Rambus Wins Pretrial Patent Ruling

[HardOCP News]

Rambus has won a pretrial ruling that basically says memory makers infringed on an element of a patent covering computer memory. The case is scheduled for trial in January.

The ruling means Whyte “already found one claim for Rambus that they won’t have to argue,” Jeff Schreiner, an analyst at San Diego-based Capstone Investments who doesn’t own Rambus shares, said today in a phone interview. “It starts to narrow the focus of the trial. You don’t get jurors lost in the complicated nature of the patents.”

 

[Nahu1337]

....Wow.

 

[PurduEE]

Looks like the FTC is seeking certiorari from the Supreme Court to review the DC Circuit's decision re the whole JEDEC and deceptive conduct.

http://www.ftc.gov/opa/2008/11/rambus.shtm

Of course, this trial will likely be completed prior to any decision by the Supreme Court to grant cert.

 

[Ice Nine]

I doubt it will fly. There's way too much documentation proving the infringers knew all about Rambus technology under NDA. With signatures on both sides, I don't see how the FTC could make the claim that no one knew about their technology and prove there was any kind of subversion going on...

I'd be more interested in hearing JEDEC's excuse for barring Rambus from presenting their technology for standardization consideration. I think it's because they DID all know about it and wanted to prevent royalties going to them (as you read in the "Deadly Menace" memo).

 

[Orinthical]

Can RAMBUS just die already?

 

[Jinto]

If they have a valid case I don't see whats wrong with them collecting their due.

 

[RangerXML]

Can RAMBUS just die already?

Nope, infact this looks very grim for just about everyone else. Except AMD/ATI and Sony who already struck deals with Rambus.

 

[MrGuvernment]

bottom line is, if Rambus had the patent, they had the patent, i dont care how much you hate rambus, doesnt mean they should have their items stolen and used for free just like no one else should

 

[Ice Nine]

Nope, infact this looks very grim for just about everyone else. Except AMD/ATI and Sony who already struck deals with Rambus.

Anyone who was smart settled up with them to get a cheap royalty rate. It's the stupid ones who signed NDA's then tried to steal it through JEDEC.

They're getting what they deserve.

 

[ICOM]

FUK!

 

[RangerXML]

Anyone who was smart settled up with them to get a cheap royalty rate. It's the stupid ones who signed NDA's then tried to steal it through JEDEC.

They're getting what they deserve.

While I 100% agree with you, in the short run its really, really gonna hurt the tech industry (picture ATI is the only enthusiast video card maker); in the long run we could have better technologies. We could be looking at the dark age of technology with the return of the Sith, lets just hope our Luck Skywalker isn't such a whiny little bitch. (sorry for the Star Wars reference)

 

[Ice Nine]

While I 100% agree with you, in the short run its really, really gonna hurt the tech industry (picture ATI is the only enthusiast video card maker); in the long run we could have better technologies. We could be looking at the dark age of technology with the return of the Sith, lets just hope our Luck Skywalker isn't such a whiny little bitch. (sorry for the Star Wars reference)

Well, that's unlikely to happen.

If the mem manufacturers can't make memory, they can't stay in business. If NVIDIA can't sell video cards/gpus, they can't remain in business. They'll have no choice but to settle for whatever terms Rambus wants or risk going under.

These companies might be stupid for doing what they tried to do to Rambus in the 90's, but I sincerely hope they aren't stupid enough to fold because they don't want to pay a royalty they should have been paying in the first place.

If anything, this has the potential to help the industry move forward. One thing we already know is that XDR & Flexphase get around all the problems that face memory architectures today (particularly trace length matching, which is one of the reasons motherboards aren't as cheap as they used to be).

If they have any brains over there at Rambus (and chances are they do), they will make DDR/GDDR insanely expensive to license - but make XDR/Flexphase licensing rates insanely attractive. This will push the memory manufacturers towards a far superior technology.

DDR2 & DDR3 is out of gas for all intents and purposes. We need something new, and XDR would fit the bill nicely - especially for video cards

 

[rflcptr]

It'd be nice to have i7s with XDR.

