Slashdot Mirror
Is Rambus Destined to Return?
An anonymous reader pointed us to an article running over at Tom's that talks
about the world
of ram and criticizes the performance of DDR. The article goes
into DDR333, DDR400, and Rambus, and explains the issues at higher
clockspeeds.
Experience tells us that Rambus is faster.
Pocket books tell us that ddr is better.
Which will your wife let you decide on?
Given the bad performances of RDRAM due in large part to its insanely high latency, and Rambus' dubious business practices based mainly on trying to milk patents to leech on the entire memory industry's back, why on earth should anybody give then the opportunity to make a come-back ?
It seems to me that Rambus has offended so much of the industry that it even intel's continued (though lately lessening) support, or perhaps especially with intel's support it will fail to be implemented by the majority of m/b manufacturers.
Other avenues for gaining speed exist- like Nvidia's extra memory controller for the gpu in the xbox and higher end nForce chipset.
With the exception of the shady business practices of Rambus, I don't fully understand why Intel dropped RDRAM in the first place. In every benchmark that I saw circa 7-8 months ago, the huge amount of memory bandwidth present gave Intel one of it's only advantages over the corresponding AMD CPUs.
It couldn't have just been the prices either, because Intel obviously knows they're not going to win that race.
Anyone?
Palaces, barricades, threats, meet promises
Sorry...
Why is Grand Theft Auto a much more serious crime than Reckless Driving?
Seriously, ignoring pure performance considerations, RDRAM is garbage. It has to be put in pairs, and if those pairs aren't made by the same manufacturer, I've seen motherboards refuse to boot. Heat is a serious issue, and I've burned one finger too many on those heat spreaders. I've also seen an analog cable coming from the cdrom get stuck between the RIMM's and melt to the heat spreader. And price is still an issue, although it's improved quite a bit recently.
Expensive + Has to run in pairs + Runs very hot == Useless to me.
---- El diablo esta en mis pantalones! Mire, mire!
This can be a problem. You should be able to make back the money so you cover your costs. Unfortunately, you may have to have deep pockets to stay in the game for a long time.
"It is a greater offense to steal men's labor, than their clothes"
It's not news that a CPU which was designed to use one memory doesn't perform as well when using a different kind of memory. The PIV needs memory bandwidth desperately. How was Intel supposed to know that all those Alpha engineers would go to AMD and give the public a decent alternative? How were they to know that the public would have the option of getting better performance for less than half the money? The PIV is the last processor of an era AMD just put an end to. It's foolhardy to derive the future of a memory technology from its performance in conjunction with a misdesigned processor. If you could test an Athlon with DDR versus RDRAM, of course, the DDR would perform better. Please let's not post "news" from Tom's anymore - Tom's has just really gone to crap.
Much like SGI boxes for the average desktop, Rambus plays it's part to a niche market. (Though I'll be damned if I'm going to stay in a job that makes this ram standard build)
When price isn't an issue, and politics not a motivater, it's amazing what ends up in the niche.
-Slashot is like a sewer, what you get out of it, pretty much depends on what you put into it.. - Updated Tom Lehr.
Canadian Bred with American Buttering
The fact is that most IC testers today support the lower-speed parallel connections. High-speed serial connections like Rambus and SERDES require very expensive mixed-signal testers with expensive and complicated load boards (the PCB between the tester itself and the chip). These high-speed serial I/Os on the memory ICs themselves are also generally much larger than on a DRAM, probably by a factor of 5. So, you don't get die savings, you don't get lower test costs, and most of all you don't have any processors whose front-side buses exploit this. Plus, you have very expensive target products in terms of motherboards to support the Rambus ram requiring tight trace routing and signal isolation, and their very limiting 28ohm max impedance (at least with the PC800 RDRAM), almost completely opposite in difficulty to DDR. So where's the advantage?
If you also figure that the memory controllers for Rambus are configured for dual-channel operation, it becomes much clearer that the advantage is not in the memory architecture itself but in the controllers. Suppose a server board manufacturer decides to support quad-channel PC2700 1GBx4. That's 10.8 GB/s of potential memory bandwith on sequential accesses! There's hope with chipsets like the Nvidia nForce420 dual-channel DDR, but the Athlon FSB is the limiting factor there. And let's not get into the infamous first-access latency issues which I hope they're finally addressing.
