Apparently Rambus DOESN'T own DDR

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
well I have grown to enjoy your posts,
but refusing to use rambus for no reason and wanting to use simms instead makes you the apparent FANATIC in REVERSE
that you so often accuse me of..

is the glass 1\2 full and 1\2 empty
who is the fanatic, for something or against something
CAMERON

CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-
 
The thing is though, that Rambus Dynamic Ram (RIMM) kinda sucks ass. I mean, 440BX and 815 SDRAM chipsets with SDRAM for Pentium III usually performed better than Rambus, correct? I mean, all these freaks tote Rambus memory for "future" uses, but right now it's price doesn't justify it's performance, and in the future, who knows what else we will have? Maybe 200 MHz Quad Data Rate SDRAM??? Sure Rambus is coming out with PC1000 RIMMs, which might do better than current 600, 700, and 800 RIMMs, but probably not much better than current DDR PC2100 SDRAM. I mean, there is a reason why Intel itself is going to be developing its own DDR PIII and P4 chipsets...

"We put the <i>fun</i> back into fundamentalist dogma!"
 
NO NOT at ALL LET ME EXPLAIN

the Pentium 3 was not made for rambus and neither was teh 815 or BX chipsets.

their maximum data rate throughput is about 7-800 MPS

since RAMBUS has a capability of 1.6 or 3.2 GPS,
the chipset and P3's cannot fill the extra capability of rambus,
so since sdram data rate is about the same or matched to
bx\815\ P3 bus throughput, then it shows the same performance as rambus..with those limitations of P3 and sdram and 815
BUT that does not mean rambus isn't faster..

when you get a chipset that has the same bandwidth as the P4
the 850 at 3.2 GPS and rambus is matched at that rate,
then you get 3.2 GPS bandwidth and rambus is 4 times faster.

its like the notepad example I like to use.
launch or do something in notepad a very simple app
in a P3 500 and then in a P3 1000,
its about the same regardless..
does that mean the 2 CPU are the same speed...
to some YES but that would be incorrect..
notepad just is not stressing the CPU's and memory enough
to show a difference between the 2.

run Mpeg 4, or MP3, or render a 3d image, and then you see the differences between the 2.
the ghz P3 is faster

so many reviews test with simply looped benchmarks, or
office apps or apps that cannot saturate past P3's
or athlon or SDRAM's ability of 800 MPS throghput

so it appears that they are just as fast as a P4.

but look closer and run very demanding software and as the throughput demands go past the 800 MPS mark and begin to stress the above CPU's , the P4 and RAMBUS
have a huge reserve and the performance curve continues to
climb while the others level or decline..

think of it as 2 glasses, one holds a pint and one holds a quart.

if you put the same amount of water in each, say 8 ounces,
the smaller glass will be half full.
as you pour the 8 ounces of water into the larger glass
even though the same amount has been poured into it
it is only 1\4 full, and this has room for more water where the other glass would overflow

I hope that helps with your good question
best
CAMERON

CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-
 
http://www.bigchart.com/news/articles.asp?newsid=776421492&symb=RMBS&sid=41488
<b> UPDATE 1-Rambus tumbles on reports of negative patent ruling
THURSDAY, MARCH 15, 2001 5:20 PM
- Reuters

(updates with Rambus comment in paragraph 5)

By Timna Tanners

LOS ANGELES, March 15 (Reuters) - Shares of computer chip maker Rambus Inc. (NASDAQ NM:RMBS) lost nearly 32 percent of their value on Thursday amid reports of a negative court ruling on its patents that could cut into its royalties, a significant portion of its revenue.

The company said no ruling on the patents had been issued, and called reports of a negative ruling inaccurate.

Shares of Rambus, which has developed technology to speed the performance of memory chips, fell $11.26 to $24.09 on the Nasdaq, where it was the second biggest percentage loser. The stock was at its lowest level since February 2000, down from a high of $127 hit in June.

Brokerage SG Cowen reported that U.S. District Court Judge Robert Payne made a pre-trial ruling in its court case with Infineon (NYSE:IFX) 🙂) siding with Infineon regarding the scope of Rambus' patents.

An article posted on Wednesday on Cahners' Electronic News, an online electronic industry news site, also referred to the ruling and said it would limit the scope of Rambus' patents. It said the decision could reject Rambus' royalty claims on synchronous dynamic random access memory (SDRAM) and double-date rate (DDR) memory. The article cited sources close to the case.

