New RDRAM chipset "beats DDR by 50%" claim

[FishTankX]

Originally posted by: SexyK
Well it looks like I'm a little late to the party on this thread, but I'll throw in my stance.

1) I think it's absurd that so many (seemingly) well educated people can slam Rambus constantly over something they apparently know very little about.
2) In all reality, the price difference between RDRAM and DDR is non-existent at best. Top DDR400 modual (the only thing that really challenges RDRAM) is extremely pricey.
3) Most importantly, I picked up a Abit TH7-II when the very first S487 P4's were coming out, well over 18 months ago. In that time I could have gone through a i845, i845pe, and a GB board and still have never completely eclipsed the speed of my i850 board. I think that speaks for itself in terms of the engineering happening over at Rambus.

I really can't understand the backlash against RDRAM, especially from a group of people who preport to be obsessed with the ultimate in computer hardware.

kramer

It's because of the misconception that RAMBUS is evil.

If Intel refused to extend AMD's X86 liscensce they would probably get the same treatment.

If intel charged a royalty on every AMD Processor just because it used X86 it'd be the same.

The rage isn't against the RAMBUS technology. I'm sure every person admits it's a good technology.

The rage is against RAMBUS, the company, and what they've done. And many people refuse to be ascosciated with such a litigeous company.


I'm a RAMBUS user, and i'm happy with my P4T-E even though i'm pissed that ASUS couldn't use decent clock generators. Running synchronous 133/1066 would have been great. I'm defecting to the SiS655 party because it'll be a boost in performance, and i'll be able to purchase cheap DDR266 memory and get near the speed of the i850E for a fraction of the price. You can't beat 120$ a gig, can you? Heh.

But that's because I need more RAM. Not because RAMBUS is bad. I like RAMBUS RAM. It's performance is great. But I cannot afford the high cost of owning RAMBUS and a decent pool of RAM at the same time.

 

[FishTankX]

Ice9:Where's your Toms SiS655 article? I do admit that granite bay is slower. But the SiS655 is a different story.

 

[DX2Player]

Not shure what you do but for 99.8% of the people 512 ram is enough

 

[Ice9]

Ice9:Where's your Toms SiS655 article? I do admit that granite bay is slower. But the SiS655 is a different story.

Maybe I fatfingered something here, but the SIS655 article was posted right here at anandtech, not at toms. I don't believe TomsHardware has done a benchmark with the SIS655 and a 3.06 processor.

At any rate, the anandtech benchmark test is here in all its 2.26ghz <giggle> splendor:

http://www.anandtech.com/mb/showdoc.html?i=1786&p=7

 

[FishTankX]

DX2 player: Sim City 4 and Unreal 2 eats RAM like candy! Must have more RAM! *GASP!*
I mainly do gaming.

 

[FishTankX]

Ice9:The i850 chipset was still beat by DDR333 chipsets at high processor speeds when running at PC800, even though it has less bandwidth. I will have to assume the SiS655 retains it's leadership position at higher speeds until I see benchmarks pointing otherwise. Sorry.

 

[DX2Player]

I game too, Im sticking to my guns that hardly anyone needs more than 512 right now
We here at anandtech hardly represent the norm, go to specialized computer sellers like alienware and voodoopc and their gaming rigs dont use more than 512 that must mean something.


BTW you know how old pc800 ram is thats like comparing sdram, but regardless ill conceed the spot to the new SIS R655 chipset. But honestly I dont think right now would be the right time to buy a new system with so many changes right around the corner

 

[JackBurton]

Not shure what you do but for 99.8% of the people 512 ram is enough

I guess I'm in the 0.2% minority.

Ice9, great posts! As for me, I never gave a sh!t about the background of these two technologies. I just live by benchmark and real word performance. Whichever is faster, I'll buy. That being said, I have DDR machines and RDRAM machines.

 

[FishTankX]

I'm refering to how scaling did nothing for the i850 when the SiS645 with DDR333 CAS 2.0 finally smacked it into the ground. Same thing with the SiS655 and the i850E. Is it just me, or is amusing how SiS seems to bury all the Intel RDRAM chipsets sometime or another with their 'More memory bandwidth than necescescary' aproach? LOL.