 

[maleficarus]

Well, that's unlikely to happen.

If the mem manufacturers can't make memory, they can't stay in business. If NVIDIA can't sell video cards/gpus, they can't remain in business. They'll have no choice but to settle for whatever terms Rambus wants or risk going under.

These companies might be stupid for doing what they tried to do to Rambus in the 90's, but I sincerely hope they aren't stupid enough to fold because they don't want to pay a royalty they should have been paying in the first place.

If anything, this has the potential to help the industry move forward. One thing we already know is that XDR & Flexphase get around all the problems that face memory architectures today (particularly trace length matching, which is one of the reasons motherboards aren't as cheap as they used to be).

If they have any brains over there at Rambus (and chances are they do), they will make DDR/GDDR insanely expensive to license - but make XDR/Flexphase licensing rates insanely attractive. This will push the memory manufacturers towards a far superior technology.

DDR2 & DDR3 is out of gas for all intents and purposes. We need something new, and XDR would fit the bill nicely - especially for video cards

Exactly! RDRAM was the better technology over DDR but the insane hate for rambus stopped the push forward and now we see DDR2/3 more or less putten along without really advancing like it should have by now.

 

[Trimlock]

Exactly! RDRAM was the better technology over DDR but the insane hate for rambus stopped the push forward and now we see DDR2/3 more or less putten along without really advancing like it should have by now.

I'm sorry but I do remember DDR1 smacking the ever long crap out of RDRAM in performance and price.

Also, until I see memory bandwidth become a factor in our performance today, which it is not, I'll glady accept XDR as a new contender to DDR.

Question: What advantages does XDR have over DDR5? I've seen GDDR5 in action and its awsome, but haven't seen XDR outside of anything other then the PS3.

 

[Ice Nine]

I'm sorry but I do remember DDR1 smacking the ever long crap out of RDRAM in performance and price.

Also, until I see memory bandwidth become a factor in our performance today, which it is not, I'll glady accept XDR as a new contender to DDR.

Question: What advantages does XDR have over DDR5? I've seen GDDR5 in action and its awsome, but haven't seen XDR outside of anything other then the PS3.

no... RDRAM beat the snot out of DDR1. You had to wait well into DDR2's lifecycle to get better bandwidth, and by then XDR was already a reality.

To name an advantage - Flexphase. The reason motherboards keep increasing in price is because of how much it costs to design them. With DDR/DDR2/DDR3 (and I assume DDR5, unless they licensed flexphase from Rambus) every trace from the memory module pin to the memory controller (either on chip or separate) had to be the EXACT same length.

Rambus developed a differential signaling technology to overcome that, meaning you could get mean bandwidth but still have differing trace lengths - it makes motherboard design far easier (and less costly).

Of course, if you don't think that Memory should be faster, then sure - cheerlead for the slower stuff that makes a lot of sense for an enthusiast, or someone who calls themselves one.

 

[FLECOM]

Exactly! RDRAM was the better technology over DDR but the insane hate for rambus stopped the push forward and now we see DDR2/3 more or less putten along without really advancing like it should have by now.

ya but it was too expensive and DDR was good enough at the time...

 

[Ice Nine]

ya but it was too expensive and DDR was good enough at the time...

Well, that means you need to believe that it was REALLY expensive to produce.

But if it was REALLY expensive to produce, how did it make it into the N64? Or the PS2?

RDRAM didn't cost any more to manufacture than SDRAM (the original, not even DDR). The prices were kept artificially high to prevent its adoption and to squeeze intel.

If you go back and look at the timeline, it's all very obvious.

 

[RangerXML]

I'm sorry but I do remember DDR1 smacking the ever long crap out of RDRAM in performance and price.

Where you maybe mixing RDRAM with SDRAM?

 

[ICOM]

Well, that's unlikely to happen.

If the mem manufacturers can't make memory, they can't stay in business. If NVIDIA can't sell video cards/gpus, they can't remain in business. They'll have no choice but to settle for whatever terms Rambus wants or risk going under.

