New RDRAM chipset "beats DDR by 50%" claim

[Cerb]

Remeber, as much as I feel RDRAM is superior, there's ONLY TWO COMPANIES MANUFACTURING IT. This has an impact on the market overall, and who's gonna support it in the end.

What really makes no sense is why they don't own up and start making it...in addition, why must they keep lowering prices? I payed for Crucial and Kingston when it was $100 for 256MB PC2100, even though I could now get it for $40, and could have gotten it for $60 even then. I will keep paying for that. Anyone who doesn't WILL encounter problems at some point down the road. DDR and DDR-II are going slow.
We should see DDR-II by the end of 2003. We should have had PC3200 as mainstream last year for the P4, with PC2700 mainstream for the Athlon (with PC3200 mainstream for the 166MHz ones). While Rambus stuff is expensive, I will give it that much...it has released the right stuff at the right time (a few months after the people here really want it), instead of us still using PC133 in new machines.

Again, why do these copanies not produce RDRAM, if it could at least make up for their losses? I wouldn't see it as a stain on Crucial DDR that they sell RDRAM, or even pay royalties. If PC2100 and PC2700 DDR doubled in price tonight, I'd still be buying the higher Crucial, Kingston, and Samsung. Even HP is known for using real Micron stuff, and that on their POS 100w PSU PCs.

 

[FishTankX]

Yeah, they certainley are a group of innovaters. It just seems sad that the company gets a bad rap because their lawyers seem to be seeing all the action.

Anyways, i'm pretty sure that RAMBUS technology will be used to good effect at some point in the future. Redwood looks very good. Could be used as a FSB, couldn't it?
Yellowstone looks like an atempt to kill DDR outright. Heh. The PS3 is gonna be a real powerhouse. I just hope they dont' screw up and arm it with a massive '16MB integrated texture cache' or somesuch. The original PS2's texture RAM was it's greatest downfall.

 

[Ice9]

Again, why do these copanies not produce RDRAM, if it could at least make up for their losses? I wouldn't see it as a stain on Crucial DDR that they sell RDRAM, or even pay royalties. If PC2100 and PC2700 DDR doubled in price tonight, I'd still be buying the higher Crucial, Kingston, and Samsung. Even HP is known for using real Micron stuff, and that on their POS 100w PSU PCs.

It's because of one reason, and one reason only: They don't want to pay royalties. As it stands right now, Infineon owes $800 million in back royalties. They are going to do whatever they can do to NOT pay that. Infineon is lucky in the fact that they're a subdivision of Siemens - lucky enough that they're a small enough division that Siemens could say "Ok, we're pulling IFX out of the dram business", and that would be the end of it.

In the case of Micron and Hynix, they simply *don't have the money* to pay it They have no choice but to fight it now. And they aren't owned by a larger company. They are DRAM manufacturers plain and simple, and the only thing they have protecting them is their respective governments.

And both of them are lobbying their governments for help. Right now, Hynix is in business because they're taking money from government controlled S. Korean banks to stay in the game. The rumor on Micron is their latest plant closing was done "strategically" - they closed the closest plant to Washington, DC in an effort to get some government sympathy (they laid off over 600 workers). They've also been pushing the FTC to drive Rambus out of commission, a move that prompted this letter to congress.

None of these 3 companies wants Rambus to have control over the memory industry, because if they did it would mean 1.5 to 3% less they could make per module. When you're losing $1B a year and have no prospects for profitability, that's huge.

 

[MadRat]

"Corporation Suicide" seems to be world-wide, almost like we're in World War III and instead of guns and rockets we're using subsidies and deficit-spending trade tactics. These companies sprang up from free money and they spend their capital as if its a burden. Alas, this is a thread all to itself and has no bearing on the DDR versus RDR argument.

NForce is not a dual-channel chipset, rather it uses its trademarked "T-bar" memory controller. NForce effectively interleaves memory using a large buffer and by using chipset-based "hardware prefetch" to make it appear that it really can perform in dual-channel. The chipset may support two 64-bit transfers to the chipset at the same time, but its still limited by a 64-bit pathway to the processor. The true marvel of NForce is its ability to divert its focus of serving memory bandwidth away from the processor and to the builtin video even while under the heaviest strain. I'd imagine that when ATI comes out with their chipsets based on their 256-bit memory controller it will rewrite the way we think about builtin video.