SiS is just clobbering intel on the ultra high end performance right now. I'm hoping canterwood will win them back but SiS will probably have reengineered their memory controller by then, and the fact that SiS tunes for performance over complete stability means that they'll always have some sort of performance edge. Not to mention price edge. SiS has really shocked me, going from integrated slow as hell chipsets to blasting away the most powerful Intel chipsets in the performance sector.

By the way, did you see how NFS4 got the stripped down version of the SIN1394XP? And was extremley happy with it? Even with DDR333 you can match the i850E, although not beat it. And the memory is really cheap. I'm just gonna get a gig of RAM, mostly because i've learned my lesson. Always upgrade with double the amount you need right now, if you don't want to add memory all the time. I thought 256 would be enough for a year, this janurary. I was *wrong*!

Anyways, I know how PC800 is still kickin it around. I mean, I'm typing on a PC800 based computer now, and I love it. It's just that PC800 is way more expensive than DDR266 right now and I can't afford to break the 512MB barrier with PC800 as this computer's main RAM.

 

[FishTankX]

Note:I know you all will pelt me with statments of how "PC800 is the same price as DDR266". Please understand I live in China where that is definatley not the case. Over here, you can buy namebrand 256MB of DDR266 for about 30$. The same amount of PC800 costs 70$. I'm sorry, but that is not price parity! And the fact that i'll be getting a *performance boost* by defecting to DDR266 just makes it irresistable.

 

[cmdrdredd]

Originally posted by: Ice9
Beware. Blatant facts follow.

Ice9: SiS655 armed with DDR400 beats the i850E. That's the conclusion of Anand's review. It's lost the performance crown. There's only one benchmark at which RAMBUS still has a clear performance lead, and that's an isolated 3d workstation benchmark. In almost every other situation the SiS655 with DDR400 will beat the i850E.

Oh god, this one is too easy

Now.

Is this the notorious "review" of the SiS655 dual-DDR400 board that they used <chuckle> a P4 2.26 as a testbed? We had this discussion on Hardforums. My id there is 'kweckstrom'. On Page 8 of the Notorious Rambus Thread from Hell(TM), I made these points:

Now, Anandtech:
http://www.anandtech.com/mb/showdoc.html?i=1784&p=7

What the HELL is going on here? A 2.26ghz CPU? How are you going to stress these new and great chipsets with an OLD CPU? I don't want to see how Ram performs on a CPU that's slow in comparison to the latest and greatest 3.06 Hyperthreaded CPU... USE THE RIGHT PROCESSOR if you're going to determine the merits of the latest and greatest board!! How can you EVEN find this review REMOTELY accurate when you're testing "speed and throughput" of a Dual DDR rig or RDRAM rig by crippling the entire test with a sub-par processor?

We know they HAVE a 3.06: http://www.anandtech.com/cpu/showdoc.html?i=1746&p=10

They use an 850E with RDRAM for *THAT* test...
We know they have Granite Bay boards.
We know they have dual DDR boarsd from SIS.
We know they have EVERYTHING they need to do a proper benchmark, but they don't.

Hey, now... has ANYONE done a P4 3.06 comparison of RDRAM vs. Granite Bay dual DDR (or any other dual DDR chipset for that matter) besides TomsHardware (who judged RDRAM as superior)? If we're gonna see which memory technology is better, let's STRESS THE DARN MEMORY SUBSYSTEM!

Interestingly enough, "Anderu" from that forum went on to Email the author of that "review" (Evan Lieb). His reply was:

First of all, this Kweckstrom fellow is sorely misinformed about Rambus
the company. It is a well known FACT that Rambus committed fraud when
they lied to JEDEC by filing for certain SDRAM-like patents behind their
back. Dare I say Kweckstrom defends Rambus the company because he is an
active Rambus investor?

Secondly, we use a 2.26GHz CPU because:

1) It reflects the approximate CPU speed most OEM PCs have been shipping
with in the past few months.
2) It happens to be a perfect CPU for enthusiasts (overclockers, gamers,
etc.), and therefore is an excellent CPU speed to show the superiority
of dual DDR memory over Rambus memory.
3) In the benchmarks we've run at 2.8GHz+ (and slightly higher, like
2.85GHz), the variance in scores simply gave us absolutely no reason to
use such a high speed CPU. I admit we haven't tested as high as 3.06GHz,
but I very much doubt the scores would change noticeably because of an
extra ~ 200MHz.