Rambus is also notorious for poor tech support. I worked for a major silicon vendor using their core, and they never responded to our requests for minimum PLL-to-Rambus core distances. It was abjectly ridiculous, but not surprising considering that regular SDR/DDR memory interfaces outnumbered Rambus designs 100:1. Have things changed? Considering what their legal bills have been lately and an erosion of their tech support, I doubt they can afford to improve it much.
If you see an interesting article on another hardware site then submit it. The submit story link is on the left hand side of all /. pages.
---
http://slashdot.org/moderation.shtml
Have a look at the SUSE linux kernel compilations with dual athlons here. As everyone with dual processors knows, compiling kernels is one of the main advantages of having dual CPUs. In this test however the athlon dual boards got worse scores than their single processor counterparts. If they had just run make with the "-j 2" option they would likely have gotten around 80% faster than than the single processor boards.
--
WHO ATE MY BREAKFAST PANTS?
on what systems you are working with? if you want a performance p4 system then obviously you use rambus... and if you want an amd system you use ddr(since there is no rambus/athlon chipset)
and until there is a rambus/athlon chipset i don't really think we can gague the real world implication of it...
either way i have better things to do with a few $100 than put it into a more expensive chipset/cpu/memory rig. if you have the extra money and the rambus system gives you what you want, then more power to you. overall, right now, you can't say either system is "the best" in ever possible catagory
I believe sex is highly over rated... unless it involves me
Good try, but you're TOTALLY WRONG.
Been reading too much Tom's Hardware Misinformation Guide?
I think the point of that article is not that the Athlon is going to burn up.
There was a rumor being spread by some kids with AMD processors that the Pentium 4 runs at half speed whenever you do more than just checking email with your computer. They had taken a new feature in the Intel processor and manipulated it into a fault. Tom's article was only trying to explain what the feature is and why it's good, since many people did not understand what it was about. He was not in any way saying that your heatsinks are going to fall off and your processor will burn up, if you use an Athlon processor. It was not an attack on AMD, it was just an explaination of a new feature found in certain other processors.
I have a few problems with RDRAM.
The first one is the price. It's simply more expensive than DDR Ram as far as I can see.
Secondly, it appears to me that we are getting to the spliting hairs, angels dancing on a needle level here with RAM. Unless it dramaticly increases my boot time, time to do things in a word processor, makes mySQL fly like a greased dolphin, gives me kickass FPS in UT, or makes my G4 fill my bowl with Fruit Loops in the morning, I'm not going to spend one cent more for RAM than I have to.
If RDRAM is 101 dollars for 256 and DDR is 100 dollars for 256, I'm going to go with the DDR and the hardware that supports it.
Who gives a crap if RAMBUS is on the way back? All /. users know that even if the number of channels double, Bill Gates will just jack up the memory usage of Windows. Then everyone will just start bitching about it. Then Jon Katz will just post an article about something that has nothing to do with "News for Nerd". The cycle never ends... ... ...
About a year or so ago, intel was doing demo's of Q3A at fry's to show off the new P4. Economy was good, I had like 4k from PTO that I aquired between jobs so I said what the hell and dropped $1500 on a 850dgb board, processor, case and 128 of ram.
I gotta say, this stuff is hot, my friends have all gone off and bought gforce3's, amd's with DDR. I thought these new cards/systems would have score winmarks well above my own (around 3800 with a gforce2gts) but I was surprised to see they only score 1000 or so more than me.
Out of curiosity, we put one of those GF3's in my system. Without fail I would score about 400 to come in around 6300 3dmarks above my buddies amd1.6. My P4 is just 1.4. Yet even with a lower clockrate the memory bandwidth made a huge difference.
I'm not trying to cause a ruckus here, anyone with deep enough pockets (or access to enough systems) can just as easily do the same testing I did. Bottom line whether or not the moderators like it is rambus systems do provide the absolute best possible performance in 3D gaming. It certainly was expensive when it came out but now with the falling prices of all ram, it's within reach of anyone that want's that extra "oomph" in thier system.
Does anyone know of any AMD boards that use rambus? I'm sorta curious what kind of scores those get in comparison to the intel one's. Anyways thats my comment.
" Why does slashdot post a story every time Tom has a new article? Agreed some of the reviews are really well done but it seems like 2/3 of tom's articles get posted on slashdot."