 

[FishTankX]

I am officially leaving the thread. Thanks for the intelligent conversation/debate Ice9. And for straightening me out on market hapenings. I had no idea about half the stuff.

My parting words for the thread are..

'Down with dumping! Long live innovation! May the best technology win'

 

[Sidulator]

This has got to be one of the most informative threads I have ever read here at Anandtech. I had no idea and it seems most people don't know either.

 

[Novgrod]

i still cannot fathom that RDRAM, or any of its successors, can still possibly be seen as the product of the future (or that Rambus will become more than a niche player) without more than the ps3 going for it. Maybe if ATi or Nvidia or AMD had even the slightest willingness to use RDRAM (though Nvidia has every reason not to; Rambus tried to make them pay royalties years ago) or if Intel had a plan to keep using RDRAM I could believe in the future of the company, but as it is all I've seen proven is that:

Rambus is making money while DDRam companies are losing money.

I tell ya what: when (not if!) Micron et al are paying royalties, and when Intel reintroduces RDRAM to its roadmap, I'll give Rambus better than a 50/50 shot.

Until then, the opinion of analysts are still just opinions.

(it should be noted that the opinions of pedants like me are still like buttholes).

 

[Ice9]

i still cannot fathom that RDRAM, or any of its successors, can still possibly be seen as the product of the future (or that Rambus will become more than a niche player) without more than the ps3 going for it. Maybe if ATi or Nvidia or AMD had even the slightest willingness to use RDRAM (though Nvidia has every reason not to; Rambus tried to make them pay royalties years ago) or if Intel had a plan to keep using RDRAM I could believe in the future of the company, but as it is all I've seen proven is that:

Rambus is making money while DDRam companies are losing money.

I tell ya what: when (not if!) Micron et al are paying royalties, and when Intel reintroduces RDRAM to its roadmap, I'll give Rambus better than a 50/50 shot.

Until then, the opinion of analysts are still just opinions.

(it should be noted that the opinions of pedants like me are still like buttholes).

I gotta admit... your dislike for Rambus despite all the facts is impressive You clearly want to keep your position, and that's fine. No one's asking you to change it.

As for Rambus being on Intel's roadmaps, they already are. Their newest network processors all use Rambus technology, and they've already stated that they're going to continue using it. For desktops? Yeah. It's there too. PCI Express is being delivered by none other than a coalition of PLX and Rambus.

Maybe you think this is "no big deal". Or hey, maybe you realize that these slots will eventually be in everyone's PC's... Rambus wins again with royalties.

Everyone seems to interpret this mess with Micron/Infineon/Hynix a different way. I PERSONALLY have zero sympathy for the companies who try to stifle Rambus' innovation by price dumping and using political influence to get their way like Micron has. People claim Rambus is litigous, but they're no more litigous than any other semiconductor company.

Micron is on their stumbling last leg, and they're the #1 supplier of DDR bar none. Hynix and infineon are pretty much tied for #2, with Hynix being held afloat by the state-controlled banks, and Infineon looking like they're going to have to pony up $1B to Rambus.

When these companies go down (and they certainly ARE going down if you follow the semiconductor biz - I certainly do), it's is going to have a huge impact on the DDR industry. No one is going to buy out any of these companies because of their enormous debt loads that they'd ultimately be responsible for. In fact, no one can AFFORD to spend that much money only to lose it all in paying off their debts

Micron alone will be the next Enron.

I think that that once this DRAM industry consolidation starts, there will be huge shortages and enormous price hikes after inventories are liquidated. The remaining 7 dram companies simply don't have the manufacturing capacity to keep up with the demand that Micron, Hynix and Infineon currently fill for SDRAM/DDR. Not even with Samsung leading the way.

This in turn will make RDRAM more attractive because it's already being produced cheaply and in volume, at half the royalty rate. Once these holdout companies are gone, the remaining companies will have to pick up the slack, and guess what: They all pay royalties to Rambus (7 out of the 10 dram manufacturers making dram TODAY already pay royalties on SDRAM/DDR). If those remaining 7 companies have a choice of making DDR at a 3% royalty rate or RDRAM with a 1.5% royalty rate, simple math will tell you what's more attractive from a manufacturing cost perspective. With the "holdouts" gone, there's no more market saturation of DDR.

