Slashdot Mirror
Intel Encounters Another Problem with RAMBUS
Palin Majere writes, "News.com is reporting that Intel is once again having problems with its RAMBUS memory chipsets. This time, it's affecting the i820 and i840 chipsets, and is located in the chipsets (MRH and MTH) that allow customers to use regular SDRAM memory instead of RAMBUS memory.
It causes memory corruption and has already caused Intel to cancel three motherboard designs as a result. " With the continuing shortage of high-end Pentium processors, and stuff like this, it's no wonder that AMD has been doing better and better.
1) To Intel: "Ha ha ha ha ha ha. Nyah nyah!"
R andom-Access-Memory, for those who aren't in the know.)
2) Let's hear it for DDR-SDRAM!
3) We don't have enough letters in front of "RAM" yet, no! I want SCFLEADDRAM!!!
(that's Super-Cali-Fragi-Listic-Expi-Ali-Docious-Dynamic-
4) AMD rulez! Oh man, I want a Crusoe. My K6/300 is just sucking lately.
---
pb Reply or e-mail; don't vaguely moderate.
pb Reply or e-mail; don't vaguely moderate.
This is just a minor nitpick, but the marginal cost will not move to zero, but to the per unit license fee that Rambus Inc. charges.
What I really don't understand is why Intel hasn't cut its losses on this. From an uninformed observer's perspective, it seems that there is a strong incentive for _all_ players to move to DDR in the near term and offload Rambus tech towards "special projects" which need tomorrows bus speeds today (which is to say not servers, or PCs, but research devices).
I'm guessing from an economic perspective that they are weighing the advantages of control (they do get sole distrobution rights from Rambus, right?) of the market greater than the traditional advantages of low fixed capital transition costs. This seems a bad strategy to me, since intel really isn't in the memory market, and that they're primary sales are in semiconductors.
Ah well... I'm sure their decision makers have more info than I do. :)
Pax -- Ob
Then, people started cloning the technology anyway. IBM sued. IBM nearly went bankrupt. It was winning the cases, but the sheer volume of them, plus legal fees, was bleeding them dry. They had two choices - open the specs, or die.
They chose not to die.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
For all their billions, I don't see how they can survive this one. A break-up, whether horizontal (the DOJ's preference) or vertical (which is what I'd like to see), could be fatal to Microsoft. It has no experience in dealing with a free market. And it has a large number of blood-thirsty competitors sharpening their knives, just around the corner.
Even if they did escape, this time, what would that do? The tower is wobbling, and each new battle will shake it a little bit more. The longer it survives, the worse the final crash will be. Sooner or later, investors, supporters and backers will realise this and jump ship. Not because they're cowards, but because nobody hangs around under those conditions.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
You know, sometimes people amaze me.
Exactly where are you getting these figures from? It's not Intel's or AMD's websites...
>To wit, SPECcpu95:
>Coppermine 800 MHz: SPECint - 38.9, SPECfp - 32.4
Taking a look at Intel's own posted benchmarks of the 800mhz Coppermine running on a 133mhz bus gives:
SPECINT - 38.4 SPECfp - 28.9
Source:
Intel's own website and benchmarks
Now, I'd compare these to AMD's benchmarks, but AMD hasn't published SPECINT results, and only publishes the base SPECfp results. (Which, by the way, show the Athlon soundly thrashing the Coppermine
In any event, comparing SPEC scores is a rather _bad_ way to judge system performance. If you know enough to extrapolate new benchmark scores from current ones, you should also know that there are much better real world tests available.
If you like, you can watch a 700mhz Athlon kick the snot out of a 733mhz PIII Coppermine running on a 133mhz bus over at Ace's Hardware
So next time, please take your results somwhere else or provide a real source for them.
Intel *had to* release these chipsets & boards to get people to use Coppermine. That's because the price of RDRAM is over 6 *times* the price of standard SDRAM. The RDRAM-to-SDRAM translation was added as an afterthought when intel finally pulled its head out of its ass and realized people are not gonna pay 6 times the price for a product that has no real advantages over the existing one.
