Slashdot Mirror
DDR3 RAM Explained
Das Capitolin sends us to Benchmark Reviews for an in-depth feature on DDR3 memory that begins: "These are uncertain financial times we live in today, and the rise and fall of our economy has had [a] direct [effect] on consumer spending. It has already been one full year now that DDR3 has been patiently waiting for the enthusiast community to give it proper consideration, yet [its] success is still undermined by misconceptions and high price. Benchmark Reviews has been testing DDR3 more actively than anyone. ... Sadly, it might take an article like this to open the eyes of my fellow hardware enthusiast[s] and overclocker[s], because it seems like DDR3 is the technology nobody wants [badly] enough to learn about. Pity, because overclocking is what it's all about."
I'm skipping it. Heck, I'll skip DD4, too. Tell me when DDR5 is ready.
This way, mine will always be faster than yours, because it goes to 11.
I want to delete my account but Slashdot doesn't allow it.
I love how overclockers spend $$$ on specialized fans and water cooling systems, just to squeeze some extra performance from their less expensive processors. I've often wondered if they took the money they had to spend on cooling and put it into buying a faster processor if they would come out ahead on price/performance.
I guess its one of those things that people do for fun, for the challenge, just to see if they can do it.
Great, now that I know how well certain DDR3 chips overclock, can I see how DDR3 compares to DDR2 in terms of raw performance and overclocking ability? I'm not an expert on how DDR2 works, so DDR3 could be better explained to me by showing me how it is better than DDR2.
This article is ridiculous. Here's a question for anyone who's passed ohhhh about the 6th grade. What is the one and only thing that has ever helped companies servive a depression? Well besides military contracts, inventions and new technology! If they wanted to make money while the economy is bad, the last thing they should ever do is sit on their current technology. People don't buy more of what they already have if they don't need to. They want something far better! So logically there must be another reason for it not catching on immediately and this article is a load of BS and incorrect speculation
Google's Super Secret Search Algorithm: SELECT @search_results FROM internet WHERE @search_results = 'good'
DDR3 is still 5-10 times the cost of DDR2, and the performance gain is not big (maybe 10% at best) on overall system performance.
Lawyers, MBA's, RIAA? A jedi fears not these things!
Though DDR3 is nice and all with it being much faster, computer speed hits quite a few bottlenecks outside of the RAM speed.
1) Hard disk speed, until SSDs become very common this is one of the causes of decreased speed because a HD can only run so fast
2) The OS. Vista is much much slower then other versions of Windows and as it is the main OS (For now) the fact that it struggles on a 1.6 GHZ dual-core CPU and 1 GB of RAM, only begins to tell you the sluggishness of the OS. And until MS fails and Linux becomes the top OS or MS manages to create a decently fast OS, this will be a problem
3) Connection speed. Its becoming where the Internet is nearly as important as the computer itself. And if you are still on dial-up (and in many places in the US thats all the connection choices offered) even a supercomputer will struggle with sites such as YouTube.
Until those problems get fixed, faster RAM won't make a bit of difference to the end-user.
Taxation is legalized theft, no more, no less.
WTF, why is everything on slashdot tagged as a hardhack? Maybe all that is left are the general equivalent to script-kiddies, waiting for someone two write an .msi they can run and it will do it for them. What has happened to the hacker mentality? I'm not talking about the illegal hacker, but the real hacker who can figure things out on their own? I have seen so many basic things labelled as a hard hack, I guess this was just the one that pushed me over the edge!
Memory bandwidth is not problem right now. I/O is too slow. We need faster disks and LANs first.
Huh. I thought it was all about the Pentium's, at least according to Weird Al.
Someone explain to me why more RAM in Dance Dance Revolution version 3 machines is such a big deal?
Or blinking lights, or loud fans, or whatever my own omniscient opinion given to you using the tone of absolute authority is what it's really about. It's a pity you took that tone.
At lest teh edturs r doin there job !!
An for wanabees. Real enthusiasts get hardware that performs without the risk of data corruption and untimely death. DDR2 is perfectly fine at the moment, especially as the impact of faster RAM is pretty small for most applications. Say up to 10% of the raw speed gain, if you are very, very lucky. That is not worth paying anything more for it, except in servers.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
I always thought the motivation for overclocking was that you pay less for more. You buy the cheaper part and try to get more bang from your buck. I admit I haven't been keeping up with the latest hardware trends, but the summary seems to suggest overclockers should pay more for their parts, which kind of defeats the purpose.