Probably because we all argue and yell at tom and each other. discussion is good...then again, any article comparing any two products is heavily debated, from distro's to text editors. I guess tom must be special.
Who is this Anonymous Coward character, how does he post so much, and why is he always such a whore?
I'm generally not too fond of Rambus. Why you ask? Look at what's it's done with the N64. Framerate problems, jaggies...etc. I'm just fine with my SDRAM, thankyouverymuch.
That Intel needs Smallbus, err, Rambus memory to keep on par with AMD chips.
This is the 'return of Rambus!'?
Please. SDRAM is the standard. DDR is entrenching into that market. Rambus? It's like the Mac - some people wonder, 'What's that?' while the techs laugh at people who have it.
Rambus is a horrible. The technology? No, the company. Not by speed, but for business should we continue ignoring it. They are a horrible company, and despite their products, should not be dealt with as a result.
The technology.. Isn't really any better or worse than SDRAM/DDR save for price. I've seen boxes refuse to boot when two different brand yet same speed/size SDRAM chips were inserted into a computer. I've seen bits 'o SDRAM cause page faults, kernel panics, etc.
I've seen Rambus do the same. *shrug*
It wasn't a bad article... I mean, the facts -do- show that the p4 runs better with RDRAM, and he addresses the consequences of that quite well, and quite neutrally. For that I commend him.
But he does misrepresent some issues. For example, signal integrity issues. I can say with complete assurance that Rambus is loaded with signal integrity issues. These issues get -very bad- as the clock frequency goes up. Also Rambus is -not-, strictly speaking, a serial bus. First, it is 16 bits wide, while pure serial would be 1. Second, the depiction of a DIMM as being a unterminated stub with significant SI issues is correct, but this doesn't go away with rambus, and this definition of "serial" fails as well. While the signals do pass through a RIMM continuously, eliminating the RIMM itself as the source of major SI problems, you still have each and every RDRAM device itself acting as an unterminated stub, each of which causes reflections of its own. Especially for devices with tolerances as low as RDRAM, this can be difficult to manage. While in the balance I'd have to concede that at a given clock frequency RDRAM has the SI advantage, remember that RDRAM needs 4x the clock frequency of DDR to match bandwidth.
Or you could have 2 channels of rambus, and only need 2x the frequency. Well, 2 channel DDR is becoming a reality. Not only does nForce support it, Sledgehammer will as well. Neither of these are Intel platforms, but I would guess that going dual-channel would be a natural step for VIA and others competing with Intel chipsets. It would especially make sense for p4, as it would more than make up the memory bandwidth disparity that currently exists.
Speaking of nForce, another thing I take issue with is the suggestion that the nforce's DIMM-slot population problems are indicative that DDR is crippled by SI issues. I think more likely is that this was the first chipset designed by a company whose experience lies solely with graphics cards, on which the ram is directly soddered to the PCB. Lack of experience in the harsher SI conditions of a computer motherboard are to blame.
Speaking of DIMM population, it's hard for me to see only having 2 DIMM's on some boards as a particularly black mark for DDR... That leaves you with 2GB per channel, the same as RAMBUS.
So, he was right about some things, insightful on others, but the picture is -not- so clear-cut in the image of rambus Inc.
The enemies of Democracy are
QDR is coming out soon, (though they are calling it something else as I recall, no idea why, QDR is such a nice logical name, even the laymen can understand it) and it (seems to be?) but a mere advance of DDR technology.
:) (it never would, latency is too high, part of the base rambus technology)
:)
Not to mention how far up Nvidia has managed to scale DDR RAM. Heh. I would like to see RAMBUS get that.
RAMBUS would settle down good in a video toaster type of applicance, but that is about it. Video editing seems to be one of its few strong suits.
Besides, I would like to see a Motherboard that is halfway cheap and can support 3-4GigaBytes of RAMBUS RAM.
Need help treating your acne? Come here!
The days of cheap memory are over.
They say this because of the huge expense needed to provide 512MB or more of ultra fast memory. But what if they added yet another level of "cache"?
Put in 128MB or more of super-fast RAM (faster than today's RDRAM or DDRAM, maybe using an exotic bus) backed by gigs of cheap, easy-to-make memory (PC266 DDRAM or slower). The cheap ram is still orders of magnitude faster than a disk drive. Manage them with hardware that does page swapping similar to virtual memory.
