Rambus wins judgement against nVidia

[alyarb]

Why all the rambus hate? Without them we wouldn't have ..... XDR for ps3?

so you are grateful to rambus because they made the PS3 possible? there are other technologies that could've filled the gap if rambus never existed, and if they hadn't "invented" XDR others would have. The fact is development of the playstation began during an inconvenient time period and the playstation's hardware is a mere snapshot of that time period.

 

[thilanliyan]

so you are grateful to rambus because they made the PS3 possible? there are other technologies that could've filled the gap if rambus never existed, and if they hadn't "invented" XDR others would have.

Sarcasm?

 

[GodisanAtheist]

For some reason i keep getting a 404 error whenever I try to read the main link. Could someone quickly sum up what Rambus patented and how Nvidia managed to step on their toes? I gather its about Nvidia's memory controller, but are there any specifics?

I mean, what the hell...

 

[waffleironhead]

For some reason i keep getting a 404 error whenever I try to read the main link. Could someone quickly sum up what Rambus patented and how Nvidia managed to step on their toes? I gather its about Nvidia's memory controller, but are there any specifics?

I mean, what the hell...

There is a Google cached version. Just search for the article, its how I found it.

 

[Acanthus]

Although Rambus is far from ethical, they were forced out of the industry by price fixing in the ram cartel.

This, and they basically patented DDR signaling and were ignored by the entire memory industry for royalties... forcing them to litigation.

I'm not a big fan of rambus, but the memory industry is just as bad if not worse.

 

[Cogman]

For whatever reason, people assume a company is up to no good if it has filed many lawsuits against other companies. What people don't seem to realize is that is the ONLY way patent violations are enforced. No government agency is in charge of that.

Intel, AMD, nVidia, etc, all file their fair share of patent suits. The difference is, they actually produce products.

I have no problem with a company defending its right to produce something, I have big problems with a company that does nothing for society except leach off of successful companies.

 

[EarthwormJim]

Intel, AMD, nVidia, etc, all file their fair share of patent suits. The difference is, they actually produce products.

I have no problem with a company defending its right to produce something, I have big problems with a company that does nothing for society except leach off of successful companies.

So how do you feel about ARM? They don't produce anything (I'm assuming you mean actually manufacture/build).

What's wrong with a company that just creates IP? Not everyone wants to get into the manufacturing business, and instead would rather specialize in research.

Is it RAMBUS' fault that a memory cartel ganged up to kick them out of the industry? Keep in mind these tech companies found success in the memory industry using RAMBUS' IP.

 

[Cogman]

So how do you feel about ARM? They don't produce anything.

What's wrong with a company that just creates IP? Not everyone wants to get into the manufacturing business, and instead would rather specialize in research.

Is it RAMBUS' fault that a memory cartel ganged up to kick them out of the industry? Keep in mind these tech companies found success in the memory industry using RAMBUS' IP.

ARM produces processors, not only that, but they are VERY friendly with letting just about anyone use their tech for pretty low prices. Rambus doesn't produce jack, and they let nobody use their tech/charge exorbitant prices for it.

If anything, I kind of wish that AMD and Intel where more like ARM. But that would be bad for AMD and Intel, so I understand why they aren't.

BTW see the reports above. Rambus used some pretty underhanded techniques to both make the companies use their IP, and getting their IPs. The fact that they where kicked out of the industry by equally crooked industries doesn't really make me shed a tear, they haven't demonstrated any desire to get back into the industry, only to sue anyone that crosses their path.

 

[lopri]

What's wrong with a company that just creates IP?

It is much complicated than that, but a patent has never been granted to mere 'ideas'. (that includes Rambus) Such practices can potentially infringe on many rights guaranteed in the constitution, such as free speech, instead of encouraging innovations.

The official stance of U.S. patent office and the Supreme Court is often called "machine-or-transformation test".

http://en.wikipedia.org/wiki/Machine-or-transformation_test

 

[EarthwormJim]

ARM produces processors, not only that, but they are VERY friendly with letting just about anyone use their tech for pretty low prices. Rambus doesn't produce jack, and they let nobody use their tech/charge exorbitant prices for it.