Rambus spokeswoman Kristine Wiseman called the article inaccurate. "There was no ruling that has been issued as of yet," she told Reuters.

A Richmond, Va., attorney representing Rambus referred calls to the company. An Infineon spokeswoman declined to comment.

"It looks like it's down on an apparent negative ruling in the patent lawsuit with Infineon, although there is no definitive news from the court," said Tim Ghriskey, portfolio manager of the $4 billion Dreyfus Fund, which does not own the stock.

Rambus collects royalties from Samsung 🙂), Mitsubishi Electric Corp. 🙂), NEC Corp. 🙂), Toshiba Corp. 🙂) and Hitachi Ltd. 🙂), and has pending lawsuits to collect royalties from Hyundai Electronics 🙂), Micron Technology Inc. (NYSE:MU) and Infineon.

Rtr 17:20 03-15-01</b>
 
Dude(cyberimage) you must of been burned big time by RAMBUS. You must of bought the stock when it was at $117 a share.

I have never seen someone try to put such a SPIN on events like you have with these posts. Well except for bush's press secretary about CO2 emissions-LOL(LOL about the stupidity of it not the CO2 emmissions).

Your like the engineer in the engine room of the TITANIC saying 'she'll be okay, the hole isn't that big, we will just plug it up with some cloth and we should be able to get home'.-LOL

Well sorry to tell you, this ship is sinking in my opinion and by the looks of the drop today in the stock price many others think the same thing.IF the ship is listing to one side get in that life boat and row, row your way to safety for godsake-LOL.

<P ID="edit"><FONT SIZE=-1><EM>Edited by onipion on 03/16/01 05:21 AM.</EM></FONT></P>
 
Hey Cyby/Fugger, both RDRAM and RAMBUS, Inc. suck! Ok? If Intel hadn't given RDRAM it's blessing, would you be all over it? Doubtful! Put another way, if AMD was backing RDRAM and Intel DDR, which memory would you be doing your sales pitch on? On second thought, don't answer that. I already know the answer.
 
There are new allegations of racketeering against Rambus. What kindof terrorist business are they???


<i><b><font color=red>"2 is not equal to 3, not even for large values of 2"</font color=red></b></i>
 
Cyber,
The only correction they have issued at this time is regarding the summary judgement portion of the article. Not that a preliminary ruling not in Rambus' favor will be forthcoming. Where you quote the editor of electronic News as saying, ''It was an error that came at an unfortunate time,'' said Tom Moran, ElectronicNews executive editor. ''We have corrected it. We stand by the rest of the story.'' Bloomberg quotes him here (http://quote.bloomberg.com/fgcgi.cgi?T=marketsquote99_news.ht&s=AOrEg0xTpUmFtYnVz) as saying ``It was a relative minor error that came at an unfortunate time,'' said Tom Moran, ElectronicNews executive editor. ``We have corrected it. We stand by the rest of the story.'' The "relatively minor" portion of his quote was omitted from the version of his statement in your post, a small but significant difference. Regardless of that, it is far too premature to call the story "bullshit". This is a complex case that is going to be very interesting to follow, and Rambus is far, far from being in the clear.

Mike
 
To quote Goodfellas:

"Buh-bye, d1ckhead. See you in Attica, d1ck."

-------------------------------------
Nature abhors a moron. -- HL Mencken
 
No kidding! There are a whole myraid of reasons to avoid RDRAM and RAMBUS, the toll collector. How can anyone feel warm and fuzzy about a company with such questionable, non-ethical business practices and tactics. Is it any wonder that RAMBUS and Intel gravitated toward each other.
 
Rambus' RDRAM generates more bandwidth at the expense of latency. This is why the P3 on an SDRAM solution is better than a P3 on a RDRAM solution. The P4 was designed for RDRAM from the start and operates quite differently than any previous processor, and it shows in its terrible performance when compared to an Athlon on DDR-SDRAM. Bandwidth is not the most important factor in memory, which has been proven time and time again as memory bandwidth was increased, yet very little performance was gained. You see the most gain from latency improvements. Only recently with streaming media have you seen an application where bandwidth was more important than latency for anything but highly specialized projects. This technology isn't useful for the average consumer, yet Intel users are forced to simply suck it up and buy the inferior and more expensive solution.