You could get good system performance and lower overall cost.
It runs REALLY hot. The machine one of our techs has needed an extra fan to cool the memory!!
But really Rambus is not the solution, but another technology will finally arrive. Damn it, I want those quantuam computers with 3D optical storage!
How much does Tom get kick-backs for supporting Rambu$? They are one of poster boys for Patent reform for both consumers and the patent holders.
The one thing it's good at, it doesn't get to do...
Then the political aspect is ignored and he talks almost exclusively about technical issues about why Rambus might theoretically be better, and uses existing intel chipsets as evidence.
Hello? Answer the question, please? Has Intel ever come out with a non-crippled DDR chipset for the P4? How do Intel's DDR P4 chipsets compare to non-intel DDR P4 chipsets? (ARE there any non-intel P4 chipsets?)
How much of the problem is political, and how much of it is a real technical issue?
Thomas Pabst (who we all respect) posted scathing reviews not only of Rambus the company but also of Rambus the technology. If he is recanting he should do it in person, not through a couple of stoolies. By withdrawing such a controversial statement Tom's site is calling into question both the technical and political validity of his write-ups.
Don't get me wrong, THG rocks and I respect Tom's advice. He knows 10x more than me about hardware. But he should explain why this review is so opposed to the ones he wrote himself...
Now, unless you're doing video encoding, I don't know why you'd want a P4 in the first place...
:)
But to even entertain the thought in the first place, you can't be on an ultra-tight budget! So why try to save what will ultimately be a small % off the total price at the expense of performance?
It just doesn't make sense.
Though I still want to see a dual-channel DDR chipset for P4.
The enemies of Democracy are
Intel uses testers from schlumberger (their reps are quick to point that out). Typically, a tester cost anywhere from 1-10 million dollars + plus they require a lot of maintenance, calibration, etc. Basically, the faster and more pins you need, the more it cost.
I've worked with some schlumbeger 'KX' testers, they're a big pain in the ass, unreliable, and are badly designed: just shutting the thing off can break it!! (especially if you use the emergency off button).
There is another choice, however, you can use 'bist'(built-in self test) and have the chip basically test itself :-). This allows companies to get away with using cheaper, more reliable testers.
Um... How was that FUD?
It's true. If the heatsink falls off your Athlon it is toast. (note that just in the last week or so a board was released that supported the XP's thermal diode... but for all other boards/chips, you still get toast)
Tom isn't the genius a lot of people think he is (or that he'd want you to think), but that was not FUD.
The enemies of Democracy are
First, it looks to me like RDRAM is still about double the cost of SDRAM, according to Tom's Hardware's own price guide.
They have $93 for 512mb SDRAM and $175-250 for 512mb RDRAM.
My question is this: Let's say I have a choice between 512mb of SDRAM and 256mb of RDRAM. Would the SDRAM not almost always be faster because RAM, however slow, trumps swap space every time?
In other words, isn't the amount of memory I have more important than how fast it is?
Many moons ago, I had a SGI O2 workstation. Tremendous memory bandwidth, but memory that cost 10x more than anything else. As a result, it could be embarassed by lesser machines, since I couldn't afford to load it up with RAM.
I see Intel repeating the same mistake when it decided to focus on RDRAM.
Apple is putting L3 cache in their G4s so that the use of expensive RAM is confined to a relatively small and affordable amount. I can upgrade my PC133-equipped G4/450 dual processor to the latest 1ghz dual processor, put my 1.5gb RAM in it, and fly. That seems like a good compromise to me, maybe better than going to DDR, which I would have to buy new.
Thoughts?
D
I've noticed that alot of slashdot posters hate rambus with a passion. Why is this? I have not really been paying attention to the history of rambus vs. DDR. Other than hearing stuff with "rambus sucks" or somthing like that in them. Could someone post an article or explain the history of rambus and why people hate it?
I work for a test and measurement company and we sell logic analyzer tools for both DDR and RAMBUS. I service at least 1 site of each of the major computer manufactuers and I can tell you none of them are even considering RAMBUS. In fact, I can't remember the last time someone asked me about it. The only thing I have consulted customers on is the future of DDR. If anyone was interested in RAMBUS, I'd at least be hearing murmurs. Keep in mind I am only looking at computer manufactuers, not the 3rd party Asian motherboard manufactuers. Who knows what they are doing.
