, and AMD to the DDR-DRAM. If DDR-DRAM becomes the memory of the next several years, Intel may find itself backing the losing side.

I have been a strong supporter of Dell and Intel for years, but I am now considering other manufacturers using an Athlon chip just so I can get the DDR-DRAM memory instead of Rambus.
Print the post Back To Top
No. of Recommendations: 0
Only one point.
If they loose in court to Hitachi they loose big because noone would pay any royalties to them if they see they don't have to. So maybe RMBS is now a Gorilla candidate but witout winning the process and without cheaper RAMs they won't cross the chasm.

Efti
Print the post Back To Top
No. of Recommendations: 0
Laugh in the face of danger. Alot of controversy on this memory and company. I wonder if Intel had this much negative response when the pentium first came out.
Dell has already shipped systems with the rdram in them and taking up a very big ad campaign for it. Slogan- R is for speed.
I bought into this company last June and have been waiting for the big explosion. I saw it back then as a gorilla in the making. With the backing from all the big players in the industry, I seriously doubt this is going to bust. And as for the lawsuit, it is not over rdram, but earlier versions of sdram. They will still get royalties from the rdram manufacturing no matter what the outcome.
A very informative site, go read. Optomist and pessimist alike-
http://www.rambusite.com/NewellSummary.htm
Print the post Back To Top