... really careless[ly] written submission[s] that the editors [have] to fix [up] before they can be [read] by us[?]
Generally, bash is superior to python in those environments where python is not installed.
It's always nice to see a tech writer full of teenage enthusiasm, but this article goes a little over the top.
It's supposed to be an article about a performance enhancement, and there's barely any performance values at all (except for the theoretical peak throughputs on page 3, and we know how much that means).
I think what the guy really wanted was to write about planes, not about computer hardware.
I was asked to purchase 4 of the fastest desktop PC's I could find (just less than a year ago) for a contract placement; I ended up going with the QX6850 and for 3 machines DDR2-800, and with the extra money on the 4th machine, going with a capable motherboard and DDR3-1333; hoping for at least some speed difference.
First of all, as the technology was brand new at the time, the motherboard, although capable of 1333mhz ram, it only detected it as 1066 (we double checked they sold us the right stuff), so we manually set the RAM in the BIOS to run at 1333.
After all the setup, on otherwise almost identically configured machines, we found absolutely 0 performance gain on the DDR3 machine over the other 3 DDR2-800 machines. Although one might argue that our applications we were using to test were not so memory intensive, the fact is it was a computationally intensive task that regularly accessed about 200-300mb of data from ram. I would think that even if everything would be pre-fetched into the 8MB CPU cache before it was used, we should at least see some small difference.
In the end, it seems that we spent an extra $800 for no noticeable performance gain.
Tepples only has 251 more posts to hit the big 18k total comments. Cheer him on!
The problem with sales of DDR3 is probably down to the fact that most computers are being sold with versions of Vista 32. To really make a difference with faster SRAM would probably also mean having a larger amount of SRAM. But the problem is Vista 32 only addresses just under 4GB of SRAM. As many computers for the home market max out by being supplied with 4GB of SRAM as new, the computers are effectively un-upgradable unless a different operating system is used. This leaves an interesting situation:- 1. Run Vista 64 - but then a large portion of hardware and software will not work as you cannot sign unapproved drivers. 2. Run Linux 3. Buy a MAC And there is the problem, the alternatives to using DDR3 are in fact called using a different operating system and adding standard DDR2 memory, which is more cost effective that paying the extra for the DDR3 SRAM. The only area I could possibly see DDR3 technology in the commercial hardware is in cutting edge graphics cards, where performance is everything. Unfortunately, as Vista 32 is already at is technology limit and that many potential users of this technology will baulk at the price, I can see games console manufacturers adopting this technology before the general PC market.
Wake me up when it is affordable and widely supported. Those nifty new phenom mobos I was looking at for example don't support it.
I don't understand why your average person would need more than 4GB of ram for a desktop computer anyways. What could you possibly run that would take up that much ram? Are you going to be using Pro Tools and Photoshop and Final Cut Pro all at once, as well as ripping a DVD and playing Duke Nukem Forever? Then maybe you'll need 8GB of ram, but other than that I don't see how it's necessary and I don't see any major advantage to DDR3.
The main limitation these days is not ram speed or size, or even processors, as the quad cores offer more than enough horsepower for most people, even professionals. These hard drives are what's slowing us down. Once we can switch over to a flash drive that can load data much faster than they can today and are relatively cheap, we'll be set.
As U/I=R => I = U / R and U * I = P, we get that U * U / R = P. The power consumption is NOT linear with the voltage but square. So if you lower U by 17% then P gets lower by 21%.
I've rarely seen a more confused article than that one. He clearly seemed like he grabbed buzz-phrases out of fliers and pasted them into an article.
I run large electromagnetic finite element problems on a machine with 16GB of DDR2 with two Xeon Dual Cores at 2.66 GHz on XP64. One job takes about four days. I can run two in parallel before memory gives out. If my firm had $20K available I'd get a machine with 64GB of DDR3 at "1600 MHz" and dual quad-core at 3.2GHz. I could run larger jobs or more in parallel and they might take only three days - two iterations per week.
He's like a jet engine supporter immediately after World War II:
"Jet engines are inherrently capable of much greater speeds than propeller engines!"
"OK, so show me one that goes much faster than the prop driven spitfire?"
"Well, uh, there's the Gloster Meteor!"
"It does about 500mph*, right? That's not bad compared to the Spitfire XIX's 460mph. But there are tons of Spitfires out there, available cheaply vs. paying several times the cost for the Meteor for about a 10% speed improvement." *note: F-3 variant, not the "overclocked" tweaked versions that set speed records.