Instead of focusing upon some benchmarks with overclocked CPUs squaring off against one another, why not pay attention to another article from Tom's that shows some benchmarks of the P4 2.0 ghz operating at factory spec on a number of different P4 boards with different memory configs(P4 + i850 + RDRAM vs P4 + Sis645/ViaP4X266/ViaP4X266a/i845 + DDR). As you can see from that article, at least on the pre-Northwood P4s, DDR did pretty well on the non-Intel chipsets, particularly on the Sis645 when PC2700 was used.
Tom doesn't even write half of the content for Tom's Hardware anymore, anyway... People really should learn to read the author credits on each article...
...
Not that Tom and Tom's Hardware aren't a nVidia/AMD owned FUD machine with less brains that a small primate... Not passing a -j n to build kernels on a SMP AMD machine = stupid, showing your monitor refresh rate for LAME encoding = stupid, selecting items as "winners" that perform in the middle or toward the bottom of your own benchmarks = stupid,
Though some might argue that the role of the preview option is to avoid such errors, I am inclined to agree with you. Either due to a lack of willingness to bother to preview, or simply because mistake initially escape your view, an "Edit" option would be most helpful. I'm almost entirely certain that many would argue that it would be abused (for instance, posting something insightful and receiving a bonus, and then swapping to an ASCII penis). Of course there are numerous safeguards that could be utilized to reduce the likelihood of this.
I'll take the Glad Bag!
Remember "Bring 'em on"? *sigh
they've relied alot on patents with a few shady practices. Your average slashdot reader doesnt know anything about ram design though, so they tend to just follow the loud critcizers..
-
It didn't stop a bunch of fools from seeing that and saying "oops, my heatsync fell off! AMD 5UX0RZ!!! 1N73L is 1337!". Understand this: When we're talking about people too retarded to put on a heatsync correctly, the facts are just an unfortunate mishap.
It's been a long time.
Yes, I realize he doesn't write most of the articles. But it's called Tom's Hardware. By virtue of the name alone, the quality of the content, regardless of who it is written by, reflects on him.
If he doesn't want to look like a dipshit, he shouldn't put dipshits on the payroll.
The enemies of Democracy are
After Tom's Hardware printed that article where all they glossed over the PS2 and Gamecube as if they were useless (and all the information about those 2 consoles seems to have come straight from Microsoft) and then devoted 12 pages to details about the Xbox, I'll never really believe what they write about products again. I'm not even really sure if I believe all their AMD benchmarks, anymore. They were clearly bought by MS for that, so they probably were bought by AMD for those other articles.
If you doubt these words take a look at Compaq's
Alpha EV7 with it's RAMBUS controller ON CHIP.
Just because PC architecture is limited by chipsets with limited memory bus bandwidth does not mean there are no other uses for such a memory architecture.
Peter
www.alphalinux.org
For a simple enough high speed link, you can do I/O bist using external loopback. Or alternatively, use a second know good device on the loadboard to test it. Of course it should be used only for detecting manufacturing defects, you need to do a lot of testing up front to convince yourself that the part will actually meet specs across process variations.
Funniest... post... EVER!
To be honest, I really don't plan on buying another desktop system ever again. If needed, I'll build one out of spare/cheap parts. Why? Because I use my laptop anymore 99% of the time.
There will NEVER be a laptop with RIMMS in them because they are too damn hot. Unless the design of them drastically changes in some unknown way, this "NEVER" is a fact.
I think there are some DDR laptop solutions in the pipe now. Yet, there is the problem of the slower system bus speeds on laptops, so it will not matter much until that's fixed too.
I think the author went out of his way to say that your Athlon is going to burn up if your heat sink falls off. Actually they made a video.
And while I do remember quite a few of those plastic 486 fans falling off, I have yet to see an Athlon fan ever come off on its own during normal (ab)use.
I lost faith in THG when they started sticking $300 GeForce3 cards into systems with onboard graphics (for testing purposes). Talk about not comprehending the value segment.
If you guys think DDR isn't fast enough, just TRY to pass Paranoia on maniac mode... It's got more steps than a ladder to the moon. :D
Woah, you meant RAM?
---- I'll take you in a Hunt deathmatch any day.