___
___
If you think big enough, you'll never have to do it.
I've read a quote from the CEO of Rambus that it would have succeeded if Intel hadn't gotten in a hurry and pushed it.
Wouldda, Couldda, Shouldda...
Supporting the AMD Athlon is great, but alas, I haven't heard any news about any Athlon-compatible motherboard chipsets that will support the upcoming DDR SDRAM running at 266 MHz.
Once that happens, THEN we might seriously consider the AMD Athlon as a serious alternative for high-end server and workstation applications.
Raymond in Mountain View, CA
Opening specs in a stupid way (like IBM lisencing an OS and chip without any control over them) might be stupid, but building entirely closed systems is even stupider. Look at Apple. In the late 80's it was pretty clear that the Mac OS was superior to anything on the PC side, and Macs were competitive with anything on the PC side. In large part what killed them is that they got fat and arrogant. They overcharged their customers, ignored their needs, and generally pissed away their lead. Now that's not strictly a result of having a closed system, but without the competition from Mac clones, they got fat and lazy. As a result, PC's ate their lunch with inferior products, simply on the basis of better prices, and the increased varieaty and responsiveness of a competitive marketplace.
If IBM had successfully prevented cloning and Apple had pursued a substantial cloning program, we might very well all be using Macs right now. And in that case, IBM's PC revenues might very well be even less than it is now.
You're wrong on that count, I think... According to the benchmarks i'd seen in the past, the Athlon has mantained a sizable advantage over pre-coppermine Pentium III's... However, when the coppermine arrived with it's full speed cache, they're basically neck and neck in integer performance and AMD still has the edge in floating point performance...
Unfortunately for AMD, they have no SMP chipsets on the market, so the high end of the markets are still all intels.
I believe what you meant to say was that royalty-based consumer technologies don't get adopted unless they're supported by Intel.
_________________
rooooar
... the harder they fall. It would be nice to see Intel taken down several notches, IMNSHO. While Intel's stuff generally works (unlike the computer industry's other monopoly), it is over-priced, and Intel has a record of engaging in anti-competitive practices. I don't want to see them fail, but I think their grip on the PC industry needs to be loosened up a bit.
Intel tried to design a system that would be expensive to clone, and would corner them the market. It's failed.
Sure looks that way, although I would be warry of counting my chickens before they are hatched.
Back in the late 1980s, IBM tried a similar tactic, with a closed, proprietary, and expensive system bus called MCA. It completely flopped. People never learn.
Rule #1 of the Hardware Industry: Don't Try to Make Money Licensing Your Design. It is too easy for someone else to make their own design without paying you.
Intel's best hope of survival, never mind market domination, is to open the RAMBUS specs completely.
I wouldn't go that far. They haven't bet the farm on RAMBUS. Intel has other products outside of the world of memory. Perhaps you've heard of their Pentium line? :-) The failure of RAMBUS won't exactly feel good, but it won't kill Chipzilla.
dragonhawk@iname.microsoft.com
I do not like Microsoft. Remove them from my email address.
Funny how anonymous posters typically champion or defend the status quo, while registered users typically champion or defend the underdog.
Just an observation.
Napster-to-go says "Fill and refill your compatible MP3 player", which is a lie. It's not MP3. It's WMA with DRM.
The point is that the problem with Geforce cards isn't with the chipset, it's with the mobos makers that didn't put an extra capacitor on the motherboards (or on the graphic card manufacturers who didn't put an extra capacitor on their video card...I blame both for being cheap bastards personally -- I mean, come on...I've heard of being cheap, but still).
The problem with AGP 2x isn't with the chipset, it's with manufactures who decide to save .005 cents on a motherboard by not putting in an extra capacitor.
Good post!
Moreover, Rambus costs $1,000 for 128MB. Check out:
http://www.mushkin.com
..one of the few online dealers where you can even find it. Intel it betting a LOT on RAMBUS, and pissing off everyone with their flaky i820 and i840 chipsets. Check out how many big OEMs now offer Athlon system. Intel has really left the door open for AMD. If AMD can get to market with a chipset that supports DDR-SDRAM, 4X-AGP, and SMP, they will put a serious hurt on Intel.