"Well, but the point isn't that it's better today. It's a better technology! It'll be better in the future! Props will never go supersonic. Jets can potentially go several times supersonic."
"That is cool. Doesn't really help me today though, does it? I'm still paying several times the cost for a small improvement, today."
"Yeah, but if you don't buy jets now, how will their prices ever come down? How will we ever reach the heady perfection they're capable of?"
"Again, not helping me today, is it?"
"But! But! It's really cool!"
Yes, the technology shows promise. But its future promise with only small increases today doesn't justify its current high cost.
If more people bought it, the cost would come down over time and more investment would mean unlocking more of that promise. Which is great in the future but gives you very little today in exchange for that high cost.
The argument he seems to be making is that everyone should adopt it right now, not because it actually gets them much for their money but because their investment will enable him to buy even faster stuff for a lower price later.
Not really compelling.
Tell me I'm not the only person who saw that and thought Dance Dance Revolution had some kind of new RAM feature. Go on, tell me!
But does it run Linux?
I haven't bought in to DDR3 yet for one simple reason...I was an early adopter of RDRAM, and look where that went.
I want a new quote. One that won't spill. One that don't cost too much. Or come in a pill.
"One particularly important new change introduced with DDR3 is in the improved prefetch buffer: up from DDR2's four bits to an astounding eight bits per cycle. This translates to a full 100% increase in the prefetch payload" *If I remember correctly*, the reason the "prefetch" is increased is because the internal clock speed is still slow.. eg. dram running at 1000mhz internal/external with 4bit prefetch would have the same bandiwdth as dram running with 500mhz/1000mhz internal/external and 8bit prefetch. because it easier to make the internal clock run slower but have more "prefetch" and let the bus run at high speed. it's hard to get a chip to run at 1.6ghz but it's easy to have the chip run 400mhz with a 1.6ghz bus. this would add latency because all internal calculations are going the same speed, but the data transfered is now higher. bus speed needs to increase in mhz or width to make it go faster. trying to add width sux because you now need more traces on the mobo, and no one needs a 500pin ddr3. more than likely, ram will keep increasing in "mhz" of the fsb to keep up with bandwidth needs, but latency will keep increasing as "prefetch" is constantly increased to keep internal bandwidth symetrical with the fsb. this author doesn't talk about anything like this and i think he gives a very biased view on ddr3 vs ddr2. I do think power efficiency is a good point he makes though.
Well, it was bound to happen sooner or later.
Aide-toi, le Ciel t'aidera - Jeanne D'Arc.
I really like ECC memory. I've built a lot of machines here and there for various purposes and people and can safely say that ECC should be mandated. People oft talk of blue screens, panics, etc. A lot of these are due to bad memory. And these types of errors can often be hard to replicate. I believe EETimes had an article about how bad bad memory is.
Anyhow, I'm on a 975X chipset with 4GB of DDR2 800 MHZ unbuffered ECC memory machine now. Not a single unforced error since I bought this machine and assembled it December, 2006. Nothing. This unit is primarily a gaming rig, the 3DMark 2006 score is 11500 with an 8800GTX, all with ECC memory.
The most irritating thing for me is, looking at the great new CPUs available, is the utter lack of any unbuffered ECC memory in the DDR3 range. This to me is unacceptable. I refuse to compromise so I will wait. Intel has a motherboard featuring DDR2 800 fully buffered memory for the 'high end workstation' , D5400XS, this is $600+. Supermicro offers something similar.
The X38 does DDR2 ECC, and for whatever reason the X48 took that away. I don't get why Intel wants to deny us DDR3 unbuffered ECC? Its really a selling point, a very good thing. If you overclock, its nice to have because it can tell you the limits minus the guesswork, not that I would bother with OC personally.
Fundamentally, without ECC, do you even know if the memory works at all? My experience leads me to believe that without ECC present, the answer is no at all.
Legalize the constitution. Think for yourself question authority.
This article seems to be paid advertising, with some bending of the facts.
And the fact is that the double prefetch buffer is the sole reason for the double bandwith and the double latency. As a matter of fact the speed of the individual memory chips on the ram module are the same as on ddr2 (see that table in the article, just divide the ddr3 clock by 2 to get the corresponding ddr2 speed) - but instead of reading 1 bit from 4 chips at once into the prefetch buffer (that is four bit prefetch buffer), they are reading from eight at once (so that's the 8 bit prefetch buffer), so double the amount of data can be read in the same time (hence the double bandwidth). However because the memory chips are the same speed as in ddr2, the time needed to program the individual chips stays the same - so because the clock is double the speed, it takes twice as many clocks to tell the individual chips which bit we want to fetch. And that bullshit about lower latencies is also not quite right: ddr3-1600 cl6 is the same latency as ddr2-800 cl3 - and such modules have been sold for years.