There has never been a doubt that Rambus owned the patents on double and quad pumping memory. The issue is that Rambus applied for several memory patents, then joined Jedec and used its membership to move the industry towards the technologies it held patents for, applied for new patents and modified existing patents as the industry discussed its future direction, and left Jedec after it was so vested in the technologies there was no turning back, securing Rambus' future with licensing fees. When DDR SDRAM came to market, it declared these patents and demanded money, making it clear that any company that went to court to try to overturn the patents would pay significantly higher licensing fees than those that just paid them to begin with. Around here we call that extortion.

Think of it this way. You decide to take a vacation and go fishing with a couple co-workers. You all talk things over and decide that one of you will bring the food and haul the camper with his SUV, the other will bring the bait and canoe, and you'll bring the firewood and cooking utinsils. When you get up to the lake, miles from civilization, the guy with the truck and food suddenly says, "hey, you know what? I want you to pay me $5 for this turkey sandwich. Oh, and I brought my own firewood, utinsils, and bait, so I won't be needing yours." Some people might get pissed and pay the $5 for the sandwich, and others might take a canoe paddle to the guy's head.

Rambus is simply exploitive and overly litigous. It has set memory technology back a year or two if they lose, and 4 or 5 years if they win. They've increased the costs of memory for everyone as memory companies are forced to pay lawyers or Rambus. Through back-room deals with Intel they've forced an inferior memory technology upon the consumer. They did not wait until their fabrication processes were advanced enough to make the technology affordable before bringing it to market. Personally, I find these acts to be distasteful enough to boycott their product, which I am doing, along with a great many others. I am also boycotting Intel because of the role they played. I could potentially end up with an inferior machine (though CERTAINLY not today) but I think the sacrifice would be worth it if Rambus can be made to bleed a bit.

Hopefully, Rambus will lose, and all of these companies that were strongarmed will be able to countersuit and force Rambus to declare bankruptcy, or at least have stock so worthless the company can be bought up and disassembled by a more honorable company.

/Athlon-1.2GHz@1370MHz(137MHz*10)/Asus_A7V133/
 
http://dowjones.work.com/index.asp?layout=display_news&searchtext=rambus&doc_id=DJ200103161120DOWJONESDJONLINE000377&source= Dow Jones Business News

It aint over, but according to "Dow Jones" the ruling has been issued and apparently is not in Rambus's favor. Not a knock-out, but helps Infineon's case
 
Sojourn
Agree-Great post. It's what I have been trying to say on the JEDEC issue but a lot clearer and with more detail to back it up.
Mike
 
That way of testing a processor is BS. Absolutely no one on earth encodes MPEG 4, runs MP3, and renders an image at the same time. If I am spending $500 for a processer (P4) and four times as much for RAM, I damn well want that to actually give me a benefit while playing Unreal Tournament or something. But in the majority of cases, like with Pentium III, normal SDRAM on 440BX or 815 chipsets performed better, much better in some cases, than RDRAM in the 820 or 840 chipsets. Not to mention that Rambus requires dual channel implementation, correct? Meaninh no matter what you have to have atleast 2 RIMMs in the system, further adding to cost.

You say "the Pentium 3 was not made for rambus and neither was teh 815 or BX chipsets." Obvisously Intel thought Rambus was made for the Pentium III, being that they tried to force it, unsuccessfully, on consumers. The fact remains that RDRAM has huge latency penalties compared to SDRAM. What would your opinion on the subject be if DDR Pentium III systems show performace increases over both regular SDRAM and RDRAM systems?

For such an expensive memory type, it should actually help make the P4 perform how it should, which it obviously doesn't.

"We put the <i>fun</i> back into fundamentalist dogma!"
 
>Hopefully, Rambus will lose, and all of these companies that were strongarmed will be able to countersuit and force Rambus to declare bankruptcy, or at least have stock so worthless the company can be bought up and disassembled by a more honorable company.<

Somehow, I have a vision of Rambus Lawyers being "disassembled" via the "drawn-and-quartered method." LOL!



<P ID="edit"><FONT SIZE=-1><EM>Edited by P4Fool on 03/16/01 01:13 PM.</EM></FONT></P>
 
US court deals setback to Rambus in royalty suit
RICHMOND, Va., March 16 (Reuters) - A Virginia federal court judge in a pre-trial ruling released on Friday rejected computer chip maker Rambus Inc.'s (NasdaqNM:RMBS - news) definitions of eight disputed terms in a lawsuit seeking royalties from German semiconductor maker Infineon Technologies AG (NYSE:IFX - news) and its U.S. subsidiary.