In other words, isn't the amount of memory I have more important than how fast it is?
I think this is probably true.
However, as you point out, the price difference between RAMBUS and SDRAM is now very small. According to Sharky Extreme the difference between 512 MB of SDRAM and RDRAM is about $80, and DDR RAM (PC2100) is actually more expensive than RDRAM! So if you plan to put 2GB in your machine, SDRAM is appreciably cheaper, but if you plan to do that, you probably plan on some serious hardware as well, so you'll probably spend $3000+ (probably coulnd't get a motherboard that would take 2GB SDRAM anyway...).
My point is that both SDRAM, DDR RAM and RDRAM have come down in price dramatically in the past year (although memory prices seem to be on the rise again). The price difference is very small when compared to the total price of the machine, so why bother? I have nothing against DDR RAM, but it'll have to win on technical merit nowadays.
As an example, I had to specify and buy a PC for my job some weeks ago. Now, this PC will be running a very specialized application, and nothing else. No CD burning, MPEG/MP3 encoding, no image processing, and no games. I like a cool machine as much as the next guy, but I simply could not justify putting more than 512 MB in this machine. Same for the hard drive, 40GB should be more than enough. A decent monitor was a requirement however. So why save $80 on memory when we're spending $700+ on the monitor alone?
Summarizing: if 512 MB is enough for you, why bother? If it isn't enough, you'll likely spend a lot of dough anyway, so again, why bother?
MSN 8: Now Microsoft even has bugs in their ad campaigns.
You arguement is invalid. DDR isn't so cheap anymore. The only thing DDR has going for it--besides the Athlon--is super high density. 1GB modules.
The falling off heat sink thing is FUD. In the several year period working repairing PCs on a daily basis, I came across ONE heatsink having fallen off. It was on an old K6-2.
I've never seen a newer Socket 370 or Socket A HSF fall off. They are simply on much too securely. This included boxes that were moved by idiots to different locations so gently that the HDDs were destroyed. The HSF was still attached though.
What percent of CPUs do you think have their heatsinks fall off? Less than 1% probably.
Making people worry that their computers will melt down over a less than 1% chance of something happening, is spreading Fear, Uncertainty, and Doubt.
Ok, I'm going to try to keep this short even though I could talk about RAM for days on end.
.5 cycles per random access (to DDR) will make the difference. It won't. First, we're talking .5 cycles out of ~10 (for a purely random access), or a 5% increase. Second, bursting from the RAM almost entirely disguises this so that CAS makes up less than 1% by the time its all said and done. This happens because burst accesses (which are almost all that takes place in a PC, due to the cache structure) pipeline and end up hiding the CAS time over a long period.
There are quite a few inaccuracies in Tom's article here. From calling Rambus serial to some of his crazy statements about DDR not handling impedance balancing as well as Rambus, I'd call him insane.
However, its a statement like 800 Mb/s/pin is better than 333 Mb/s/pin that really gets me fired up. Yeah, those are the speeds, but he's losing so much of the big picture that its scary. The more important thing to account for is the geometry of the RAM configuration which includes things like the number of data pins in parallel. For instance, a DDR bus has 64 while Rambus uses 16. Lets see here, multiplying 64*333 and 16*800 (pins*(Mb/s/pin)=Mb/s total) shows that DDR333 is now twice the performance of Rambus. Yikes.
And then there's his comments about CAS latency. Tom says that CAS 2 will be the factor that makes DDR333 outperform Rambus because CAS 2.5 is too slow. Lets just sketch out what CAS means: it is the time between when a read command is issued to the RAM and when the first piece of data comes out, measured in clock cycles. So basically, he's saying that
Gads, I've been doing too much work with RAM lately.
Sure we can, you've never seen the i820 PIII chipset with Rambus? It wasn't as fast a the BX with PC100. The PC800 also cost 4x more.
The P4 is a dog. A P4 platform requires: a more expensive CPU, more expensive RDRAM, and the expensive i850 chipset motherboards. And then, it still isn't much faster in apps where it isn't beat by the Athlon.
But, I would recommend it highly to people who use the Lightwave 3d modeling app, and Windows Media Encoder!
If I don't need to insert data I use hiew (hacker's view).
I've used it to fix other people's software many times. One case had tcl source code stored in an Oracle database with no official way of altering it. So hiew it was...