Long-term, DDR has to give way to something better, and Rambus is clearly the only memory technology innovator who's had the foresight to develop technologies that meet future needs. JEDEC on its own hasn't moved fast enough to keep up with tomorrow, and that's keeping the door wide open for Rambus to capture the memory market.

Rambus is *quite* secure in its position. Their own bottom line shows that.

 

[Goose77]

well this is another post gotten out of hand. I just have to say some things here. DDR will be around for a while. With the amount of support and how popular it is, its not going anywhere. And all i care about is price/performance. The cheaper the better. Im glad that they're dumping vast amounts of DDR and dropping the price. And if they go out of business, well.. oh well! too bad for them.
No matter how much better RDram is than DDR, i wont buy it, and as the market shows, neither will any one else(in large numbers).

on another note, these court battles wont end there. im sure they will file for appeals. this long court process only give RAMBUS more time to shread more important documents and to get away with more. even when the court process is done, it wont be able to find out who really came up with the tech first. sure rambus got it to patten first, but that doesnt necessirly prove that they thought up of the whole thing first! they were just faster to the patten office!

And for all those companies that are gonna go under.. well, i'll have to see it to believe it first. I got no stock in them so i dont really care. i just dont think that they all will tank, no matter what ICE thinks or says! There have been many companies that people thought they were gonna tank but ended up being sucessful, and there were companies where people thought they would be sucessful but ened up failing.

In this world there are no guarantees. and no matter how much proof u provide i will never believe u. your a popous ass and the way you go about these forums lacks much to be desired. Maybe if you took a different approach to explainning your thoughts, ideas, and fact more people would be willing to receive any info u have to offer! arrogance can only get you so far, and then u just look like an ass.

 

[Ice9]

Ahh, goose. One of my favorites in this thread! I was wondering when you'd chime in with your wisdom again

Only you could say this thread is "out of hand" after others have have called it "of the most informative threads at anandtech."

Bravo!

DDR will be around for a while. With the amount of support and how popular it is, its not going anywhere.

That's fine. Rambus is gladly taking your 3% royalty payment Or did you forget about that again?

No matter how much better RDram is than DDR, i wont buy it, and as the market shows, neither will any one else(in large numbers).

Tell it to Sony and Dell, who buy it in large numbers right now, and will continue to do so. Sony will be buying it in massive quantities to support the PS2 through the PS3. *HOW* many millions have they sold of those little black boxes again?

Heck, I'll bet you even have one If you do, Thank you for Supporting Rambus!

I looked at your system rigs profile:
Memory: 384 MB of samsung(pc3200) <--Samsung pays royalties on SDRAM and DDR. Thank you for supporting Rambus!

on another note, these court battles wont end there. im sure they will file for appeals. this long court process only give RAMBUS more time to shread more important documents and to get away with more. even when the court process is done, it wont be able to find out who really came up with the tech first. sure rambus got it to patten first, but that doesnt necessirly prove that they thought up of the whole thing first! they were just faster to the patten office!

With regards to patents: Getting your patents filed with the Patent office first is THE ENTIRE POINT to the patent system!! Getting the idea, submitting it, and having the ability to license it. If you don't believe that this is one of the most important lifelines to a semiconductor stock, look at the patent portfolio of ANY ONE OF THEM.

As for the legal end, i'll summarize this for you, since you probably don't follow the legal proceedings as closely as I do. This is also all public information and verifiable as FACT.

After the CAFC decision in rambus' favor, infineon had a specific timeframe to file "en banc" (a full review of the CAFC ruling). They did so on the last day possible, the day of the filing deadline (which happened this past week).

That buys them maybe another 2-6 weeks, though 90% of these requests are denied and no one really expects to win en banc. They just file for it because "hey, the worst they can do is say no", which they do most of the time.

After that, the only choice is the appeal to the supreme court after it's been played out in the lower courts.

But you're forgetting something: Micron, Infineon and Hynix no longer have gold-lined pockets deep enough to wait that long. They have to do something to become profitable *NOW*. All Rambus has to do is sit back and let 'em go about their legal wranglings while they keep collecting royalties. They're finally vindicated, and everyone knows it... While those guys wrestle around with each other selling their drams for $2 a piece, Rambus still gets 3% of the cost to manufacture that.