If anything, I kind of wish that AMD and Intel where more like ARM. But that would be bad for AMD and Intel, so I understand why they aren't.

BTW see the reports above. Rambus used some pretty underhanded techniques to both make the companies use their IP, and getting their IPs. The fact that they where kicked out of the industry by equally crooked industries doesn't really make me shed a tear, they haven't demonstrated any desire to get back into the industry, only to sue anyone that crosses their path.

ARM licenses processor designs, they don't build them.

 

[DrMrLordX]

I used to think the problem with RAMBUS was that they never actually made anything, yet expected to be paid for their designs/patents anyway. The fact is that that's not what's really wrong with them.

The problem with RAMBUS, Inc. is that when they actually tried to get their memory designs implemented in silicon and used in a PC platform, their tech sort of fell flat compared to designs based on JEDEC standards. Forget for a moment that, at least according to RAMBUS/RAMBUS apologists that other JEDEC members appropriated RAMBUS' technology by using approved JEDEC memory standards; the fact remained that designs (SDR, DDR) drafted and approved by a lumbering standards committee wound up being more useful than RAMBUS' own creations.

Sure, price fixing hurt RAMBUS' ability to saturate the marketplace with their memory according to Intel's plan, but that wasn't the reason why the Pentium 3 was really no better with RDRAM than it was with SDR DRAM. It also wasn't the reason why the Pentium 4, a processor more-or-less built from the ground-up to NEED RDRAM's bandwidth, wound up being faster on dual-channel DDR DRAM platforms than on any RDRAM platform Intel bothered to release. Hell, dual-channel DDR DRAM platforms PRODUCD BY NOW-DEFUNCT SiS defeated RDRAM when hosting Pentium 4s. What does that tell you about RAMBUS, Inc?

Aside from a few industry wins in which honest-to-gosh RAMBUS, Inc. memory found its way into consoles (and why RAMBUS designs were chosen in those instances is anyone's guess), RAMBUS memory designs have been duds from a technological point-of-view.

RAMBUS, Inc. is good at drafting patents and then slipping them in where they don't belong or fooling judges into believing that someone stole their designs. RAMBUS, Inc. is full of jaded cynics who think it's better to just sue the hell out of people than to try and license tech as a primary source of income.

 

[SlowSpyder]

I used to think the problem with RAMBUS was that they never actually made anything, yet expected to be paid for their designs/patents anyway. The fact is that that's not what's really wrong with them.

The problem with RAMBUS, Inc. is that when they actually tried to get their memory designs implemented in silicon and used in a PC platform, their tech sort of fell flat compared to designs based on JEDEC standards. Forget for a moment that, at least according to RAMBUS/RAMBUS apologists that other JEDEC members appropriated RAMBUS' technology by using approved JEDEC memory standards; the fact remained that designs (SDR, DDR) drafted and approved by a lumbering standards committee wound up being more useful than RAMBUS' own creations.

Sure, price fixing hurt RAMBUS' ability to saturate the marketplace with their memory according to Intel's plan, but that wasn't the reason why the Pentium 3 was really no better with RDRAM than it was with SDR DRAM. It also wasn't the reason why the Pentium 4, a processor more-or-less built from the ground-up to NEED RDRAM's bandwidth, wound up being faster on dual-channel DDR DRAM platforms than on any RDRAM platform Intel bothered to release. Hell, dual-channel DDR DRAM platforms PRODUCD BY NOW-DEFUNCT SiS defeated RDRAM when hosting Pentium 4s. What does that tell you about RAMBUS, Inc?

Aside from a few industry wins in which honest-to-gosh RAMBUS, Inc. memory found its way into consoles (and why RAMBUS designs were chosen in those instances is anyone's guess), RAMBUS memory designs have been duds from a technological point-of-view.

RAMBUS, Inc. is good at drafting patents and then slipping them in where they don't belong or fooling judges into believing that someone stole their designs. RAMBUS, Inc. is full of jaded cynics who think it's better to just sue the hell out of people than to try and license tech as a primary source of income.