Not so fast. The scumbag RAMBUS patent bastards haven't won anything yet. Lets wait and see what the courts decide. Don't confuse RAMBUS with a company that develops technology.

 

[ssrat]

It was WAY to expensive for anyone at the time except for businesses, had to be bought in pairs, and was also hard to get, since the price scared regular stores from carrying it.

Add to that it was insanely linked to some of the most expensive boards at the time, and you can see that even without them violating JDEC (at least in spirit) it was not what your average buyer could afford

As for the performance issue, yes it WAS faster, but if you did a price/performance check DDR won out especially since you only needed ONE to get the machine working

 

[TechLarry]

Just how the hell does this company keep getting over like this. I'm totally amazed...

 

[ICOM]

Just how the hell does this company keep getting over like this. I'm totally amazed...

What? You haven't heard of the law firm RAMBUS, RAMBUS and RAMBUS?

RAMBUS chases ambulances. Let's hope this is the last gasp for the scum company.

 

[qwakrz]

JEDEC meeting to design standards

JEDEC "Well guys, we have agreed that we need to use identical trace lengths and that this new design for a memory subsystem will work great, and we can all use it to our benefit so we just all need to sign up to it and..."

RAMBUS "Hang on guys while I answer the call of nature, be back in 5"
....

RAMBUS on phone to company "Hey frank, nip down the patent office with this fax will you, I need it entered while its fresh in my mind"


......

JEDEC "ok, you just need to sign here Mr Rambus"

RAMBUS "Ok, all signed. Now due to this using stuff that our company has patented you will have to start paying us royalties, we will make them nice and cheap but sorry guys"

--------

Rambus have tied the market up with patents that are now hindering the production of fast memory and also hindering development of future technology.

Their design of RDRAM was faster than SDRam but had the same disadvantage as the original P4 chips in that if the memory locations were not sequential then it really hit the performance of the chip making it a lot slower in general use.

 

[Ice Nine]

So what is that XDR stuff that's in Cell? lol

You mean to tell me they didn't develop that?

Don't confuse development with manufacturing. When a software company develops something, it's sold to a publisher which does the duplication, marketing and sales - and the software developer gets a royalty on sales.

When a musician writes a song, he goes to a music label to do the duplication, marketing and sales. When people buy the CD, the musician gets a royalty.

When Rambus develops a memory/interface technology, they go to the memory manufacturers to do the manufacturing, marketing and sales. When a product uses their technology, Rambus gets a royalty.

See how simple it is?

 

[Ice Nine]

JEDEC meeting to design standards

JEDEC "Well guys, we have agreed that we need to use identical trace lengths and that this new design for a memory subsystem will work great, and we can all use it to our benefit so we just all need to sign up to it and..."

RAMBUS "Hang on guys while I answer the call of nature, be back in 5"

Yep - they added the patent *claims* (not patents) because JEDEC was cherry picking them. Rambus has every right to add patent claims to an invention - that's patent law. Rambus knew they were being robbed.

Rambus have tied the market up with patents that are now hindering the production of fast memory and also hindering development of future technology.

Their design of RDRAM was faster than SDRam but had the same disadvantage as the original P4 chips in that if the memory locations were not sequential then it really hit the performance of the chip making it a lot slower in general use.

Yeah. That XDR stuff. The industry sure is hindered.

And no - RDRAM on the P4 was the fastest memory technology it had for a LONG time. Go back and read all the benchmarks. What you've said right there is outright BS. It took a LONG time for DDR to catch up to RDRAM, and by then Rambus already had XDR prototyped.

It's the memory manufacturers own foolishness that held back the industry.

 

[Nenu]

As I remember, only the fastest RDRam bested SDRam.
The slower RDRam versions were higher bandwidth than SDR but the latency killed performance.
Rambus also ran pretty hot and cost too much.
(I seem to remember early modules were easy to damage too?)

What nailed the door shut for Rambus was the lack of consumer adoption (it only being available on Intels P4 didnt help them and then Intel dropping Rambus when the going got tough) + lack of progression to bring heat down and performance up while reducing the price.
Also what pissed techy consumers was that it prevented better SDRam offerings coming from Intel while they were tied to Rambus so general performance dropped unless you had deep pockets and didnt mind running a very hot system.