Of course ddr3 is better, because it has higher bandwidth, and absolute latency is not worse than ddr2's. Also there are in deed technological advances (e.g. the lower voltage). But this article is still not exactly honest.
I run memtest for a minimum of 48 hours on any new system I build and have never had any problem with RAM that has passed that. This is the best I can do without the premium of paying for ECC capable motherboards and RAM.
And conversely; a reasonable number fail. Working as a system builder and memtesting every machine I build, I've seen around 15% failure rate in all brands of RAM.
[clever sig]
Did the customers actually want vista, or was it forced upon them by their choice of computer?
A better metric would be to see how many copies of vista sell with XP sitting next to it, both available equally.
Otherwise, it's like saying that 99.999% of people prefer to breathe oxygen - like there's a choice.
Build a good load balanced web cluster some time. It's easy to saturate a GbE connection to the back-end storage boxes.
Yet somehow millions of non-ECC machines give the impression that they're working on a daily basis.
Hey, look! It's Bono's brother.
I have yet to have a customer offer a contract for a full ECC system, but I expect that type of customer to be ordering in the dozens or hundreds. For one machine or 100 I would still give 48 hours of memtest. For ECC I would still give 48 hours of memtest.
If you can recommend a better test, I am open to suggestions.
DDR3 is worthless anyways. I've only seen about a 5% or so increase in performance in various tests (Gaming, video editing, loading huge photographs,) not much to justify the extra 50% increase in price.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
Failures at build never see the customer. One would hope that hardware failing build has at least enough warranty to recover failed parts from supplier/manufacturer. If not, how can you provide any warranty to your customer?
I've seen ECC errors on server logs quite often. Then you replace the memory and its gone.
The thing with ECC errors you would never know unless the error causes a crash if it happened.
You are misguided in your thinking, *severely*
You could have errors all the time, you could have silent corruption, you name it, and the memory "works"
There is nothing besides "getting lucky" and the corruptions causing a panic that is between you and your computer basically being worthless in terms of reliability.
http://www.eetimes.com/news/latest/showArticle.jhtml?articleID=199601761
Impression.
So you are lobbying to have sucky memory which only saves you 1/9th the cost ? (Garbage memory generally has 8 chips, real memory has 9)
Are you a cheapskate or do you just like to expose the entire computing world to single bit errors because you "know better."
Ever stop to think that now with 4GB configs being common that the rate of single bit errors might have gone up considerably since you "experience" began?
No one said ECC means dont test the memory, what people "in the know" know is that your little memtest86 run means jack split about long term memory viability/reliability.
Theres a whole topic on XS speak freely about this post, half of it was plagarised from wikipedia, and then copyrighted for the authors own use, not to mention the images are stolen from other sites and then stamped with the benchmark reviews bookmark. The owner of that site should be ashamed of themselves
When Nehalem comes out, with its new socket, motherboard manufacturers are likely going to mandate the switch to DDR3. Right now, it's pretty easy for motherboard manufacturers to push hybrid solutions or redirect consumers to DDR2 1066MHz (which is equally useless unless you're OCing something like a Wolfdale). They won't have that excuse in the new motherboards and I seriously doubt they will stick with DDR2, especially if prices come down by then.
Wow, the first thing I thought of was that it would be the most useful degree you could choose nowadays, and that even an undergraduate degree would immediately pay for itself in corporate America.
Failures at build never see the customer. One would hope that hardware failing build has at least enough warranty to recover failed parts from supplier/manufacturer. If not, how can you provide any warranty to your customer?
Yeah, that's true. I wasn't arguing with your comment, merely commenting that a reasonable number of modules fail when first tested. Not maintaining machines I've had no experience to comment on ongoing failures.[clever sig]
Running memtest is of course a good idea, but with ECC you can (assuming you have software support) know about errors during normal operation. I've had two on my desktop in the last 30 days. (975X chipset, 4GB DDR2 667MHz memory.) (This is infrequent enough that it would usually pass memtest for your 48 hours (and has). (Memtest stresses the memory more, but still.))
When I was building my latest system, several months back, I initially wanted to go with DDR3. Money wasn't an inhibiting factor for me. But, I ended up going with DDR2 anyway. Here's why:
I couldn't find DDR3 memory confirmed to work with the motherboard I wanted!
