NVIDIA Violated Rambus Patents on GPU Memory Controllers

Terry Olaes

Terry Olaes

I Used to be the [H] News Guy
Joined
Nov 27, 2006
Messages
4,646
Speaking of the ITC, a judge there ruled that NVIDIA has violated three Rambus patents, though two other patents in the complaint were cleared. The patents involve memory controllers related to the GPU. This complaint was originally filed back in 2008 and there could be other companies that are in violation now that the precedent is set.

The case also names more than a dozen companies whose products use Nvidia controllers, including Hewlett-Packard and motherboard makers Asustek Computer Inc. and Biostar Microtech International Corp. Most of the Nvidia chips are made in Taiwan, and then sold to other computer-parts makers.
Click to expand...
 
Ah, I remember this from 2006 as well, surprised it took this long to settle. Le Sigh....sue happy USA :D
 
I hope I'm not the only one sick of this bullshit with Rambus patenting technologies it did not develop and shaking down companies for billions.

Gee, thanks Rambus. My next nVidia card just got more expensive to pay your lawyers. While you're stealing money from my pocket, you might actually develop some technology that I actually give a shit about.
 
To be fair Rambus did develop technology that was worlds faster then DDR. It was called RDRAM and when RDRAM was running PC800 (DDR400) DDR was running or I should say [email protected] the math my friend. =)
 
maleficarus said:
To be fair Rambus did develop technology that was worlds faster then DDR. It was called RDRAM and when RDRAM was running PC800 (DDR400) DDR was running or I should say [email protected] the math my friend. =)
Click to expand...

They haven't done anything but patent claims since then though. RAMBUS supposedly worked with other companies developing standards, while secretly patenting the very things they were proposing to the standards groups - for obvious reasons. I think this quote from the article will allow you to infer what my opinion of RAMBUS is:
In the Micron case, a federal judge had said Rambus deliberately destroyed documents to hinder the defense by chipmakers in patent litigation
Click to expand...
 
maleficarus said:
To be fair Rambus did develop technology that was worlds faster then DDR. It was called RDRAM and when RDRAM was running PC800 (DDR400) DDR was running or I should say [email protected] the math my friend. =)
Click to expand...

I have to agree here...I have a machine, nearly seven years old, still running strong with 2GB of RAMBUS memory. At the time I got it, that machine was insanely fast, and those RAMBUS modules were a big part of that. Moreover, that was a huge amount of memory...I had bragging rights for years with that one. ^_^
 
RAMBUS again? I wonder if they are in the right this time? My guess, based on their history, would be no. It will still be a couple years b4 this one is actually settled. So there is really not much to see at this point in the game.
 
I think Rambus is in violation of my patent on being an asshole! I should sue them!

maleficarus said:
To be fair Rambus did develop technology that was worlds faster then DDR. It was called RDRAM and when RDRAM was running PC800 (DDR400) DDR was running or I should say [email protected] the math my friend. =)
Click to expand...

True, but the latency was very high. Memory latency has always been a bigger issue than memory bandwidth. The moves from SDR to DDR, DDR to DDR2 etc. have not improved performance that much because while bandwidth doubled, latency stayed the same or actually went up.
 
didnt rambus make crap ram that no one really used? They don't deserve shit.
 
Konig-Wolf said:
I have to agree here...I have a machine, nearly seven years old, still running strong with 2GB of RAMBUS memory. At the time I got it, that machine was insanely fast, and those RAMBUS modules were a big part of that. Moreover, that was a huge amount of memory...I had bragging rights for years with that one. ^_^
Click to expand...

Really?

I still have a p4 with 1gb (256mbx4) of rambus ram. When I first got it, it had "only" 512mb, but I really never felt it was fast.
 
Nanan said:
Ah Patents, stagnating technology since 1836.
Click to expand...

if rambus actually made this product and used it, then what is the problem, it is okay to just take someone else's idea and use it, and make money from it.

i thought in order to get a patent you had to have a working version?
 
RDRAM was retarded, because you had to have any memory slots without memory in them filled by some other chip. Not to mention it was really expensive.
 
