backtop

Print 33 comment(s) - last by fuser197.. on Jul 17 at 5:25 PM
Rejection of claims is just part of the process says Rambus

Rambus and NVIDIA have been fighting a legal battle in court over alleged patent infringement by NVIDIA. Rambus filed suit against NVIDIA in June alleging that technology having to do with memory circuits in NVIDIA GPUs infringed on 17 different Rambus held patents.

NVIDIA general counsel David Shannon said in a statement, "[The patent office] has now initially rejected all of the patent claims asserted by Rambus against NVIDIA in the ITC."

Rambus says that the rejection of the claims is all part of the process. Rambus said in a statement, "The recent action by the [patent office] to re-examine Rambus patents is part of the process and Rambus will have an opportunity to respond."

Reuters reports that Rambus has spent $300 million on legal disputes with different chipmakers since 2000. Rambus wants NVIDIA to pay royalties on the memory circuits that are used in its GPUs. The dispute is far from over at this point for either side.

Analyst David Wu from Global Crown Capital told Reuters, "NVIDIA appears to have won one round in a multi-round contest, so we have a lot more decisions to make before they can declare final victory."

Part of NVIDIA's defense against the Rambus allegations was to tell the patent office that the patents were based on old ideas. Rambus has already asked the ITC to drop some of its claims against NVIDIA in the suit.

Comments     Threshold


This article is over a month old, voting and posting comments is disabled
All Rambus ever has done...
By MrBlastman on 7/15/2009 10:56:44 AM , Rating: 5
Is piss, moan and sue. I think I've been reading about them doing this for years. They apparently are still pissed off about Intel dropping them years ago and continue to stomp and shake their rattle.




RE: All Rambus ever has done...
By FITCamaro on 7/15/2009 11:01:00 AM , Rating: 2
About their only win was with the PS3. But yes they've largely become a company that sues to stay in business. Other than the PS3, does anything use their technology?


RE: All Rambus ever has done...
By MrBlastman on 7/15/2009 11:20:41 AM , Rating: 2
The N64 did actually. I remember Nintendo bragging about it at the time. Anything else currently? No so sure.


RE: All Rambus ever has done...
By FaaR on 7/15/2009 12:31:48 PM , Rating: 2
XDR memory is apparantly used in heavy-duty routers and such, that need very fast memory to cope with the packet load of today's 10gbit networks.

So they do have design wins other than the PS3. If they hadn't, it's unlikely they'd still be in business today.


RE: All Rambus ever has done...
By hduser on 7/15/2009 12:34:45 PM , Rating: 2
Don't forget about PS2 which had RDRAM in it.


RE: All Rambus ever has done...
By rs1 on 7/15/2009 3:33:17 PM , Rating: 2
I have an 8 year old 1.3 GHz P4 box that actually uses RDRAM. It still runs today, I use it to host a couple low-volume servers, and it works fine for that task. Of course, it's horrendously slow compared to any modern system, but I remember that at the time I got it, its memory performance was near unbeatable.


RE: All Rambus ever has done...
By Samus on 7/15/2009 6:57:12 PM , Rating: 2
RDRAM had minor improvements over SD RAM with the P4. In the P3's case, it was actually slower.

DDR came along and completely killed the technology for PC implementation.


RE: All Rambus ever has done...
By TA152H on 7/16/2009 6:21:27 AM , Rating: 2
Actually, you're wrong, of course. Why even make such a stupid post?

For a long time, most people considered anything but RDRAM for the Pentium 4 a mistake, and this was long after DDR was available for it from companies like VIA, SIS, and even Intel.

Intel killed RDRAM, and would not update the i850 chipset, or even allow it to be used with later south bridges (although Gigabyte did it), with the consequence that RDRAM died that way. SIS made some noise about creating a RDRAM chipset, but never did, so all that was left was DDR and RAMBUS dropped later specifications of RDRAM with no support from Intel.

Inasmuch as RDRAM easily outperformed DDR when they were both well supported, and really even after the DDR chipsets were newer than the i850, it's unclear that DDR killed it. Intel killed it, and with no other real allies, that was enough.

With regards to the Pentium III, you're also wrong, but not by as much. It wasn't universally slower on the Pentium III. The i840 actually beat the i815 and even the overclocked 440BX on many benchmarks, although it lost to the overclocked 440BX on many as well. It was probably the fastest legitimate chipset for the Pentium III at 133 MHz. The 440BX could be said to be slower, faster, or equal depending upon how you valued the different benchmarks. The i820, though, was overall slower, and was much more popular than the i840, even though the cost difference was minor, consider the costs of the memory itself at that time, and the performance difference pretty dramatic. I don't know why anyone would buy an i820 with PC800, when a i840 with PC600 was faster, and had greater options down the road. But they did.