The one thing that will keep Rambus Memory, Inc. afloat is the fact that Intel supports them, and it will be the memory in Play Station-2s. But if the price does not fall down to somewhere within the range of DDR-SDRAM (which already exists for video cards) it is going to fail in the desktop market and take Intel's i820, i840, Pentium-III, and Williamette with it.
I'm really looking forward to buying an Athlon box as soon as SMP and DDR-SDRAM support are a reality. The VIA KX133 chipset is already a very nice stable platform for uniprocessor/SDRAM setups.
---
True - but as I understand it the issues have been more to do with volume up to this point
What I really don't understand is why Intel hasn't cut its losses on this. From an uninformed observer's perspective
I think that long-term Intel probably wants to move the memory controller onboard for all but high end MP systems - one advantage that RamBus drams have today is that they allow more concurrency in the dram system - in particular more 'open banks' (ie sense amps with cached data) and more concurrent RAS cycles in the array. Making use of this sort of stuff is very difficult for a memory controller which necessarily sees transactions serially over a (relatively slow [compared with rambus speeds or cpu clocks]) slot1 bus. With CPU clock speeds getting faster and memory not the CPU architects are in a bind - they are spending big on things to make up for the slow memory latency (not memory transfer rate - read latency is the first order effect here that's the killer) like big caches. But their wonderfull superscalar and/or VLIW CPUs are stuck talking to a potentially fast memory system through a slow serial pipe.
The concurrency in the multi-bank architectures in the memory system can really only be used effectively directly from the CPU where the concurrency from the CPU architecture is directly expressed. My guess is that long term the Intel designers would like to pull the RamBus controller onto the CPU die so that they can attack their latency problems.
For the record there's another way this can work too - RamBus is narrow - you can toss 2-4 of them onto a die (if you can afford the area and power costs) where you can only afford the pins for a 64 or 128-bit bus. You don't have to run the RamBuses in lock step - instead you interleave the shit out of them (4-way for 4 interfaces - every 4th cache line from a different bus) this aagain allows you to increase your concurrency - at the expense of the customer having to stuff all 2/44 buses identically (2/4 simms at a time).
Having said all this I think that competing technologies are trying to push at the multi concurrently active bank thing too. I think that Rambus just started evangelizing that first.
MCA was a great idea. It worked like a charm on their 'larger' systems, and filled the need for a faster-than-ISA bus. VLB wasn't yet commercially produced, PCI was still a twinkle in someone's eye. IBM had peripherials available for MCA, it was better than ISA by an order of magnitude, and they could control it. What would you do?
.sig: Now legally binding!
I think the fact that there's just NOW coming to market a decent Chipset for the Athlon has hurt AMD quite a bit. I also think AMD should come out of the closet a bit and share what they know of why their Irontgate chipset isn't always compatible with AGP2x as it's spec'ed to be.
.18 micron processes (last I heard they were still at .25) and/or copper interconnects yet, and if not what the current planned dates are?
OTOH, Intel is having a MUCH harder time with the new boards (i820 and i840) - the number and seriousness of the errors on these things in crazy.
And combined with the disaster-in-the-making known as IA-64 (personally, I think it seems like a good idea on paper, but there are so many problems I don't think that anything good will come of it), and their production problems on high end Pentium III chips, Intel is not doing at all well.
By comparison, AMD is doing good. The Athlon is doing great and it seems that the architecture will hold up for quite some time (unlike Pentium IIIs, which IMHO are pretty much on their last legs as a viable design for new chips - hence Willamette and IA-64, neither of which will be here for at least 6-9 months). The chipset problems are a disadvantage, as is the lack of availability of SMP Athlon boards.
BTW, does anyone know if Athlons are being made with
Taking a look at Intel's own posted benchmarks of the 800mhz Coppermine running on a 133mhz bus gives:
SPECINT - 38.4 SPECfp - 28.9
The fastest one is one in a Dell system, which has the numbers I originally quoted (SPECint: 38.9, SPECfp: 32.4). Proof is here for SPECint and here for SPECfp .