 

[Ice Nine]

As I remember, only the fastest RDRam bested SDRam.
The slower RDRam versions were higher bandwidth than SDR but the latency killed performance.
Rambus also ran pretty hot and cost too much.
(I seem to remember early modules were easy to damage too?)

What nailed the door shut for Rambus was the lack of consumer adoption (it only being available on Intels P4 didnt help them and then Intel dropping Rambus when the going got tough) + lack of progression to bring heat down and performance up while reducing the price.
Also what pissed techy consumers was that it prevented better SDRam offerings coming from Intel while they were tied to Rambus so general performance dropped unless you had deep pockets and didnt mind running a very hot system.

Incorrect. RDRAM was faster than both SDRAM and DDR by a large margin for a long time.

RDRAM did not run hot at all. They simply included heat spreaders to evenly distribute the heat across the module (no "hot spots"). Somehow that was made into FUD.

What nailed the door shut for RDRAM (not Rambus) was the memory manufacturers colluding to keep the prices high so no one would WANT to adopt it. It was an illegal practice, and if you haven't noticed these companies have all been fined millions of dollars and some executives are doing jail time. No one from Rambus is doing jailtime because they haven't done anything illegal - but it always makes me laugh when everyone points the finger and calls Rambus the bad guys

Go back and read the reviews of P4's running RDRAM. Go back and see just how "hot" they were. They were faster than SDRAM, they were faster by DDR, and they didn't run any hotter. And the cost was artificially high because the memory manufacturers didn't WANT RDRAM to be adopted - they illegally squashed it.

 

[Nenu]

I have no interest in revisiting the topic personally, I am recalling my experience as a retailer/manufacurer at the time.
If you care to provide links, I may read them.

When Rambus was released and for some time after, it was too hot, too expensive and not fast enough for those 2 penalties.
Only the very fastest RDRam outperformed SDRam. I dont care what evidence you have, that was the performance at that time that I and my customers observed.

The only systems that could use RDRam were those with the P4 CPU which already ran very toasty.
Adding Rambus to the mix made for one hot expensive machine.

 

[Ice Nine]

That's how most people handle the Rambus issue. They don't care what evidence you have They're always going to be the bad guy.

I don't know what you thought you saw back then, but RDRAM never got hot in any system I was running. Today's DDR2 gets far hotter than RDRAM ever did.

The P4 on the other hand........

 

[Nenu]

That's how most people handle the Rambus issue. They don't care what evidence you have They're always going to be the bad guy.

I don't know what you thought you saw back then, but RDRAM never got hot in any system I was running. Today's DDR2 gets far hotter than RDRAM ever did.

The P4 on the other hand........

lol, its you who is typical, I didnt paint them to be the bad guy, I simply related my experience.
I agree with you that Rambus got stiffed and a very bad picture of them was painted that was not correct.
It doesnt help that the comsumer didnt get a good deal though.

 

[ICOM]

That's how most people handle the Rambus issue. They don't care what evidence you have They're always going to be the bad guy.

I don't know what you thought you saw back then, but RDRAM never got hot in any system I was running. Today's DDR2 gets far hotter than RDRAM ever did.

The P4 on the other hand........

I know its just wiki but it's a fair source of information for those who don't know what RAMBUS is about:

http://en.wikipedia.org/wiki/Rambus

From the site:

As a company with no chip production facilities of its own, Rambus conducts business by filing patents and then licensing technologies

The majority of Rambus's revenue is believed to come from lawsuits, rather than developing and selling its own technology

This is a company of lawyers. They use unscrupulous and evidently unlawful practices to patent technology that they didn't develop. Their business is based primarily on lawsuits because that's what they're best equipped to handle. It's just a company of scumbag lawyers masquerading as a technology firm.

 

[Nenu]

Dude, you need to read more on the History of Rambus.
Peoples ramblings on Wikipedia arent always correct

 

[Ice Nine]

It's not at all. Most of their staff is engineering.