GORANKAR said:
RAMBUS again? I wonder if they are in the right this time? My guess, based on their history, would be no. It will still be a couple years b4 this one is actually settled. So there is really not much to see at this point in the game.
Click to expand...

did you bother to read the article, or even the first post....


Speaking of the ITC, []a judge there ruled that NVIDIA has violated three Rambus patents,[/b]
Click to expand...
 
I used to have a P4 system with 1066 RDRAM... I though it was pretty fast. Damn ram was so friggen expensive at the time though :(
 
Rambus did quite a bit of illegal patent scheming and did everything they could to get out of antitrust problems. They were on the memory standards board long enough to suggest technology to be used for DDR based technology and once the standards were set they left the board and forgot to mention that all the stuff they offered was going through a patent process and some of the stuff they patented they did not even develop. Which is why people hate their guts.

If they had been open about it the board would have gone a different route and maybe a faster route.
 
JimmiG said:
True, but the latency was very high. Memory latency has always been a bigger issue than memory bandwidth. The moves from SDR to DDR, DDR to DDR2 etc. have not improved performance that much because while bandwidth doubled, latency stayed the same or actually went up.
Click to expand...

It was also four times the price with little or no real world performance gains. Intel using Rambus ram was the reason I got my first AMD processor. :D
 
This is not good .
Rambus has been on a roll a lately with samsung settleing with them for almost a billion and now nvidia and its affilates losing to rambus . I fear we are going to see a return of this monster to the pc world and it will not be a good thing . Memory prices will probably skyrocket again and will be locked into some kind of memory scheme that promises faster performance while delevering no or at best a minimal boosts in performance .

What should happen is the industry just boycotts rambus all together and refuses to do business with them .
If a product has rambus in it . Then it should treated like the plaque and avoided at all costs.
Rambus should be just like abestos in the mind of the consumers. the mere mention of it should send people fleeing in terror .

All this is being said in the intrest of affordable computers with great performance.

If rambus gets their way computers will no longer be afforadable.
Their performance will look great on paper, but in reality it will be no better than what you could get on todays top end technology . After having paid a premium for SuXuRDDR or whatever rambus wants to call their next wonder technology . We will all be the poorer not just literally but our computers will suffer Too.
 
rambus hasn't sold a product in years. They're strictly a paper r&d/patent trolling company. The problem is that through their abuse of the DDR1 standards committee they managed to get patents on enough key technologies that making high speed ram without stepping on one is almost impossible.
 
you mean the rambus used in the p4 that one were outperformed by AMD64 and AMD64 x2. AMD over took the market and stock went as high as $80. That rambus?
 
piscian18 said:
didnt rambus make crap ram that no one really used? They don't deserve shit.
Click to expand...

No, it was good ram, but they shot themselves in the foot by not licensing anyone (barely) to make it............the price was too high.........and DDR passed it up by being decent and affordable.

Rdram disappeared faster than a quart of vodka at an AAA meeting.:D

Now all they do is bum around someones basement telling stories of how "we could have been the best", and "remember when people bought our stuff?".
 
I don't think anyone is questioning the validity of the RAMBUS patents. It's the way they gained the knowledge for the patents progress that is disturbing.

They joined the JDEC standards group and stole information and processes from other vendors who where divulging in good faith. They then left the group and after obtaining patents on other vendors processes behind their back, sued everyone into oblivion.

Call it what you will. It may be legal, but RAMBUS stole every thing they patented. They never paid for R&D like everyone else. What sucks is knowing every ram purchase I make they get a kickback from now. For what? Questionable morality?
 
Mega2 said:
you mean the rambus used in the p4 that one were outperformed by AMD64 and AMD64 x2. AMD over took the market and stock went as high as $80. That rambus?
Click to expand...

Nope, the rambus that was outperformed by Athlon/duron with PC133, and DDR333
 
BallerX said:
I don't think anyone is questioning the validity of the RAMBUS patents. It's the way they gained the knowledge for the patents progress that is disturbing.