In a ruling that sent Rambus shares plunging, U.S. District Court Judge Robert Payne said the company's definitions of key terms defining the scope of four patents at the center of the dispute were ``at odds'' with claims made by inventors in seeking the patents with the U.S. Patent and Trademark Office.

Payne also said Rambus, and an expert witness who testified for the company, appeared to be trying to broaden the scope of its patents on synchronous dynamic random access memory (SDRAM) and double-date rate (DDR) memory, which are at the heart of its royalty suit against Infineon.

``Also, it was difficult to credit (the expert's) testimony on disputed terms because it reflected the general, and disturbing, tendency of Rambus to distance its current constructions from what the inventors said in...the specification, and, in doing so, to use the claim construction process to broaden claims,'' Payne wrote in the 77-page ruling.

Rambus filed suit in federal court in Richmond, Virginia, alleging that Infineon had infringed on four Rambus patents and seeking royalties from the German semiconductor company.

The pre-trial ruling, released to attorneys for both companies on Thursday, set the stage for a jury trial on April 10.

Rambus said it would not drop its royalty claims against Infineon despite the judge's ruling, which adopted Infineon's definition on each of eight disputed terms underlying the patents.

``Based on this interpretation, Rambus maintains its allegation that Infineon has infringed these four patents. Rambus is prepared to protect its intellectual property from those who infringe and looks forward to presenting its case to the jury,'' the company said.

Rambus shares plunged 26.8 percent on Friday, falling $6.45 a share to a year-low of $17.64 in afternoon trading on the Nasdaq, a day after shares tumbled $11.26 a share to $24.09 on unconfirmed reports of the ruling.
http://biz.yahoo.com/rf/010316/n1631222.html
 
you mist read my post,
I did not mean those test all at the same time..

also, P4 and rambus are faster in games that are heavy loading like Quake which use OGL , more memory and CPU..

Quakes get 60 FPS faster than on athlon..


normal SDram does not perform better, it only seems that way when you use apps that do not stree the memory or CPU bus.
that is why I listed the above apps to test..

you are arguing for this moment about what people use.
software is ever changing and will get more complex and stree components more very soon.

P4 and rambus are ready for that and can handle it
sdram is NOT..

latency is an ooverused term which tom and others write about by looking at spec sheets rather than the whole picture..
the 850 chipset and Pentium 4 have software, and architecture that compensates for this, uses longer pipes moving more data,so it does not really matter.
if you have one with latency of 2 moving 500 MPS
and another type of ram with latency of 4 moving 20 MPS
even though the higher latency ram looks slower it isn't as it is moving 4 times the data..

CAMERON

CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-
 
no you are totally wrong

latency is an overused term which tom and others write about by looking at spec sheets rather than the whole picture..
the 850 chipset and Pentium 4 have software, and architecture that compensates for this, uses longer pipes moving more data,so it does not really matter.
also rambus uses doubling data rate technology by using both sides of the clock signal, so latency becomes less of an issue
if you have one with latency of 2 moving 500 MPS
and another type of ram with latency of 4 moving 20 MPS
even though the higher latency ram looks slower it isn't as it is moving 4 times the data..

if what you say had any truth to it, rambus and P4 would not be faster than sdram in all the applications
that it is,
and DDR would not be faster than sdram since it has higher latency.

you need a certain amount of latency when you are moving huge amounts of data expecially with clock doubling signals
or you get misfires.

cache, longer pipes and algorhythms can compensate for these

CAMERON

CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-
 
OOPS,

LOOKS like some of you guys jumped the gun and the articles were full of crap..

turns out the judge posponed the case when rambus
cited and proved that INfineon has withheld certain documents that prove that Infineon knew about the patens much earlier than claimed, and they have to go back to germany and find them under discovery

NOT good for infineon, since its a jury trial,
one lie you are caught in and the jury sides with rambus
no matter what other things happened..

and it still does not disputer the patent of rambus..
all Infineon is claiming is rambus did not tell it about
patents and apparenty with the ruling on fri by the judge,
he thinks maybe they are not being forthcomming abou that
since rambus cited certain doc that prove their case..

we will find out in APR.
better buy the stock now its only going up !!

can't wait !