If Rambus has yet to win a case, how can the majority of their revenue come from lawsuits? That makes no sense.

They have their technology in a lot of electronics. Not just the PS2/PS3.

 

[Ice Nine]

This is a company of lawyers. They use unscrupulous and evidently unlawful practices to patent technology that they didn't develop. Their business is based primarily on lawsuits because that's what they're best equipped to handle. It's just a company of scumbag lawyers masquerading as a technology firm.

How did a bunch of lawyers convince Sony, IBM and Toshiba to put their interfaces into their products?

See, now THAT'S just Laugh-out-loud funny!

 

[xX_Jack_Carver_Xx]

My issue with Rambus is more fundamental.

What they patented was the application of glue/interface logic that many applications and industrial computers had been doing externally with discrete chips, INSIDE the chip.

Simply taking industry known memory system designs, reducing the schematics to onchip circuits and then PATENTING it is ridiculous.

It is called OBVIOUS under the art. It is not something you can get a legit patent for, unless the Patent Office is pathetic in doing it's job.... oops, it is, dayem.

For example, I designed and built a memory system for a 68040 Mac Quadra accelerator, in 1992 in the days of DRAM on 32bit SIMMs. It used buffer/latches to operate 4 32bit SIMMs simultaneously, so when the 68040 did a burst access... 4 32bit longwords in a row, the memory array accessed all 4 at the same time. The first 32bit longword took however long 70ns DRAMs took to access, but the remaining 3 32bit longwords were also ready in parrallel, so they were delivered with 0 wait-states, one clock cycle each.

This adds up to a huge speed increase, witht he 68040's writeback cache operating in burst mode as well, it made for a very fast system.

DDR simply implements the equivalent ON CHIP. I wasn't alone, there were dozens, hundreds, thousands of unique memory system designs floating around "IN THE ART" in all manner of inductrial and specialty computer systems. Every one was dealing with the same speed constraints of DRAM and it wasn't rocket science as to HOW to get performance improvements.

In the big picture, all Rambus did was take the OBVIOUS manners of increasing performance for accessing DRAM cells and implement some of these tactics on-die.

It simply is not patentable to anyone familiar with memory design at the time. But the USPTO does NOT.... hasn't since Ronnie gutted the agency.... have actual technical experts who can access the true PATENTABILTY of devices and inventions and grant/deny based on science and engineering.... instead the agency was converted into a PIECE WORK sweat shop that gets paid by the volume NOT the quality of the patents they grant.

It is a defect, courtesy of the Republicans, which is a threat to the progress of mankind and the health of our economy and standing in the world. The PTO must be purged and redesigned with quality experts and no incentive for volume over quality workmanship.

/rant

Whether Rambus gets away ultimately with this or not, the controversy hampers the progress of memory design, and mankind needs to put an end to it and move on, we have more important things to do.

 

[ICOM]

Dude, you need to read more on the History of Rambus.
Peoples ramblings on Wikipedia arent always correct

I believe I stated "I know it's just wiki"...

 

[Ice Nine]

OK, name a technology that's better than XDR that's available today.

 

[ICOM]

How did a bunch of lawyers convince Sony, IBM and Toshiba to put their interfaces into their products?

See, now THAT'S just Laugh-out-loud funny!

You've got to be completely fuking kidding...right? "How did a bunch of lawyers..."

Lawyers, especially the RAMBUS scumbag variant, could sell the Pope condoms.

C'mon Man...

 

[Ice Nine]

You've got to be completely fuking kidding...right? "How did a bunch of lawyers..."

Lawyers, especially the RAMBUS scumbag variant, could sell the Pope condoms.

C'mon Man...

OK, you're proposing that lawyers approached Sony, Toshiba and IBM and said "Look, you are going to use our products in a new technology you are going to jointly develop with us."

Did they do it at gunpoint? I mean, you're saying that Rambus is a group of lawyers that doesn't actually develop anything - yet they somehow managed to squeeze some "technology they didn't develop" into Cell?

Are there other people who believe what you do? Because WOW.... that's really farfetched.