RE: All Rambus ever has done...
By fuser197 on 7/17/2009 5:22:21 PM , Rating: 1
You might be right, but you sound like an asshole.


RE: All Rambus ever has done...
By Calin on 7/16/2009 7:38:15 AM , Rating: 2
Dual Channel SDRAM and single channel DDRAM were somewhat close to the performance of the RDRAM - until dual channel DDRAM appeared.
I still have an computer with RDRAM - but I found that upgrading its 128 MB of RAM was more costly than a new, decent mainboard/processor/RAM combination


RE: All Rambus ever has done...
By XZerg on 7/15/2009 11:11:11 AM , Rating: 4
It wasn't because of Intel entirely. The issue was Samsung and other DRAM manufacturers collaborated on price fixing the RDRAM. The high prices is what caused the take up of RIMM down to negligible. The manufacturers have been found guilty and been fined for it.

Rambus has managed to come out with quite a few good ways to improve frequency and performance while lowering power usage and for cheap.

The issue with the company however is that their legal side has given them such a huge negative publicity that it is harder to see past it and towards the technology they offer.


RE: All Rambus ever has done...
By brybir on 7/15/2009 11:49:27 AM , Rating: 2
This is a critical part of the story although I am not fully versed in it. The DRAM manufacturers were found to be engaging in price fixing in many countries and faced some pretty hefty fines. Would not suprise me if they somehow colluded with each other to either drop the SDRAM prices a lot or artificially raise RDRAM prices to make it exit the market.

I remember back when I had a Gateway system with RD-RAM and at the time it was touted as "so much better" than the standard DRAM offerings at the time. But that RD-RAM was expensive and you had to buy it in pairs (I think) which made upgrading costly as you could not always just "add mor".


RE: All Rambus ever has done...
By AlexWade on 7/15/2009 12:40:56 PM , Rating: 2
Yes, you had to buy it pairs like SIMMS chips. But unlike SIMMS, requiring the pair was for performance and not necessity. RDRAM also require a terminator on the unused slots, much like SCSI cables.


RE: All Rambus ever has done...
By Solandri on 7/15/2009 4:12:02 PM , Rating: 2
quote:
This is a critical part of the story although I am not fully versed in it. The DRAM manufacturers were found to be engaging in price fixing in many countries and faced some pretty hefty fines. Would not suprise me if they somehow colluded with each other to either drop the SDRAM prices a lot or artificially raise RDRAM prices to make it exit the market.


The DRAM manufacturers "colluded" after Rambus tried to torpedo their SDRAM specification committee with a submarine patent. They weren't happy about that, and (IMHO individually) chose to punish Rambus for it. I'm still scratching my head over how the judge could decide that the no-patent rules of the DRAM specification committee weren't enforceable. That kinda defeats the purpose of ever making such a committee. I suspect it had more to do with, as you say, the DRAM manufacturers being engaged in their own hanky panky, rather than Rambus' case actually having merit.

At the time, Intel was doing everything it could keep up the appearance that it had made the right decision in selecting RDRAM. But the leaks from the low-level Intel engineers was that it was horrendously complex to design with, and in some cases impossible. My impression from following it at the time was that it was a niche product which Rambus salesmen managed to pitch to Intel as a mainstream product. Just because it was faster than SDRAM doesn't really count for much. It's fairly common for a less capable but easier to use product to win in the market over better but harder to use/design with products. Windows and the iPod are great examples.


RE: All Rambus ever has done...
By brybir on 7/15/2009 12:15:11 PM , Rating: 2
I think you are also right about them being perhaps a solid engineering company.

I read things like:
http://www.rambus.com/us/news/press_releases/2009/...

And think "perhaps they are innovating". If that is to be believed, that they can produce memory with 40% less energy requirement at twice the bandwidth of GDDR5, I would say that is fairly impressive if it can be had at a competitive price point.

Then again, who knows whether its just FUD or not.


RE: All Rambus ever has done...
By hduser on 7/15/09, Rating: -1
RE: All Rambus ever has done...
By invidious on 7/15/2009 12:59:24 PM , Rating: 3
quote:
If it wasn't for the competition Rambus provided, you'd still be using EDO memory.


Uh no? Your crystal ball is broken, just because they sped up innovation doesn't meant that in their absence innovation would have stopped entirely.