Does anyone remember what made Rambus memory so bad compared to other technologies? I seem to remember that it ran at a higher speed, but only 8 bit wide compared to DDR that was 64 bit wide per channel... maybe? That sticks in my head, not sure if it's correct though... maybe something else? Latency maybe? The chipset was just borked? What exactly was the problem with Rambus memory? On paper PC800 Rambus looks better than DDR400, but that was almost never actually the case from what I recall.

 

[EarthwormJim]

Does anyone remember what made Rambus memory so bad compared to other technologies? I seem to remember that it ran at a higher speed, but only 8 bit wide compared to DDR that was 64 bit wide per channel... maybe? That sticks in my head, not sure if it's correct though... maybe something else? Latency maybe? The chipset was just borked? What exactly was the problem with Rambus memory? On paper PC800 Rambus looks better than DDR400, but that was almost never actually the case from what I recall.

PC800 RDRam needs to be in dual channel mode to compete with DDR400. RDRam had significantly higher latencies than comparable DDR memory, 50% more I believe. It usually was faster than competing technologies, but barely (or in the P3 days actually slower).

It was also more expensive, partially due to a larger die size and a memory controller on each chip (not to mention the CRIMMS needed in unpopulated slots). Paired modules of memory aren't a big deal today but a few years ago it was. Also I believe wiring for RDram on motherboards was more complicated, and physical distances mattered more than with SD or DDR memory. Adding multiple chips to a channel really negated performance since latency was directly related to physical distances.

RDram seemed sort of like a brute-force method to increasing memory bandwidth compared to DDR memory. Lots of catches and complications involved in implementing it, let alone licensing it.

RD memory was never accepted on AMD platforms, while Intel has always dominated the market, not having the support of the other major play didn't help things. Rambus also shot themselves in the foot, not every memory manufacturer was licensed to make RD memory.

Do I think it's right how Rambus was booted out of the industry or had its IP used without authorization? Not really, but Rambus isn't the most honorable of companies and really tried hard to make things difficult for themselves. I am glad we're using DDR derived memory and not RDRAM today.

 

[evolucion8]

Rambus used a 16-bit wide bus with speeds up to 800MHz, which is still slower than a 64-bit 400MHz DDR, the Rambus would have to own a 32-bit bus to be as fast as conventional ddr.

 

[EarthwormJim]

Rambus used a 16-bit wide bus with speeds up to 800MHz, which is still slower than a 64-bit 400MHz DDR, the Rambus would have to own a 32-bit bus to be as fast as conventional ddr.

That's what I said, PC800 rdram would need to be in dual channel mode for it to compete with ddr memory. Single channel rdram was competing with sdram.

 

[Dribble]

A more in-depth article:
http://www.businessweek.com/news/20...ttlement-with-rambus-unlikely-after-loss.html

Quote from article - Shannon is the nvidia spokesman:

In a separate proceeding, the U.S. Patent and Trademark Office is taking a second look at the Rambus patents. The three found to be in violation by Essex were rejected by the agency, Shannon said. Rambus is appealing that decision in a process that will take more than a year, and in the meantime the patents remain valid and enforceable.

“We’re not going to pay on patents that are not valid,” he said.

Shannon said that, should Nvidia lose both before the ITC and the patent office, Rambus would have to accept limits on patent royalties because of an agreement reached last year with the European Commission.

 

[SirPauly]

delete, sorry, wrong thread.

 

[SlowSpyder]

I wonder what Nvidia uses that Rambus sued over that apparantly AMD/Intel/Matrox/etc. does not use. Or has AMD/Intel already been sued too?

 

[Juncar]

I heard that AMD/Intel/Matrox already pays royalties for those licenses.

 

[Seero]

It is interesting that Rambus is entitled to "collect 100 dollars" upon each players in the field. I wonder what the patent is, why everyone is using it, and why ain't there work around. More interestingly, Rambus didn't stop people from using it back then, and now suing varies companies for royalties. Did they actually put a patent on something trivial?

Maybe those who got sued knowingly use Rambus' technology without paying, and stall them endlessly for years, hoping that hoping that those patent will become invalid before a case can be put together.