And for all those companies that are gonna go under.. well, i'll have to see it to believe it first. I got no stock in them so i dont really care. i just dont think that they all will tank, no matter what ICE thinks or says!

That's ok, neither do a lot of idiot analysts that cover these semiconductor stocks. No one thought Enron or Worldcom would hit the crapper either. Heck, no one even knew they were in trouble. In the case of "The Three Bears" (Micron/Hynix/Infineon), the trouble is obvious to anyone who reads semiconductor news Just because you've decided to put your head in the sand and say "no no no, they'll never go out of business" doesn't mean it won't happen

There have been many companies that people thought they were gonna tank but ended up being sucessful, and there were companies where people thought they would be sucessful but ened up failing.

It's all in the numbers. There is no other gauge anymore. You either pay your bills and generate revenue, or you don't. Times have changed since the internet boom, y2k and Enron. We're back to good old fashioned profitability now that the .com bubble popped. "Creative Accounting" will now land your top brass in JAIL, your people LAID OFF, your factories and plants SHUT DOWN an your stock price WORTHLESS. Look at the economy. It's happening all over. I've survived 3 rounds of layoffs at my job in the past year alone.

In this world there are no guarantees.

Sure there are. Death and Taxes.

and no matter how much proof u provide i will never believe u. your a popous ass and the way you go about these forums lacks much to be desired. Maybe if you took a different approach to explainning your thoughts, ideas, and fact more people would be willing to receive any info u have to offer! arrogance can only get you so far, and then u just look like an ass.

"You're welcome."

Seriously. You are only insulting my credibility because you don't have the facts to support your own argument, or are simply ignoring the ones presented. You don't have to believe anything I say. The facts speak for themselves. You can take them all at face value, or you can subvert them into supporting your own opinions.

If it's all just the same, I don't think you're a pompous ass, just a guy trying to look like he knows something about the semiconductor industry, patent law and finance. Don't quit your day job, assuming you have one.

My apologies if this post seems kind of harsh, but the simple statement of "no matter how much proof you provide, I will never believe you" says one thing about your character: You are incredibly closed minded. I hope you never serve on a US Jury.

That's not what any good technological discussion needs.

 

[Novgrod]

Ice, i'm going to pass over your attack on a blindingly dumb defense of DDR that involved such pedagogery as "and no matter how much proof u provide i will never believe u. your a popous ass and the way you go about these forums lacks much to be desired."

Instead i have a question I ask in all honesty:

the most I can find is that a federal appeals court has thrown out a settlement against Rambus. This allows Rambus to go on with its lawsuits to allow it to try to make Infineon et al pay royalties.

Rambus still has another suit to win before it can collect royalties, no? And since Rambus lost one suit big, and won only in overturning that loss--with one adamant dissent--I fail to see how royalties are a guaranteed matter (though perhaps I'm just not looking hard enough).


TIA for information.

 

[DannyBoy]

FishTankX, you made 39 posts in this thread alone :Q:Q

 

[Ice9]

the most I can find is that a federal appeals court has thrown out a settlement against Rambus. This allows Rambus to go on with its lawsuits to allow it to try to make Infineon et al pay royalties.

Correct, this is the CAFC ruling.

Rambus still has another suit to win before it can collect royalties, no? And since Rambus lost one suit big, and won only in overturning that loss--with one adamant dissent--I fail to see how royalties are a guaranteed matter (though perhaps I'm just not looking hard enough).

Well, of the 10 dram manufacturers making SDRAM/DDR, don't forget that 7 of them already pay royalties.

Anyway, this means the case goes back to the lower court. The lower court will have to agree to hear it again (the en banc proceedings) or deny en banc. Here's what could happen:

The lower court could deny the en banc, time's up, game over, unless you want to go to supreme court and appeal. The supreme court is NOT obligated to hear the case and would likely refuse. The supreme court is HUGE on the "proper channels" being used. Either way, Infineon has to weigh the options of going to the supreme court and risk paying triple damages. It will be incredibly difficult for infineon to win with Rambus being recognized as the legitimate patent holder because of the reinstated markman ruling granted by the CAFC.