The reasons listed above do not apply to a 3.06GHz CPU, as this CPU does
not reflect the average CPU speed the majority of OEM PC buyers receive
(from Dell, HP, IBM, etc.), and it doesn't even reflect what the
majority of hardcore enthusiasts (overclockers, gamers, etc.) purchase
either.

Of course, you can read my responses on those pages, but i'll give you the gist of it.

A 2.26ghz CPU will NOT stress a memory subsystem the way a 3.06ghz CPU will. *ANY FOOL KNOWS THAT*. If you are trying to claim a victor in memory technology superiority, USE THE RIGHT CPU FOR THE JOB, PLEASE.

Oh wait, I don't think that chipset supports Hyperthreading, does it... I think they conveniently left out that little fact.

I'm not sure what this Evan Lieb guy had to gain from using such a junky CPU for this test, but we KNOW from previous Anandtech reviews that they have the correct tools to PROPERLY benchmark this memory subsystem.

So hey, does something smell funny to you?

Now you're going too far. Who are you to question the benchmarks and reviews donw on AT or any other site? You obviously don't do them so you should not be questioning the validity of the numbers. If it were your review then sure, but it is not. You have no place telling people how to review the chipset and motherboards. How do you know that a 3Ghz P4 won't still be better with the DC DDR? And the SIS655 does support HT FYI.

 

[cmdrdredd]

and every RDRAM board I used didn't overclock well at all

 

[Dug]

and every RDRAM board I used didn't overclock well at all

lol

 

[dexvx]

Now you're going too far. Who are you to question the benchmarks and reviews donw on AT or any other site? You obviously don't do them so you should not be questioning the validity of the numbers. If it were your review then sure, but it is not. You have no place telling people how to review the chipset and motherboards. How do you know that a 3Ghz P4 won't still be better with the DC DDR? And the SIS655 does support HT FYI.

So you're implying that the reviewers know everything there is to know about the product they review? I'm not saying pointing any fingers here, but it doesnt exactly take a genius to receive a testbed and perform tests on it (premade tests), and write down the results. It goes to the next level if a reviewer theorizes why the numbers are there in the first place and how different variables can offset the values. What this Ice9 guy says a more powerful processor requires more memory bandwidth, which will cause i850e to pull further away from DDR/DC-DDR in benchmarks. Although a more powerful processor does require more memory bandwidth, I also believe that the increased speed will not skew the overall result as much as he thinks (then again HT might make a difference).

If we take every review site's words as the holy word, then I'd be believing Van Smith saying in his CosbyBench (I cant remember the exact name) is the godset of all benchmarkes, showing that a Celeron p4 1.8Ghz/128KB is faster than a p4 2.53Ghz/512KB, which also happens to be about half as fast as an XP 2000+.

 

[dexvx]

Originally posted by: cmdrdredd
and every RDRAM board I used didn't overclock well at all

You sound like you went through a lot.

Well for starters theres the Abit TH7-II, which dominated the P4 overclocking scene even long before the Northwood popped out. If it werent for the fact that many TH7-II's were paired with crappy DRCG's and the lack of HT support, it would probably be one of the best boards ever.

Then theres the P4T-E/P4T-533C, only bad thing is that it is capped at 156Mhz FSB, most i850E boards are capped at 156Mhz FSB. But the Asus P4T533 is capped at 200Mhz FSB.

Then theres the fact that most all Samsung PC800 16 device RDRam manufactured after 2001 can reach PC1066 with little difficulty. Some of the newer revision Samsung 32bit PC1066 can top at at PC1333. At 166FSB with DC-PC1333, it can even give DC-DDR400 a run for its money.

 

[cmdrdredd]

but at a high price...I looked at the 32bit PC1066 512MB Samsung and it's $299 at pricewatch!

YIKES

 

[Ice9]

Now you're going too far. Who are you to question the benchmarks and reviews donw on AT or any other site? You obviously don't do them so you should not be questioning the validity of the numbers. If it were your review then sure, but it is not. You have no place telling people how to review the chipset and motherboards. How do you know that a 3Ghz P4 won't still be better with the DC DDR? And the SIS655 does support HT FYI.

Going too far? That's funny. I question the review because I feel it's just plain *wrong*.

If you think this Evan Lieb guy covered all the bases in claiming DDR400 the victor over RDRAM based on a *2.26ghz CPU*, you're crazy. I have a 3.06HT machine. I've done my own benchmarks. I've compared them to popular websites. 2 year old 850E still wins.