Now, I'd compare these to AMD's benchmarks, but AMD hasn't published SPECINT results, and only publishes the base SPECfp results. (Which, by the way, show the Athlon soundly thrashing the Coppermine
You are a liar. AMD Athlon 750 MHz, SPECint: 33.0, SPECfp: 26.5. (800 MHz Coppermine: 38.9 and 32.4). Proof: SPECint and SPECfp . AMD hasn't published results for the 800 or the 850, because those processors have such poor performance.
In any event, comparing SPEC scores is a rather _bad_ way to judge system performance. If you know enough to extrapolate new benchmark scores from current ones, you should also know that there are much better real world tests available.
SPEC is not the be-all and end-all of benchmarks but it is *the* standard benchmark for scientific commputing, and by far the most respected CPU benchmark in the world. I would *really* love to see TPC-C results for Athlon, but they haven't been published yet (gee, I wonder why?)
If you like, you can watch a 700mhz Athlon kick the snot out of a 733mhz PIII Coppermine running on a 133mhz bus over at Ace's Hardware
And I should trust some ma and pa benchmarks over the most professional and industry standard benchmarks for what reason? For starters, spec is a dot-org and Ace's is a dot-com, so SPEC is inherently less biased. I don't trust benchmarks from dot-com sites.
So next time, please take your results somwhere else or provide a real source for them.
Well, it is next time, and the source is all above.
Your statement, "Intel has never had a processor of [sic] equal processor speed that outperformed Intel", is patently false. AMD's Athlon outperforms a Coppermine of equal clock speed by greater than 17%, in both integer and floating-point operations. Why do people still believe that Intel remains unchallenged?
Coppermine is faster than Athlon on industry standard benchmarks. Athlon is only faster on old benchmarks comparing Athlon to Katmai. On recently taken benchmarks, Coppermine wins hands down.
To wit, SPECcpu95:
Coppermine 800 MHz: SPECint - 38.9, SPECfp - 32.4
Athlon 750 MHz (the fastest for which SPEC is available): SPECint - 33.0, SPECfp - 26.5.
Athlon 850 MHz (scaled from the above, which is generous because the 850 according to some reports is very slow due to cache speed): SPECint - 37.4, SPECfp - 30
I would also LOVE to see the TPC-C benchmarks for Athlon, but they aren't even published yet. The Athlon's utterly pathetic L2 cache performance, as well as its lack of support of MP, makes Intel by leaps and bounds the winner in this area. If Athlon ever supports MP, Foster will already be out, which will seriously clean up in that area.
The only benchmarks which Athlon actually does better than Pentium III, is back when comparing Katmai. Coppermine is much different than Katmai, and performs better than Athlon at almost every benchmark. AMD won't tell you this, of course, and is maintaing benchmarks of Athlon vs. Katmai, not Athlon vs. Coppermine.
It is remarkable that Intel with its five year old architecture still beats the pants of AMD's massively hyped, brand new microarchitecture. And, Intel has a brand new microarchitecture (which has a 400 MHz FSB, a trace cache, a 3 GHz ALU, and a few other as-of-yet unannounced MAJOR features) coming out in about six months, which will most likely completely put AMD out of business (especially if Intel can catch up with its manufacturing problems, which is the real issue here).
Really??? I got an AMD 486 Dx-2 50 I'll let go really cheap.
AMD doing Better and Better??
I think the fact that there's just NOW coming to market a decent Chipset for the Athlon has hurt AMD quite a bit. I also think AMD should come out of the closet a bit and share what they know of why their Irontgate chipset isn't always compatible with AGP2x as it's spec'ed to be.