 

[DrMrLordX]

More interestingly, Rambus didn't stop people from using it back then, and now suing varies companies for royalties. Did they actually put a patent on something trivial?

RAMBUS, Inc. never stops people from violating their patents until technology based on their (RAMBUS') patents is present in numerous commercially-available products. It's how they do business.

 

[lopri]

Does anyone remember what made Rambus memory so bad compared to other technologies?

It ran very hot. All Rambus sticks had heatspreaders, iirc, which wasn't a common practice back then. (Strange Rambus didn't patent heat spreaders on memory sticks.. I guess they missed it? :biggrin: )

 

[Obsoleet]

RAMBUS, Inc. never stops people from violating their patents until technology based on their (RAMBUS') patents is present in numerous commercially-available products. It's how they do business.

That's just bad business then on the part of those who feel free to violate the patents, should see that coming.

Rambus is getting justice in this case (or any case where it's ruled their patents are being violated), its the name of the game. No use to cry about it, pay up or get sued. The ruling will/did decide what's right and wrong here.

 

[DrMrLordX]

That's just bad business then on the part of those who feel free to violate the patents, should see that coming.

It is my understanding that there is some question as to whether or not RAMBUS' patents are valid (at least the ones that nVidia has allegedly violated anyway).

In the broader computer memory industry, it seems that the existence of standards organizations like JEDEC have lead to a good deal of backroom deals and buddy-buddy arrangements in which patents from numerous participants are routinely violated with the understanding that everyone will be violating the IP rights of everyone else and that nothing much will be done about it as long as everyone more-or-less behaves. It's a messy situation, I suppose, but apparently that's what it takes for JEDEC drafts to be released and to be useful for consumers and the industry as a whole.

RAMBUS either didn't like it, pulled out after making the mistake of letting Intel leak enough information on their upcoming tech for JEDEC members to start copying bits and pieces of it (and including said tech in JEDEC standards), and sued the bastards; or they knew about JEDEC culture and participated for the express purpose of seeding JEDEC standards with tech they had already patented/were in the process of patenting/intended to patent later so they could sue up a storm and make money later on.

Yes, this does make most, if not all, JEDEC participants colluding IP thieves of a sort, though it certainly doesn't excuse RAMBUS from poisoning JEDEC standards the way they did. When you get right down to it, the situation has gotten so bad that you can't really develop memory or memory controllers that are in any way, shape, or form relevant to commercial applications without violating some patent that RAMBUS has filed. If you want to produce memory for a PC or a video card, or produce a memory controller to handle said memory, RAMBUS is probably already there with a patent, and they might not even offer to accept a license fee since they'd rather see if your product/business will be successful before they decide to enforce their IP. They can make more money suing for damages later than they can trying to collect a licensing fee, especially since a license fee might cause products based on said tech to lose their value edge in the open market.

Rambus is getting justice in this case (or any case where it's ruled their patents are being violated), its the name of the game. No use to cry about it, pay up or get sued. The ruling will/did decide what's right and wrong here.

That would somehow imply that courts and judges are ever really aware of what's going on in tech cases. The consumer rarely wins when this crap goes to court.

Even if you view RAMBUS as being "right" in the eyes of the law, the fact remains that they are obnoxious and that they constantly waste time and money for us all by suing anyone left in the PC industry that has produced a memory controller of any kind. It's an abuse of the court system.

If RAMBUS, Inc. wants to make money, let them sell us XDR and ask us if we want that memory in our system (and let them ask Intel and/or AMD to support it). In most cases, if you ask people whether or not they wanted bonified RAMBUS-based memory platforms in their machines, they would say "no", and for good reason. RAMBUS either produces lousy designs or they roll out excellent designs that are 4-5 years behind the rest of the industry. Do you want XDR in your PC when you can get triple-channel DDR3? I would hope not.

RAMBUS could disappear tomorrow and JEDEC would lumber on without them, unhindered. They don't need them or their technology to continue producing better and better PC memory standards, and one could argue that JEDEC members would have gotten by just fine without RAMBUS' tech when drafting the DDR standard, signaling technology be damned.