RE: All Rambus ever has done...
By hduser on 7/15/2009 1:32:14 PM , Rating: 2
I'm exaggerating of course. What the memory manufacturer's have done prior to Rambus has been evolutionary FPM->EDO->Burst EDO->SLRAM. Frankly the innovations were few and far between. Intel got fed up with the pace of innovation and chose RDRAM to feed it's upcoming netburst architecture because RDRAM was revolutionary not evolutionary. Since that, you got SDRAM->DDR->DDR2->DDR3. But hey, I'm sure those memory manufacturers would be happy to churn out the same old design year after year.

But at least you did acknowledge that Rambus sped up the innovation curve, directly or indirectly.


RE: All Rambus ever has done...
By XZerg on 7/15/2009 2:25:02 PM , Rating: 3
I doubt if Intel chose RDRAM because it was revolutionary. Making a decision only on such basis would have run any company down very fast. Companies have to show the need for so and so devices to stay competitive and also attract customers to purchase your devices.

Most likely there are two probable reasons I believe why Intel chose RDRAM for their P4 platform:
1. Intel was banking on higher frequency lower IPC to be the way to go, at least that's the architecture P4 was.

2. To support such architecture you need a high bandwidth ram to supply data to make the system perform optimally. RDRAM's high frequency, 800mhz at the time of release, was much much higher than the PC133/PC100 frequency of SDRAM we were selling at the time as the other option. DDR came after and that too didn't jump to 400mhz until later.

The benefits of RDRAM however was "marginal" compared to SDRAM at that time. Why? Because at that time AMD was the King of the town in terms of performance and had 10-20% lead over a P4 CPU's performance. So the overall system performance was about similar but you had to pay an arm and a leg and possibly liver if you went with Intel instead of AMD.

The other factor that lead to RDRAM's demise was the fact that all the DRAM manufacturers had heavily invested in SDRAM and DDR SDRAM to just flip over and focus on RDRAM. So the quick ramp up of DDR SDRAM and the lower prices made sure RDRAM died quick and painful death.

If you recall Intel had to step back and provide a chipset to support SDRAM on P4. That was even more ridiculous given Via was supplying a chip to support DDR SDRAM for P4. AMD had long stepped into DDR SDRAM too.

So it wasn't so much about revolution but what the architecture required to perform optimally.


RE: All Rambus ever has done...
By hduser on 7/15/2009 3:36:49 PM , Rating: 2
I kind of remember that era a bit differently. RDRAM was already picked by Intel. SDRAM was officially rolled out after that a year after that to counter RDRAM.

While I agree with most of your post, there were other reasons for RDRAM's failure in the market place.

1. Real performance gains were slim and cost was higher.
2. Ramping production of RDRAM using the P3 CPU and 820 chipset was a disaster.
3. Programs didn't take advantage of the hyperpipelining and memory throughput P4/850 chipset combined with RDRAM higher latency.
4. Negative press from the above points and Intel "backdoor" deal made Rambus bashing a daily occurance.
5. Possible collusion from memory makers prevented RDRAM into the market place.

In spite of all this, I still think RDRAM was revolutionary. Perhaps it was an elegant solution to a problem that didn't really need addressing. Even current benchmarks for DDR2 show less than 5% real gains across different speeds of DDR2.


RE: All Rambus ever has done...
By jconan on 7/15/2009 11:03:32 PM , Rating: 3
That wasn't the only thing, Rambus demanded royalty payments for RDRAM greater than SDRAM. That was what did it for RDRAM because Rambus was greedy. If it was affordable and Rambus had thought about the future they wouldn't be in the demise.


RE: All Rambus ever has done...
By hduser on 7/15/2009 11:34:48 PM , Rating: 2
At that time when Rambus wanted to enforce royalties in the early 2000s, they wanted MORE royalty for SDRAM/DDR and less for RDRAM to level the playing field and spur RDRAM adoption. I believe they wanted about 1-2% for royalties on RDRAM and about 3-4% on DDR. So I don't quite understand your assertion.

Rambus couldn't make RDRAM affordable because it wasn't in charge of manufacturing. Initially RDRAM was more difficult to manufacture and the memory maker had no interest in driving the price down. And it's entirely possible that memory makers had colluded on pricing to make SDRAM/DDR cheaper (already proven in court) and kept RDRAM high(remains to be seen but not far fetched now given the former admission of collusion).


RE: All Rambus ever has done...
By Solandri on 7/15/2009 4:20:24 PM , Rating: 2
quote:
Whether you believe that Rambus owns DDR memory IP or not, DDR was created as a direct challenger to RDRAM. If it wasn't for the competition Rambus provided, you'd still be using EDO memory.

The idea behind DDR is so simple that I think anyone who gets a 1-minute explanation would agree that it's obvious and not worthy of a patent.

Your computer synchronizes events with a clock. At a tick of the click it flips a timing signal. If it's currently 0, it flips to 1. If it's currently 1, it flips to 0.