And while Intel may have they're bugs, they're very public these days, and therefore they're fixed very quickly, and you can typically get the fix without too much trouble or cash. Granted, that's not necessarily the case with the RAMBUS issue, but who has the money to buy the stuff at this point??? (:
Yes, I may get flamed for saying that bit about the Irongate, (Some think it's the Athlon MB Manufacturers not meeting spec, and nVidia hasn't entirely sidestepped blame..) but it's AMD's Processor AND Chipset. If something's not meeting spec, they should do something about it, or at least make the consumer aware of a problem, and what to watch out for.
So while AMD may truly look to be doing "Better and Better" both chip makers still have their own problems to deal with.
The benchmarks I've seen agree with what you are saying. Don't forget that AMD is integrating 256K of L2 cache similar to the coppermine. If all goes well then AMD should be able to recapture the speed crown until Williamette comes.
It's really great to see AMD doing well, but people get a little carried away rooting for the underdog. I am an AMD fan, but what I really want to see is AMD and Intel both competing hard. Both with good size market shares. Each keeping the other honest and forcing technical innovation.
AMD is making doing better and better because their chips are getting better and better. I used a K6 back when no average user had ever heard of AMD. Now the athalon is all the rage. people are using these chips because they're DAMNED GOOD, not because intel's chips are crappy - that's just a perk :)
-FluX
-FluX
-------------------------
Your Ad Here!
-------------------------
"It is seldom that liberty of any kind is lost all at once." -David Hume
Historically, not a very good argument. Remember the original IBM PC? They published all the specs openly, and then were immediately surpassed in the market by a startup known as Compaq, who could build on (& tweak) the design without having to recoup the millions (or more?) that went into the original research. IBM all but died in the PC market, shortly after initiating it.
Of course, then there's also the Apple story. Keep it all locked up & sue anybody whose product shows the slightest resemblance to yours. I remember reading a quote several years ago that went "You're nobody in the computer industry until you've been sued by Apple."
This seems to indicate that neither strategy is a good way to go. The companies that end up the winners are the ones that wait for somebody else to make a breakthrough, then make a small improvement (be it speed, on-board cache, or whatever) and sell at discounted prices. The only expenses to be recouped are the costs of reverse engineering the original, not development from scratch.
- - - -
Correction: DDR-DRAM is much faster then Rambus. 100MHz DDR-DRAM has bandwidth of 1.6GB/s, same as Rambus. 133MHz DDR-DRAM has a bandwidth of 2.1 GB/s (that's giga *bytes*, btw, not bits). DDR-DRAM, as well as regular SDRAM, also have a significantly lower latency.
And as it happens now, memory bandwidth is not the bottleneck. Even 800MB/s of regular PC100 SDRAM is plenty for 99% of applications, including the latest 3d games. Just about the only thing that would make use of the higher bandwidth is large databases. Too bad you can't put more than 512MB of RDRAM in a machine... ;-)
However, lower latency is guaranteed to boost performance a bit, no matter what kind of application you are running. This is where standard SDRAM and the upcoming DDR-DRAM have an advantage over Rambus. Not to mention the cost...
So, the whole situation can be summed up in one sentence: Rambus is just an inferior product with a ridiculously high price.
___
___
If you think big enough, you'll never have to do it.
DDR-SDRAM is great, but while you can get them in quantity, FINDING a motherboard that supports it natively is quite something else. :-/
I want to know when will the VIA Apollo KX133 chipset be upgraded so it will support DDR-SDRAM.
Raymond in Mountain View, CA
While I don't think that this is the death of Intel, for they have too many fingers in too many lucrative pots, it suggests that they have misstepped badly with RAMBUS are are going to lose their market dominance in CPUs if AMD keeps it's act together... which other than Irongate and a scarcity of Athlon MBs, they have been doing fairly well.
As far as Irongate's AGP issues, there isn't a whole lot of difference performance-wise at this point between AGP 1x and AGP 2x. Maybe in the next iteration of video cards we'll see a more significant difference, but I'd rather have a CPU that is 15%-20% faster than sweating about a 4% hit on the AGP bus. Some people feel that Athlon is not being entirely honest & ethical with the issue, which may taint their reputation in the long run.
My next machine will be an Athlon based system. I've suffered extreme technolust since they were released, and they just get better and better.