And if the SIS655 does support HT, and HT has been shown to make a dramatic difference on the P4 3.06, then for GOD'S SAKE, TAKE THAT INTO ACCOUNT when reviewing a chipset and calling it "superior".

Maybe people should be more accurate and say that "DDR400 is superior when you use slower processors that don't support Hyperthreading". If you ask me, that's all they have the *RIGHT* to say with this poor review that doesn't cover all the bases.

It's my speculation that this evan lieb guy is a child who simply can't afford a P4 3.06HT, and anand is too busy playing Unreal to lend it to him If he even trusts him enough to lend it to him at all.

 

[Ice9]

Originally posted by: FishTankX
I'm refering to how scaling did nothing for the i850 when the SiS645 with DDR333 CAS 2.0 finally smacked it into the ground. Same thing with the SiS655 and the i850E. Is it just me, or is amusing how SiS seems to bury all the Intel RDRAM chipsets sometime or another with their 'More memory bandwidth than necescescary' aproach? LOL.

I wonder why SIS is moving away from DDR and moving towards quad channel PC1200 then!

If DDR finally smacked RDRAM into the ground, why go back to RDRAM? *smirk*

 

[fkloster]

I have owned PC-800 for 4 years now and it still is a competitive memory solution. I used it in my veritable Asustek P3C-E (i820) back when most people were using the popular 440BX w/PC-100/133

 

[Ice9]

Here's an interesting article:

http://www.planetanalog.com/story/OEG20021016S0037

I can't read these kinds of articles without re-iterating just how brilliant I think Rambus engineers are.

I mean really... how is JEDEC or ANYONE going to compete with that??

 

[DX2Player]

Originally posted by: FishTankX
I'm refering to how scaling did nothing for the i850 when the SiS645 with DDR333 CAS 2.0 finally smacked it into the ground. Same thing with the SiS655 and the i850E. Is it just me, or is amusing how SiS seems to bury all the Intel RDRAM chipsets sometime or another with their 'More memory bandwidth than necescescary' aproach? LOL.

Ya you seem to have lost track of the inital post, SIS is the very one thats producing the R659 with 9.6GB/sec bandwidth so it looks like it will be smacking its own DDR solution into the ground with uber bandwidth.


BTW for those who think RDRAM cant oc well, Here is 172 FSB, and Here is 165 FSB both are using the newer 32bit RDRAM as well.

Pics of RDRAM and Samsungs roadmap here note that recently DDR2 has been pushed back to 2005 actually, not really good as RDRAM will have no compatition.

 

[DX2Player]

Originally posted by: cmdrdredd
but at a high price...I looked at the 32bit PC1066 512MB Samsung and it's $299 at pricewatch!

YIKES

Thats funny because i just went onto pricewatch and foud it for $214; besides normal people would just buy 2x256 sticks for $86 each

 

[FishTankX]

Originally posted by: Ice9

Originally posted by: FishTankX
I'm refering to how scaling did nothing for the i850 when the SiS645 with DDR333 CAS 2.0 finally smacked it into the ground. Same thing with the SiS655 and the i850E. Is it just me, or is amusing how SiS seems to bury all the Intel RDRAM chipsets sometime or another with their 'More memory bandwidth than necescescary' aproach? LOL.

I wonder why SIS is moving away from DDR and moving towards quad channel PC1200 then!

If DDR finally smacked RDRAM into the ground, why go back to RDRAM? *smirk*

*Sigh* You are the most ardarent RAMBUS suporter I have ever seen.

SiS and the entire market was in a rare position, where SiS saw a way to match/beat the i850E with cheap DDR. They tried. They scored, big time. They matched/beat(? Sure not sure on beat yet, seeing that they used a 2.26).But, because they had them beat on the price side of the equation, while matcing them on the performance, they still had them beat. 1 win and 1 draw still equals win. Also, they won *bigtime* for the people who need massive amounts of RAM as 512MB PC1066 modules can run for 300$. They won me, and trust me, i'm a *strong* RAMBUS suporter. I believe RAMBUS is the future. But the future isn't now and my wallet is screaming at me for spending 190$ on my P4T-E and 120$ on my 128X2 RDRAM sticks back in janurary. I need something *cheaper*.