Most en banc filings are denied because of the time and brain power the CAFC judges use in making a ruling.

However, the lower court could also could agree to hear this case again (granting en banc). The problem for infineon here is that Rambus will *DEFINITELY* file for a summary judgement (a quick resolution without a jury trial based on current evidence) and is expected to easily win. The CAFC has made this all but inevitable with its verdict, not to mention the reversal of the original markman ruling placing Rambus as the patent holder. No judge will deny a summary judgement with the CAFC's stamp of approval all over it.

There are a LOT of links to dockets and other legal documents on www.rambus.org and in the forums at www.rambusite.com.

 

[Novgrod]

perhaps I'm just obtuse, but if the lower court does not hear the case, how does that make infineon begin paying royalties?

And it is my opinion, through an admittedly limited study of the judicial process, that such a strenuous dissent from one of the three judges in the appeals court makes the outcome of any future case less certain.

TIA for responses.

 

[Ice9]

perhaps I'm just obtuse, but if the lower court does not hear the case, how does that make infineon begin paying royalties?

Because the appeal verdict holds Rambus as the legitimate patent holder (reinstated Markman). Rambus would be free to seek a summary judgement against Infineon and demand they pay $1b in back royalties as soon as this en banc business is over (2-6 weeks). Infineon has the right to file for en banc, and they did so on the very last day they were entitled to. If they hadn't, Rambus would have been in court that very next day getting their summary judgement

And it is my opinion, through an admittedly limited study of the judicial process, that such a strenuous dissent from one of the three judges in the appeals court makes the outcome of any future case less certain.

Doesn't even come into play. The ruling is just that. The entire panel, dissenters and all, agree to the final ruling. The lower courts must adhere to that ruling. They can hear the case again, but within the guidelines of the CAFC verdict. This means Infineon would have to quickly come up with a new defense other than "Hey, they snooped in on our JEDEC meetings".

Even if it did matter, the single dissenting judge is a newcomer to the panel. If I remember correctly, he has been on the panel for a year, while the other two have been there for decades. His lack of experience would be REALLY easy to exploit in court by Rambus' lawyers, and the opinions of the other two judges would weigh far more in the eyes of any jury. One "Green" judge from the CAFC panel does not a verdict make

 

[ElFenix]

great, so rambus broke a contract and they're going to get away with it. yay legal system!

 

[Novgrod]

Last note for tonight:

If watching Rambus over the past three years has taught me ANYTHING, it's never to bet on lawsuits, since the company has seemed both sure of victory and hanging over the jaws of defeat on a number of occasions, and analysts (Edelstone in particular) went from praising it as the Second Coming to kicking it out of bed in the span of a couple hours before.

I would also opine that the next case isn't all that clearcut; the dissenting opinion, combined with the earlier finding against Rambus, adds uncertainty.

I suppose we'll all know the answer before long.

 

[dexvx]

Does this mean I can buy RMBS stock and wait for it to appreciate?

 

[MadRat]

Originally posted by: dexvx
Does this mean I can buy RMBS stock and wait for it to appreciate?

Its all speculation in today's market.

 

[Ice9]

great, so rambus broke a contract and they're going to get away with it. yay legal system!

What contract did they break?

If watching Rambus over the past three years has taught me ANYTHING, it's never to bet on lawsuits, since the company has seemed both sure of victory and hanging over the jaws of defeat on a number of occasions, and analysts (Edelstone in particular) went from praising it as the Second Coming to kicking it out of bed in the span of a couple hours before.

Eh. Edelstone is an analyst like any other. Our own semiconductor analyst rates Micron as a "buy" to this day, even after 2 years and $2B lost. Are you gonna rush out and buy some MU Shares based on his recommendation?

I would also opine that the next case isn't all that clearcut; the dissenting opinion, combined with the earlier finding against Rambus, adds uncertainty

There's nothing IFX can do with it... the CAFC verdict was a clear 100% reversal of the grand "fraud" ruling, and not even the great Ken Starr can twist it into something in IFX's favor. If they try, it'll just be a stream of objections that'll have to be sustained.