It will be quite some time before Intel has anything in market to compete with Athlon, and by that time it might be too little too late. Their most recent efforts have yielded uncertain results in comparison to the Athlon.
-- benton.
Rambus (Inc.) is a company!
They(Rambus Inc.) have designed a memory type called DRDRAM that only uses a 16bit wide external databus, and 8*16bit wide bus internaly.
As always when it comes to Intel it's only MHz that counts, not what they do with those. DRDRAM can handle 800MHz but as the bus is only 16bit wide it wont be very much faster than the 64bits(At most twise.).
I'd put my money on SLDRAM, it will be atleast twise as fast as DRDRAM and is, unlike DRDRAM, an open standard. SLRAM shouldn't have a problem doing 3GB/s+, at much lower clock speeds.
"Last words are for fools who haven't said enough." - Karl Marx
It's failed.
I hope Intel, and other chip manufacturers learn from this. Secrecy and control aren't cool. They can, and will, turn around and bite you.
IMHO, Intel's best hope of survival, never mind market domination, is to open the RAMBUS specs completely. Do a hardware variation of the GPL. If they don't, it's going to bleed them dry. If they do, sure, there'll be clones, but Intel will still exist.
Given the choice of pride or survival, Intel needs to think about that survival option a bit more.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
The MRH's are Memory Repeater Hubs, and come in two flavors - MRH-S and MRH-R. The -S does translation from rambus to normal SDRAM. The -R is to let you get aronud the 2GB limit on rambus. A rambus channel can have up to 32 256Mb devices, or 1GB. An MRH-R has two channels dangling off it, doubling capacity. Put two MRH-R's on one channel from the chipset, and you can have 4GB.
One interesting thing about this solution is that it takes time to go through the chips, increasing the already high latency of rambus.
I wonder if Intel would alter their decision on Rambus, were they able to go back in time and do so. They might pull it off yet, but it won't be easy. If it does work, it will only be because they are Intel.
The enemies of Democracy are
Reading the article my take on it is that the problem is in the device that does the SDRAM to RamBus conversion (ie it's a channel adaptor that lets them mix and match rams types) - and the problem only occurs when you use ECC.
I can think of 2 reasons this might happen - either they got the ECC logic wrong (probably likely), or there's a noise problem on the sdram side when they drive 72 data pins [for ecc] rather than the usual 64 (less likely). Either way it isn't a RamBus problem.
There's a lot of noise made about the various merits of memory types - my personal take on it is that it's mostly a wash, RamBus drams do have some advantages - but for main memory systems they are more in the future (and revolve around how many chips it takes to make a minimum memory sized system as memory continues to move down the memory density curve - M$ may of course make this moot). Their main disadvantage is cost - and it's rather a chicken and egg sort of thing - if people use them a lot the marginal cost of RDRAMS will probably go close to 0 - but if people don;t use them in volume because they cost more that won't happen. Remember in the core of a RDRAM is the same core that's in an SDRAM it's just the interface circuitry to the pads that's different.
Rambus is a design for a memory system from Rambus Inc. It is extraordinarily fast on paper. Intel chose their design and decided to support it on a lot of their new products.
The implementation took a long time to get around to getting around. It is now here. Intel bet a LOT on Rambus, because it would give them significant control over a lot of markets. (IE: They own rambus designs)
Rambus is significantly different from the DRAM used commonly today. It requires changes to how stuff is laid out on the motherboard. And it is manufactured differently, to very demanding tolerances.
It is now in production and is competing with DDR-DRAM, which uses existing manufacturing processes, generally works with existing chipsets, and is easy to support. And it doesn't require a fan setup for the memory alone. And runs far cooler. And gives almost as good performance when set up correctly as a RAMBUS setup. And is also capable of being manufactured in quantity, whereas RDRAM is extremely difficult to manufacture. DDRDRAM is also about a fifth of the cost of a RDRAM setup.
You do the math, and read up on it a bit.. I think you will agree that for all intents and purposes (read: mainstream pcs, servers, et al), Rambus is DOA.
Toodles.
-troll taker