Now, if they come up with a way to get quad channel RDRAM (either dual RIMM 4800 or if they're really serious about 4X16) and it smacks DDR into the ground, that's fine. But that board won't earn them a ton of money. The majority of the market in DDR. What it will do is earn them the prestiege of having beat the canterwood, and a *very* good workstation chipset.

I still don't understand what's so hard to understand about how the board fits into the scheme of things.

What's the point of having the Geforce4 Ti4200 if the Ti4600 blows it out of the water? It's a lower cost sollution that meets the demand at the market at significantly reduced prices, from the Ti4600. And I wonder, which does Nvidia make more money from? Somehow have the feeling that the R659 will be in a similar situation. It will be an expensive, profitable high performing sollution relegated into a niche market. That's my opinon. You have the right to disagree.

Now, from what I can see with this new chipset it will have to support alot of RIMM slots because unless they have 64Bit RDRAM (Unlikely, Tomshardware definatley had a roadmap and I believe that 64 bit RDRAM wasn't suposed to be on the market until 2004) they'll need alot of slots to even get close to matching the chipset's defined memory limit of 16GB and having any less than 8 slots would only provide the user with a limited amount of memory upgrades. 4 slots would provide you 2 upgrades with Dual RIMM4800 or 1 upgrade with quad PC1200. 8 slots would provide double that. Still not very atractive with the PC1200 sollution.

I don't see this as being atractive for the average user just wanting a stable high performance sollution, not necescescarily bleeding edge. For that the SiS655 will do fine. Thus, they're not "Smacking their own chipset into the ground", persay. Just having all bases covered. Offering people choices. The more 'Bases' SiS has covered in the motherboard market, the better they'll do. I think they're being just like ATi, eyeing the highend of the market. Once canterwood comes out they'll need something to smack Intel's highend back into the ground. And quad channel RDRAM just might do it, even if it's unpractical for the average user and cumbersome to upgrade (in PC1200 16 bit modules configurations).

I guess in the industry speedcrown means all. Just look at the i850E chipset. It isn't very affordable for the average user. Heck, most Dell people are still on the i845D or E or PE. It doesn't have very good marketshare in terms of volume. But hell, it's on top! It's used as *The* platform in most all videocard benchmarks with Intel chipsets. Why? Because it's got the speedcrown. (They can still definatley beat them in that one benchmark..) And if SiS wants to hold the speedcrown they'll have to have an answer to the canterwood, something to bring them back into performance leadership. The R659 looks like a good candidate.

Oh, by the way Ice9. If I finally do upgrade to a SiS 655 PC266 rig, I'll be able to achieve 97% your system's memory subsystem performance at 1/3 the RAM cost. It's a good deal, no? I like good deals. You go on saying RAMBUS is the fastest. That doesn't mean it'll make it an ideal sollution for anyone out there. For the majority of us RAMBUS may be the best out there, but it's not the best out there for us.

P.S. Fkloster, sorry to jump ship, but i've learned my lesson. The $$$ just worthit for a 'Higher performance' system, espically when you're eighteen with a lousy job. I wish I had went with the i845D. That way I could directly jump to the SiS655. Besides, the performance difference isn't that big. Heh.

 

[fkloster]

P.S. Fkloster, sorry to jump ship, but i've learned my lesson. The $$$ just worthit for a 'Higher performance' system, espically when you're eighteen with a lousy job. I wish I had went with the i845D. That way I could directly jump to the SiS655. Besides, the performance difference isn't that big. Heh.

No worry's Fish....I already own 4-256mb of PC-800 so I'm gonna give this Sis-659 gig a spin....

 

[Ice9]

*Sigh* You are the most ardarent RAMBUS suporter I have ever seen.

I'm also the most factual in here.

SiS and the entire market was in a rare position, where SiS saw a way to match/beat the i850E with cheap DDR. They tried. They scored, big time. They matched/beat(? Sure not sure on beat yet, seeing that they used a 2.26).But, because they had them beat on the price side of the equation, while matcing them on the performance, they still had them beat.

I think that's silly to assume without stressing the chipset with a *FAST* processor. I am NOT stupid enough to believe that this chipset will behave in an identical manner with a 3.06ghz P4 with hyperthreading.

You may choose to BLINDLY believe this. I am not complacent enough to do so. I want to see how fast this chipset performs on Intel's top of the line CPU. Nothing less will convince me of its "superiority" over RDRAM.