If that's the leg you think IFX should stand on, then IFX is doomed, period. Any judge would see it as an attempt to subvert the CAFC by vulturing a dissenting opinion, which no lawyer is going to touch. And I don't think it's ever been tried.

Regardless, IFX will settle. Their shareholders are NOT going to allow this company to potentially lose $3B in triple-awarded damages when they're already in a hole $800M.

 

[FishTankX]

Originally posted by: MrFiTTy
FishTankX, you made 39 posts in this thread alone :Q:Q

Well, i've decided to stop debating here. This thread, and Ice9, kept me in this thread till 1AM, twice. It's horrible for my job. Heh.

But I enjoyed every post, and the learning that came with it... It's so sad that there is so much marketing and financial crap behind the war in hardware. I wish it was like the videocard market, where the best and fastest technology won. Where performance dictated everything. Where good technology ultimatley won out and every round kept you on the edge of your seat. Man.. I wish the memory market would turn out like that. But unfortunatley, the memory market is mainly determined by capacity and price, with memory speed taking a back seat much of the time.

But I bet you there's more information and intelligent debate in this thread than all other RAMBUS related threads from the past year on anandtech put together.

But RAMBUS is something I feel passionatley about. I, at the time earning 100$ a month, spent an extra 150$ on getting a RAMBUS armed system because I believed in the i850's capabilities. I love my P4T-E and I love my RIMMs. Just desperatley need more RAM and I am sick of the rediclous markup on RDRAM over here. (Beijing: 256MB PC800 95$ :Q). If I could get someone to ship me some damn RIMMs from America I would stay with my i850. I'm very sad to sell it.

My present feelings is that if the best technology/performance, not the best lawyers and the lowest price won out, RDRAM would have a decent presence today. But the fact that Team DDR has managed to get Intel to defocus on RDRAM takes away RDRAM's lifeline in the computer market. The i850 and the i850E have managed to fend off team DDR (Until they had to brute force them out with 50% more bandwidth and a significant latency advantage). That in itself is amazing. While DDR has finally caught up, it doesn't mean it's superior technologically. It's just cheaper. And it's cheaper because of a price dumping war. That in itself is nothing to be proud of. I swear though, watch out. In round 2, DDR hits frequency ceiling (Ala P3) RDRAM is gonna come out swinging just like a williamette. And I bet you it'll win victory over victory until it metamorphs into something like a Northwood. Then everyone will Have to accept it.

Edit:Swinging like a williamette in refrence to how the Williamette kept gaining on the suposedly more 'efficent' Athlon.

 

[Ice9]

I agree. There's just nothing compelling about DDR to keep using it long term when better solutions are out there. So far, none of those solutions come from Team DDR. By the time DDR2 hits mass production in 05, there'll be far better stuff out there. Rambus already has technologies that eat DDR2 for lunch anyway.

But either way, Rambus doesn't really need to care about DDR. They're perfectly happy taking a 3% royalty for an inferior technology. It's more money for them if the big boys wanna keep making it.

 

[busmaster11]

It's disappointing how there's so many rdram/p4 fanboys here who are eager to spend all their hard-earned cash just so they can brag about being able to run a couple memory bandwidth benchmarks faster...

 

[Ice9]

It's disappointing how there's so many rdram/p4 fanboys here who are eager to spend all their hard-earned cash just so they can brag about being able to run a couple memory bandwidth benchmarks faster...

It's not about that for me. I can buy whatever technology I want, PC's just aren't a very expensive hobby.

For me, it's about which technology wins long term. If rambus were as crappy as people in here made it out to be, they wouldn't be making engineering milestones like they are in Cray X1's and Alpha EV7's. Since the goal is to get that performance to our desktops someday, I want Rambus to have something to do with it. They simply know more than the "engineers" at JEDEC.

 

[FishTankX]

Originally posted by: busmaster11
It's disappointing how there's so many rdram/p4 fanboys here who are eager to spend all their hard-earned cash just so they can brag about being able to run a couple memory bandwidth benchmarks faster...

Hey, you! Let's see which wins. An i845D or an i850 when running at 2.8GHZ. I could do that, you know. The three choices at a time were an MSI i845D board, an i845 board, or an i850. I picked option number 3 and i've never regreted it for a moment. It's just that China inflates the prices of not such common stuff into the sky.