They joined the JDEC standards group and stole information and processes from other vendors who where divulging in good faith. They then left the group and after obtaining patents on other vendors processes behind their back, sued everyone into oblivion.

Call it what you will. It may be legal, but RAMBUS stole every thing they patented. They never paid for R&D like everyone else. What sucks is knowing every ram purchase I make they get a kickback from now. For what? Questionable morality?
Click to expand...

There is no such thing as morality in business. Blame the system, not the company.
 
Rambus created their glorious memory controller, fired all their engineers, and became a company of lawyers. Fook them.

I hate patent trolls. The patent office is way...way behind the times for technology patents.
 
MrGuvernment said:
did you bother to read the article, or even the first post....
Click to expand...

Uh, yeah, and they are not at the end of the appeals rope yet, once exhausted, we move to the fight for actual amount of damages and those appeals as well. Still a year or more out on this one imo.
Or did you take exception to my doubts regarding them being in the right? A judge spewing out a judgment makes something a legal requirement, til another judge comes along and changes it, but it does not necessarily make it the right thing. RAMBUS's past actions make wondering about the veracity of their claims an easy thing to do. They have won in court b4 and had things reversed b4. No need to take it personal.
 
What kind of bait do you need to get all of RAMBUS, RIAA and MPAA into a sturdy little air tight building? I just want them to talk to each other for a couple of weeks... with the building sealed. No food, water, light, electricity, cell service, Internet, plumbing, etc...
 
maleficarus said:
To be fair Rambus did develop technology that was worlds faster then DDR. It was called RDRAM and when RDRAM was running PC800 (DDR400) DDR was running or I should say
[email protected] the math my friend. =)
Click to expand...

Ummm... you forget, it was 800mhz on a 16bit bus vs 266 on a 64bit bus.

All you get with a narrower faster pipe are higher latencies.

Probably why P4/DDR wiped the floor with P4/RDRAM

Not to mention rambus charging a fortune for the privilege of using their technology
 
I'd say that patent law needs reform, but any time the government reforms anything it gets worse and the lawyers still get rich either way.

I like Poseur's idea, except that I think the building should also have some VX floating around in the ventilation system. :D
 
maleficarus said:
To be fair Rambus did develop technology that was worlds faster then DDR. It was called RDRAM and when RDRAM was running PC800 (DDR400) DDR was running or I should say [email protected] the math my friend. =)
Click to expand...

Ok, I'll do the math


PC133 = 133MHz * 64-bits = 1.06MB/s

PC1066 = 533MHz * 16-bits * 2 = 2.1MB/s

PC2100 = 266MHz * 64-bits * 2 = 2.1MB/s

RDRAM also had the detriment of having to send data serially. So data stored in chips further from the memory controller had to travel through every other chip on that channel. To compensate, it requires 32ns for every 2-byte memory access. CL3 DDR-266 could have 8-bytes of data out as fast as 11-16ns.
 
AVT said:
There is no such thing as morality in business. Blame the system, not the company.
Click to expand...

A company doesn't spring into existence spontaneously. There are people running it, deciding what it should and shouldn't do. The people deciding what actions they're going to do under the company's name should have morals. If a thug shoots somebody and says "Blame the game, not the player." would you nod your head in agreement and say what he did was ok?
 
JimmiG said:
I think Rambus is in violation of my patent on being an asshole! I should sue them!



True, but the latency was very high. Memory latency has always been a bigger issue than memory bandwidth. The moves from SDR to DDR, DDR to DDR2 etc. have not improved performance that much because while bandwidth doubled, latency stayed the same or actually went up.
Click to expand...

This would be true IF the RAM was running on the same memory controller AND at the same speed (Mhz). However, neither of these is true and therefore your point is mostly invalid.

The biggest issue was with the memory controllers NOT being on-die on the CPU. Once AMD added an on-die memory controller, DDR1 running at a mere 250Mhz was able to be almost as fast as Intel systems running DDR2 at 400Mhz.
 
You must log in or register to reply here.
Back
Top