1 win and 1 draw still equals win. Also, they won *bigtime* for the people who need massive amounts of RAM as 512MB PC1066 modules can run for 300$. They won me, and trust me, i'm a *strong* RAMBUS suporter. I believe RAMBUS is the future. But the future isn't now and my wallet is screaming at me for spending 190$ on my P4T-E and 120$ on my 128X2 RDRAM sticks back in janurary. I need something *cheaper*.

And that's fine... *AS LONG AS YOU UNDERSTAND WHY DDR IS CHEAPER THAN RDRAM*. I've outlined it in several posts in this thread.

Now, if they come up with a way to get quad channel RDRAM (either dual RIMM 4800 or if they're really serious about 4X16) and it smacks DDR into the ground, that's fine. But that board won't earn them a ton of money. The majority of the market in DDR. What it will do is earn them the prestiege of having beat the canterwood, and a *very* good workstation chipset.

Again, that's FINE. I'm more concerned with the semiconductor industry using the right tool for the job. DDR is out of gas, and has been for a long time. These little "incremental victories" over a 2 year old Rambus technology are just plain LAUGHABLE to me!

I still don't understand what's so hard to understand about how the board fits into the scheme of things.

Again, the right tool for the job.

What's the point of having the Geforce4 Ti4200 if the Ti4600 blows it out of the water? It's a lower cost sollution that meets the demand at the market at significantly reduced prices, from the Ti4600.

Right, but don't call it "better" simply because it's "cheaper".

And I wonder, which does Nvidia make more money from? Somehow have the feeling that the R659 will be in a similar situation. It will be an expensive, profitable high performing sollution relegated into a niche market. That's my opinon. You have the right to disagree.

I don't disagree but Rambus *WASN'T ENGINEERED* to be in a niche market. It's ARTIFICIALLY expensive. Read my prior posts. Everything Rambus engineers is engineered to be AFFORDABLE and CHEAP TO MASS PRODUCE. The fact that it's expensive is the result of LACK OF COMPETITION. There's only 2 companies making the stuff, as opposed to 50 million others producing DDR and running the semiconductor industry into the ground.

Now, from what I can see with this new chipset it will have to support alot of RIMM slots because unless they have 64Bit RDRAM (Unlikely, Tomshardware definatley had a roadmap and I believe that 64 bit RDRAM wasn't suposed to be on the market until 2004) they'll need alot of slots to even get close to matching the chipset's defined memory limit of 16GB and having any less than 8 slots would only provide the user with a limited amount of memory upgrades.

Please read my prior posts. You are missing a lot of critical details by ignoring them. Particularly the one with Samsung's roadmap CLEARLY showing 64 bit RIMM's arriving 3rd quarter 03, coinciding with SIS new RDRAM offering. It's on page 28 of Samsung's presentation, and includes a diagram of the module itself.

I guess in the industry speedcrown means all. Just look at the i850E chipset. It isn't very affordable for the average user.

Funny, Dell is selling MILLIONS of them. The Dimension 8250 is one of their top selling models, and it's based on *gasp* the 850E!

Heck, most Dell people are still on the i845D or E or PE. It doesn't have very good marketshare in terms of volume. But hell, it's on top!

You wouldn't say this if you knew the facts. Go to Dell's own forums. http://delltalk.us.dell.com

Take a poke around and tell me how many people are buying 8250 Dimensions with RDRAM Please. Do some research before you blindly make statements that simply aren't true.

Oh, by the way Ice9. If I finally do upgrade to a SiS 655 PC266 rig, I'll be able to achieve 97% your system's memory subsystem performance at 1/3 the RAM cost. It's a good deal, no? I like good deals. You go on saying RAMBUS is the fastest. That doesn't mean it'll make it an ideal sollution for anyone out there. For the majority of us RAMBUS may be the best out there, but it's not the best out there for us.

Again, read my posts on WHY RDRAM is so expensive.

You're rooting for the wrong team It doesn't really matter anymore which one wins. At least, not from Rambus' perspective.

As it stands right now, Rambus has the right to collect a 3% royalty on SDRAM and DDR. There are only 3 holdouts left to paying this royalty, and they too will succumb after Infineon lost in court. Rambus is the rightful patent holder for SDRAM and DDR technology.

So... If you want DDR, you will now have to pay more for it, especially when you consider the royalty rate of RDRAM is only 1.5%. It will only be a matter of time before all these money-losing dram houses start realizing that it's cheaper to manufacture RDRAM-based solutions.