CAMERON

CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-
 
Cyber,
You are correct there has been a delay in the trial. But the judge specifically said he was not saying that the incomplete response was intentional. And Infeneon is certainly disputing Rambus's requested INTERPRETATION of the patent(s) in question, not just accusing them of withholding information at JEDEC. It's somewhat telling that Rambus requested a much broader definition of key terms, broader apparently than they defined those terms in the actual language of the patents themselves. The judge shot them down and pretty much accused them of tryiing to re-write their patents to make them support weaknesses in their case. I don't claim to know what the outcome of this will be, but it's very hard to look at this as being just a minor housekeeping ruling. As you should know, patent cases hinge on very small, technical details. And it appears that at least four of those details broke for the Infineon camp today. For example, Rambus wanted the definition of "Bus" expanded to include and connection by wire or otherwise to their memory. Infineon was able to have the definition narrowed so that it only included a very specific type of bus, the "Multiplexed Bus". Since they don't use the Multiplexed bus in their DDR memory they feel that this is a victory for them. I don't see any way to put this into a positive light for Rambus, as much as you or they might try.

Mike
 
You know, I actually <font color=red><b>REALLY</font color=red></b> hate to do this (and to emphisize this, I even used the non html tags which I hate as well), but I have to actually AGREE with Cybey.

**shudder**

Ugh.

I feel so dirty.

But he is right. Look at the nVidia GeForce video cards. Which do the video cards gain more performance from: Faster memory, or higher bandwidth? They gain more from higher bandwidth every time. Bandwidth is their main limitation, same as the RADEON cards, same as ANY processor which pushes things to the very edge of computational limits.

The P3 didn't show any gain from RDRAM because the whole architecture wasn't designed to support the higher bandwidth. It would be like making a translater to put PC133 SDRAM into an EDO SIMM slot. You might be able to get it to work, but it will never run at it's true potential. Does that make the PC133 SDRAM slow or crappy because it hardly performs better than an EDO SIMM in that situation? No more than it makes RDRAM crappy for performing so poorly when put into a P3 system.

So why did Intel shove RDRAM down the P3 user's throats? Because their P4 is optimized heavily for the high bandwidth of RDRAM. And Intel needed the assurance that RDRAM would have been produced commonly by memory manufacturers so that it would be at decent speeds and (hopefully) affordable prices when the P4 came out.

Of course, their plan didn't work as they had hoped because not many people wanted to pay so much for RDRAM in a P3 when the performance gains were hardly worth mentioning, if not worse than SDRAM.

Will a P3 benefit more from SDRAM than from RDRAM? It's possible. It isn't likely to be much though, just as the Athlon isn't gaining much. It wasn't designed for high-bandwidth memory either. Both cores are pretty old by now.

Will a P4 and any other chip designed to take advantage of higher bandwidth memory benefit from DDR SDRAM more than RDRAM? It really all depends on the chip architecture. The P4 obviously does, otherwise Intel would have ditched Rambus. No one can convince me that Intel's deals with Rambus have provided them with more money than what they would have made had they ditched Rambus all together. Intel has lost a LOT of ground and face for sticking with Rambus. So what do they have to gain from doing it? Because their P4 works better with RDRAM than with SDRAM. That, or their executives must be smoking some really wacky weed.

Don't get me wrong, I HATE Rambus and the way that they've screwed over the SDRAM community. I think the executives should be whipped with a cat-of-nine-tails. But I also think that RDRAM itself does have more potential than SDRAM derivatives. And if we see the latency improved in RDRAM, DDR (or QDR) SDRAM won't stand a chance.

RDRAM isn't much more expensive than high bandwidth DDR SDRAM. And RDRAM prices are dropping.

The company itself may be a bane on this Earth and everyone's finances, but the technology isn't.

I believe the saying goes, "Don't throw the baby out with the bathwater."

-<font color=orange>Olestra</font color=orange> may cause <font color=red>abdominal cramping</font color=red> and <font color=yellow>loose stools</font color=yellow>.
 
You're a good man Silver. I have even had to grudgingly admit through gritting teeth that RDRAM is a decent technology, especially for the P4. Even though I still believe the P4 is a truly SUCKY product and that rambus is a SCUMMY SUCKY company, the bandwidth advantage P4 gets when paired with RDRAM is one of it's only bright spots.