Regular DRAM uses the flip from 0 to 1 as a sync signal.

DDR (dual data rate) DRAM uses both the 0->1 and 1->0 flip as a sync signal.

That's all there is to it. It just wasn't implemented before because the extra speed wasn't needed. It wasn't until CPU core frequencies started vastly exceeding the bus clock speed that there developed the need for memory which ran faster than the bus clock speed. Whether or not RDRAM existed, DDR memory would've been "invented" because it was so damn obvious.


RE: All Rambus ever has done...
By hduser on 7/15/2009 5:05:14 PM , Rating: 2
Umm, that's an overly simplistic view of the situation. Dual edge clocking has been around for a long time I agree. Some circuits used rising edge other used falling edge, hey why not use both?

What is in question is the signal clocking across a bus. Rambus perfected a technique of high speed signaling that was not used that way before. Combine that with a packetized interface and dual edge clocking and a narrow bus you get RDRAM. Rambus's contention is that SDRAM is RDRAM without the packetized interface and dual edge clocking on a wide bus. DDR added the dual edge clocking and Rambus's clock skew correction. It's been a while but that's pretty much as AFAIK.


RE: All Rambus ever has done...
By Clauzii on 7/15/2009 12:13:31 PM , Rating: 2
They do however, also make some quite nice RAM technoligies like XDR etc..


You get 3 years.
By Lastfreethinker on 7/15/09, Rating: 0
RE: You get 3 years.
By brybir on 7/15/2009 12:09:56 PM , Rating: 4
The reality is that many (perhaps the majority?) patents are not necessarily intentionally sought after and sometimes the rseults could not be used right away.

Example:

An engineer at Caterpillar comes up with an innovative new hydraulic system for use in tractors completly by accident as they were trying to do something else. However, this new idea does not mesh well with current generation tractors but will work in the next generation that is slated to start being built in 5 years. Should they lose the patent just because they intentionally plan to sit on it for several years while waiting to maybe use it in their next generation tractor?You can analogize this over many sectors of the economy where lead times in new products have differing life spans and production lead times. This makes the idea of a set time period somewhat unworkable.

And what about worst case? Say the neat hydraulic system is never used by Caterpillar. In that case, the company paid for an invention that it could liscense out to John Deere or GE or something, or it could just sit on and let eventually fall back into the public domain. What is wrong with either case? In the first case John Deere or GE would be benefiting from Caterpillars research and in the second case the public basically gets "free" research that advances the state of hydraulics and eventually everyone can use for free.

Does not seem that bad to me. Sometimes we have to remember that while the patent system has some issues it still is a major source and driver of innovation for many companies. Patent trolling is something that seems like a net loss for society but we have to be careful that in our quest to extinguish the trolls that we dont undermine the entire system in the process.


RE: You get 3 years.
By invidious on 7/15/2009 1:01:59 PM , Rating: 2
What if it takes 4 years to develope your product? You are just supposed to dump millions of dollars into the idea only to have someone else steal the patent at the last minute? Smart people came up with the rules, don't make uninformed criticisms.


What was the suit?
By FITCamaro on 7/15/2009 11:00:07 AM , Rating: 2
What was the basis of the suit for which Rambus thought Nvidia owed them money? How were the memory circuits supposedly violating their patents?




RE: What was the suit?
By XZerg on 7/15/2009 11:13:46 AM , Rating: 2
the memory controller in the nvidia chips that RAMBUS was after. I have heard the graphics for sure but could include the motherboard chipset as well.


@@$# Rambus
By fuser197 on 7/17/2009 5:25:53 PM , Rating: 1
Fuck Rambus




i just want to see the G300 series!
By superPC on 7/15/09, Rating: -1
By Goty on 7/15/2009 12:01:39 PM , Rating: 5
Can you please go wet yourself in a relevant article? Thanks.


"I f***ing cannot play Halo 2 multiplayer. I cannot do it." -- Bungie Technical Lead Chris Butcher



Latest Headlines
2/10/2012 Daily Hardware Reviews
February 10, 2012, 5:50 PM
2/9/2012 Daily Hardware Reviews
February 9, 2012, 11:54 AM
2/8/2012 Daily Hardware Reviews
February 8, 2012, 1:11 PM
Ex-Acer CEO Slapped With Lawsuit for Joining Lenovo
February 7, 2012, 5:38 PM
2/7/2012 Daily Hardware Reviews
February 7, 2012, 12:23 PM
2/6/2012 Daily Hardware Reviews -- Intel 520 240GB SSD Edition
February 6, 2012, 12:04 PM
botimage
Copyright 2012 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki