Slashdot Mirror
Intel To Drop RAMBUS In Favor of DDR RAM
El Pollo Loco writes: "Anandtech has this news article. Basically, Intel has decided to drop expensive rambus memory and instead to go with cheaper DDR memory."
← Back to Stories (view on slashdot.org)
Watch the RAMBUS litigation machine swing into overdrive, now that they can't even pretend to make things...
ABSURDITY, n.: A statement or belief manifestly inconsistent with one's own opinion.
I wonder how much of this switch was due to the recent scandals - are Intel worried about the pockets of their customers? have they decided to stick with less, ahem, notorious technologies? are they truly concerned with the performance? I notice they haven't mentioned comparitive benchmarks in the article though... not a good sign.
If their competitors follow suit, we'll see what happens.
jer
We may be human, but we're still animals
- Steve Vai
This is a bit of a shock, really. RDRAM *is* a half decent solution for some high end boxes. That article from the windbags over at Tom's Hardware showed (kinda) sucessfully that there is a great deal of potential in RDRAM. I am an AMD / DDR user, but I have to admit that the P4 with a 133x4 bus is damn nice.
tinfoilmedia
Obviously, this is bad news for Intel but great great news for the industry in general.
Intel tried to push RAMBUS heavily. In fact, it tried to ram it down people's throats (no pun intended). For various reasons, not least of all cost, not too many people were happy with this state of affairs.
I can recall when Intel were pushing RAMBUS as the best thing since sliced bread and were denouncing DDR RAM as a pile of pants but now the company's been forced to perform a complete volte face.
Why does this matter? It matters because Intel, despite it's near total dominance of the desktop market, has been shifted from leading the herd to being forced to run with it. It just goes to prove, you can lead a horse to water but you can't make him drink it.
Of course, all this doesn't change the fact that Intel is a major player, and will get its own way in lots of other areas but it's nice to see that it can't win 'em all.
Additionally, one fewer memory standard should help drive down the long-term cost of DDR RAM further (right now it's a twice what it was in November 2001 but still at least a third cheaper than March 2001).
Good news all round. Score one for the other guys.
"Accept that some days you are the pigeon, and some days you are the statue." - David Brent, Wernham Hogg
Since this move will only be one toward the end of the year, that will mean that the P4 will still be heavily based on RAMBUS memory.
Even though there are several other P4 motherboards out there supporting DDR RAM for the P4, the point is that Intel will have to support DDR and RAMBUS for their P4 fully.
I can see complications brewing... this should test Intel's mettle.
Otherwise, they may decide to leave the current P4 chips with RAMBUS for now, in which case the customers who bought into it are screwed.
I'm an AMD fan (see me twirl!), but RAMBUS have screwed Intel for the last time with overpriced and underperforming memory.
Sticking feathers up your butt does not make you a chicken - Tyler Durden
It's only 8:37 and Rambus stock is already down 6%. I imagine it will drop some more. They made an announcement yesterday that they're going to start making cheaper memory, and it boosted their stock a lot. I think they'll completely lose that gain. Personally, I'd like to see them go out of business just because of their crappy attitude in the industry towards consumers and competitors.
He been such a *whore* because he had the guts to admit that he was wrong about Rambus in the first place. He's not kissing their butt, he's just recognizing the facts : DDR just isn't good enough for Pentium 4. At these high clock rates, only Rambus delivers enough bandwidth to make the P4 happy.
And saying that Tom is accepting money is just ridiculous : he's probably the best hardwre journalist on the Net, and has no need to accept bribes from companies.
First, this is an EBM story. Why link to a 2nd hand report that has a link to the primary source right on the page?
Second, it's spurious: "An Intel workstation roadmap secured by EBN" strongly implies that this is not an official Intel announcement.
Third, while it's not such a huge deal for Intel, it's a huge furry deal for Rambus Inc., the well known firm of lawyers. Rambus Inc. is a public company, and as such has an obligation to announce significant events effecting future earnings. "A spokeswoman for Rambus Inc. said she couldn't comment on new Intel workstation chipsets supporting DDR, and referred all questions to Intel" simply doesn't cut it. Rambus Inc. might be greedy lying parasites, but they're surely not stupid enough to sit on information that they must - must - have known about prior to this (alleged) policy change.
Given that Rambus Inc. share price rose 22% yesterday based on the news that Intel had adopted the 533Mhz FSB to support RDRAM, the SEC will no doubt be having a good, long look at their disclosures and these "yes we will/no we won't" announcements, and asking who exactly is releasing them, and who is benefitting from the share fluctuations.
Let's hold fire on this until it's been confirmed by both Intel and Rambus Inc. Please. Pretty please.
If you were blocking sigs, you wouldn't have to read this.
They should boycott DDR ram memory in protest
against the terrible human rights situation in
communist East Germany.
Buying DDR RAM is supporting the communists!
I didn't see any references here, so here it is:
http://www.ebnews.com/story/OEG20020226S0040
- Mike
Ah, from the mouths of babes (cowards)...
if you check pricewatch RDRAM and DDR have almost identical costs. if someone is charging more for RDRAM then they are just lagging behind the market...
If you just read the headline above and the linked *clipping* of a story, you'd think Intel is dropping RAMBUS completely. That is not the case. Here's the original story that the clipping came from. Intel is dropping *exclusive* support for RDRAM. Here's a relevant quote:
/. earlier this week.
Also, although not new products, the next iterations of its 850 and 860 chipsets, supporting a 533MHz front-side, will support RDRAM when they arrive, probably in the second half of this year.
This ties in perfectly with Tom's Hardware review of a new, pre-release 533MHz RDRAM chipset and the Anandtech review of the new Intel DDR-based chipset linked to by
I think the reason why Intel may be supporting more DDR-SDRAM installations for desktop platforms is the fact that the latest DDR-SDRAM variants already equal the speed of RDRAM for large data blocks but substantially cost way less than RDRAM on a per 128 MB basis. I believe also that motherboard manufacturing costs are lower for DDR-SDRAM than RDRAM, too.
Intels decision to drop rambus in favour of DDR, DDRII, and dual channel versions of it isn't because of the desktop market at all. Which is what is misleading here.
Intel has always known that RAMBUS is not the best thing for servers, since servers need large amounts of memory. The serial configuration RAMBUS must be put it, along with its higher latency, leads to poor performance when in large amounts. This isn't acceptable, and doesn't happen with DDR, its the primary reason they are switching.
That and RAMBUS runs hotter then the P4 CPUs
Maybe he was right about rambus first time around, but the situation has changed since?
How about it?
You're mostly right. Most /.'ers went on an anti-Rambus rant and assumed Intel was dropping RDRAM across the entire line.
The EBN article clearly says that the new chipsets are for workstations (graphics, software engineerings, MCAD, etc.). Folks, these are machines that have at least 512MB of RAM. They are not mainstream desktops...
For now, Intel is still pushing RDRAM for mainstream desktops...
--Al
In fact, you're right. Intel showed no advantage using Rambus at first, but when the clockrates went higher and needed more bandwidth, then Intel showed why it had chosen Rambus as the leading memory technology to be paired with the P4. Right now, pairing P4 with DDR is like pairing an AthlonXP with cheap PC133 modules : it's just plain dump. You won't get the amazing performance your processor can offer. Stick to DDR for AMD since there is no Rambus support planned (I think), and Rambus for P4. They might cost a hell lot more, but if you have the money to buy a decent system with a P4, then chances are you will probably be able to afford the increased price of the memory. Never forget that the memory is the easiest way to make your computer faster. You almost never have enough RAM.
I'd never buy a P4 system without Rambus, but then, I'd never buy a P4 either because it cost just too much.
This is a really odd announcement, especially considering this article released by Tom's Hardware Guide yesterday. Basically, the article tested the upcoming P4 chips at 2666mhz and 2533mhz, and stated many times that the performance gained by using both a 133mhz FSB and 533mhz memory clock would show enormous gains over any current Intel or AMD processor. Some of the performance numbers on THG's graphs were rather impressive.
So I see one of two things here. Either this report is wildly inaccurate or misinterpreted, or Intel sure is shooting themselves in the foot yet again. As soon as they create technology that can utilize Rambus' enormous memory bandwidth, they can the technology? Something about that doesn't sound right. Intel may be dumb, but they're stubborn too. If this is true, they sure picked a hell of a time to seriously rethink their relationship with Rambus.
Rambus was just starting to become affordable (on par with DDR), there are a ton of new benchmarks showing that in several applications, the P4 at high clock speeds actually benefits tremendously from Rambus. Just when it actually starts to look good, they decide to drop it?
Check out:
Tom's Hardware
"This is because the Pentium 4 has a problem: the increase in clock speed (e.g. P4/2533 or P4/2666) will be rendered useless by the slow DDR SDRAM memory bus of the 845 platform. In the mass market, the 845 chipset dominates by nearly 100% - and this will remain the case for the next six months. But only 533 MHz RDRAM enables the processor to attain high performance. Eventually, the dual-channel DDR solution will receive some sort of technological boost, however there's still no sign of development in this area."
I guess there's only one thing left to say: GO AMD!!!
-- Adam
Before you accuse somebody of taking money or doing other things slimy and illegal, you should have something to back that up with. Basically what Tom has said all along is that Rambus gives you better memory bandwidth at the cost of latency. What he used to say is that the latency is unacceptable and the bandwidth wasn't needed and the price is too high. What he says now is that if you look at P4s with clock rates up above 2500Mhz they appear to need that super high bandwidth to perform at their best. And recent price increases in DDR have brought its price in line with RDRAM. Where do you see something fishy? It's the sign of integrity when somebody admits that a solution they said was bad has become better. I wish more politicians would have the guts to do so.
I agree that Tom's Rambus reversal isn't based on some kind of payoff, but if it's based on technical reasons, he has yet to share them with us. His article on the topic was just shoddy.
In particular, he kept comparing his custom P2200 - 2600s to an XP 2000+ and blaming the difference on memory bandwidth. Well, the name XP 2000+ *means* "about as fast as a P4 2000". If the speed goes up on the P4, of course the XP will lag. So if in "MPEG-2 video encoding, the Pentium 4/2666 is approximately 25% ahead of the AMD Athlon XP 2000+", well, 2666 is approximately ahead of 2000. Where does memory bandwidth fit in? The rest of his benchmarks are similarly unsurprising and misattributed.
Also strange was the fact that in his review of the uninspiring performance of the KT-333, Tom failed to take advantage of MSI's 166MHz fsb option to see what the chipset was like when driven to it's full capacity. He could've just taken one of his custom unlocked chips and adjusted the multiplier to match the speed of the normal chips at 133MHz fsb. Maybe it wouldn't make any difference, but it seems a pretty glaring oversight. Wouln't you at least check?
It's too melodramatic to argue that he's on the take. But given his previous opinions, I wonder why he's so desperate to read more good things about Rambus into his data than are really there.
Kill, Tux, kill!
Sorry, but Tom Pabst is hardly the best hardware journalist on the Net. In fact, I find it hard to call him a journalist, period.
His articles are continuously, and blatantly biased -- and while the target of that bias does change, it remains that most of the articles read like National Enquirer stories.
Take, for example, the KT266A vs nForce 420 test. The benchmarks show the nForce in the middle of the pack for most tests - roughly half the KT266A boards faster and half slower. And with margins of 2% in most tests. Yet the "Conclusion" was that "KT266A Trounces nForce 420D" and that "the nForce 420D is currently no match for the new KT266A". What a load of crap. Of course little things like total system cost and features were ignored - the nForce has a significantly better sound chip than the KT266A and all nForce boards have integrated network (only some KT266A's do).
The P4/2666 and 533 MHz Rambus article is nothing more than sheer yellow journalism. Benchmarking a system that won't be available until at LEAST the end of the year, comparing it against currently available systems, and concluding that "this will put it quite a distance ahead of its competition from AMD" isn't journalism. It's being a patsy to the latest company to show you a new toy.
Sure, there's the Claw hammer preview... with nothing more than a few snapshots. At least it appears to be mostly devoid of sensationalist statements though. The above review of an unavailable system would've been just fine had Tom and his staff not stooped to phrases normally seen at the supermarket checkout lines. They even tried to put in some moderating comments, but they are overshadowed by the sensationlism elsewhere in the article.
Note, first of all, that it is in fact quite true that Intel is (planning on) dropping all new RDRAM-based designs from their x86 chipset lineup soon. The last "new" RDRAM chipset Intel releases will be the upcoming 850E, which is just the 850 but speed bumped to work with the upcoming 533MHz FSB P4s (due in April), and to match them with PC1066 RDRAM. Of course, this is very old news, known from Intel roapmaps obtained by everyone and their mom back in October or November.
But that's not what the article says. It's talking only about chipsets for servers and workstations, where, indeed, the 860 is being replaced by the just introducted dual-channel DDR E-7500 (Plumas) and the upcoming dual-channel DDR Placer (as well as a just-introduced chipset from Broadcom), and where the 850 will be replaced by the dual-channel DDR Granite Bay chipset, due in Q3 or so.
Thing is, dual-channel DDR for the *desktop* won't arrive from Intel until sometime in 2003, with the Springdale chipset. (Dual-channel DDRII, in fact.) VIA and SiS are both trying to get their dual-channel DDR chipsets out in time for the 533 FSB P4s (doubtful, but they should be in full swing by Q3), but, again, if you want the very highest-performing P4 desktop, and you want an Intel chipset, you'll either need to ridiculously overpay for a Granite Bay (workstation oriented) motherboard, or you'll have to use the 850E with PC1066 RDRAM, or you'll have to wait until Springdale in 2003.
So, to reiterate:
1) Yes, RDRAM is gone from all future Intel chipset introductions save the 850E, which is just a speed bump, not a new chipset.
2) But that's not what this article is talking about; it's only talking about servers and workstations.
3) RDRAM won't be completely gone until there is a dual-channel DDR chipset to replace it on the desktop; soon from VIA and SiS, not until 2003 from Intel.
Keep in mind that Tom himself rarely writes the articles on the site these days. The article you linked to is credited to Frank Völkel, Bert Töpelt, and Patrick Schmid. The P4/2666 article is credited to Frank Völkel and Bert Töpelt. The Hammer preview is by Frank Völkel.
However, Tom Pabst's name is the one that's in that logo on the top of every page (similar to AnandTech's Anand Lal Shimpi), so by doing so he puts his integrity at stake with every article.
Ian
I don't know about the rest of the world, but where I live (Denmark), the prices for DDR-RAM and RD-RAM are nearly the same.
Ah, from the mouths of babes (cowards)...
if you check pricewatch RDRAM and DDR have almost identical costs. if someone is charging more for RDRAM then they are just lagging behind the market...
I'm not sure, but I think it might have something to do with Rambus charging higher royalties from it's patents on DDR technology.
Rambus makes money no matter which memory Intel chooses. It just might make more off of RD-RAM. Rambus isn't going belly-up any time soon.
Remember "Bring 'em on"? *sigh
I've never run into speed or compatiblity issues with Rambus; I can not say the same for DDR memory.
I can't tell you how many chipsets i've dealt with in the past year that had issues with filling all the memory sockets; general timing and compatibility issues....damn....via even has speed issues regarding filling all the dimm sockets with ram on some of their chipsets! That's rediculous.
How many Rambus issues have I had? Let's see....NONE.
Granted, Rambus was expensive, but has anyone looked at DDR vs. Rambus lately? The pricing is almost equal.
Hopefully Intel will release STABLE chipsets and possibly force DDR technology to be as cross-compatible as Rambus.
-ted
Did they really think people wanted to pay that much for memory? Why didn't they just engineer a nice DDR chipset in the first place? Seems like a no brainer to me. Uh gee, people like paying less per performance unit so...
Derek Greene
I waited until the prices dropped. When RAMBUS memory stated appearing in circuit city I was delighted. It appears that supply has finally caught up with demand and the prices appear to be as good as other RAM prices. I see not reason to give up on the speek offered by RAMBUS because of manufacturing delays.
I always thought RAMBUS sounded like a brand of condom.
Wansu, th' chinese sailor
No sign of development in dual DDR chipsets? High-end dual DDR chipsets are popping up all over, and recent leaks show that VIA is working on a desktop dual DDR chipset.
Check Pricewatch before claiming DDR costs less. Really, it's about the same price as RDRAM. If Rambus weren't such a lame and litigious company, I'd be sad at hearing this announcement. I think the technology is good. I hope their engineers, who have done a decent job, get really pissed off that their stupid marketing/legal departments stunk up the company so badly that even Intel couldn't stay in bed with them. If I were a start techie at Rambus, I'd put up my resume today!
Hey, I like this idea, but let's not float it too soon. I think they should let Rabmus go bankrupt first and buy up the IP at the liquidation auction. There is no hurry, and if this announcement is right (and applies to all lines of future Intel chips) we won't have to wait long anyway. You can bet that Rambus people will be trying agressively to sell their IP pretty soon. I hope they get stonewalled.
People don't buy x86 for speed. They buy x86 (whether it's Intel or AMD) because it's very cheap and has decent performance for its price. If you want speed and price isn't a big factor, then you don't buy x86. You don't buy RAMBUS because your CPU isn't from Intel; your CPU is from IBM or DEC (RIP) or someone else, and you use whatever type of RAM they tell you to.
Having fast expensive memory systems for x86 chips defies that principle. There is no market for it.
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
But what about the recent article about Intel's newest architecture, using 533MHz RAMBUS DRAM? That article, and the comments associated with it, state that, price-wise, the RDRAM is no longer significantly higher than DDR SDRAM, and with the new faster FSB, will blow DDR right out of the proverbial water.
Is Intel reneging on their statements of yesterday, or is Tom's Hardware or Anandtech just getting false information?
If RDRAM is no longer more expensive and is in fact a better choice for a fast system, should the company be left out in the cold? Bear in mind that I don't like the way RAMBUS has been treating people, but I do believe that the best product should always be able to live on, and RDRAM has surely been making strides.
Lack of eloquence does not denote lack of intelligence, though they often coincide.