There are so many stories in the reviews sections of NewEgg and other sites with people complaining that boards like the Asus Maximus fail to work with many and various brands of DDR3 memory.
The only brands I could confirm to work, were "out of stock" at every online shop I could find (apparently everyone was doing the same research and wanted their systems to actually work).
Short suppy, high price, spotty compatibility. This led me to decide to just build a system with double the memory @1066, instead of going for 1600 or 1800. Hopefully DDR3 will mature before I want to build another.
I run memtest for a minimum of 48 hours on any new system I build and have never had any problem with RAM that has passed that. This is the best I can do without the premium of paying for ECC capable motherboards and RAM.
Memtest86 (and Memtest86+) do not do a good job of finding RAM that is borderline working. They are only able to find RAM that is definitely not working. By "borderline", I'm talking about systems that fail under moderate to heavy load in random fashion (usually bluescreens or random application crashes). Or instances where things like QuickPar will fail to create or verify PAR2 sets.
By far, the best way that I've found over the years to truly stress test a system and verify that it will work correctly is a combination of Prime95 in torture-test mode and running some sort of heavy disk activity test. Prime95 exercises *both* the CPU and RAM at the same time, which is an excellent way of uncovering timing issues (i.e. CAS 2.5 RAM sold as CAS 2.0).
If your system can survive Prime95 + running the disks non-stop + running the video-card non-stop for 72 hours, then you are mostly home free.
Wolde you bothe eate your cake, and have your cake?
"...overclocking is what it's all about..."
While I understand it might be your particular hobby, it's not the case for most others.
I used to upgrade my computers every 3 years, which would roughly give me a 3x improvement in raw processing power (not to mention video, etc.).
IBM AT, 486-33DX, P-90, (AMD)P300, (AMD)P1.2GHz, P4 2.7 GHz.
That last one is something like 4-5 years old now (I forget) and it's only reaching the end of it's value as a cutting-edge game machine. I haven't found a lot of persuasive evidence that I need to upgrade, except now that Age of Conan is coming out, as well as some shooters I'm looking forward to.
So no, I don't see much of a compelling argument to upgrade hardware recently at all, much less push the envelope trying to overclock for that 4% performance gain.
-Styopa
SRAM = STATIC Random Access Memory SDRAM = Synchronous DYNAMIC Random Access Memory DRAM is basically a capactior, SRAM is a latch. Two different memory technologies. I believe a CPU cache is typically SRAM.
I've used ECC and I know the reasoning for it along with the technical innard-workings of it. I've seen it in operation for years and years... the problem is due to the price of ECC compatible boards they don't come standard in the main range of "servers", unless they're good. Hard press... but it's not near the point of saying ECC or nothing, unless you're going to outwardly pony up the dough for it. Kind of like only getting a car oil changed with Amsoil filters & synthetic oil every time you get an oil change done at a shop. Will be quite a bit more money comparedly, but why should you use any other kind of oil and oil filter in your car since the oil and filter are proven to far surpass any synthetic & hybrid synthetic/dinosaur-based. In fact I think all oil should be blocked and Amsoil be mandatory because it's functionally better than cheap 10w/30. (mind the sarcasm of the comparison, even if Amsoil is much better.. it's consumer choice)
I still don't understand how this explains Dance Dance Revolution 3...
As so many people have said, the price premium over ddr ram and the lack of a performance enhancement makes DDR3 not be appealing to most people. I can enhance my computer much more cheaply by stuffing more ram in it, getting a faster processor, or getting a faster hard drive. I might even spring for a better monitor before I spend the extra money on a 5% performance increase. And in a lot of cases, because of the slower latency, DDR3 can be slower than DDR2 ram.
It's amazing. Power = Current * Voltage. Lower voltage with proportionately higher current = equal power consumption.
When this "Eye opening, educational explanation" starts with such gibberish. If the only change in the memory was to shorten the distance between high and low, and all other issues such as resistance (or inductance since it practically an AC circuit) remained the same, then the power consumption would obviously be lower. But since the technology does in fact increase specifications of nearly everything, I'd imagine that the current drain would generally be at least proportionally higher.
Also it just seems more than a little childish for the author to focus so heavily on overclocking and then talk about power in terms of savings as opposed to focussing on the fact that lower voltage means shorter distances between gate threshold levels, therefore allowing higher performance of gate switching over equal capacitance circuits.
LAME!!!!
Also check out the specification update for the X38 chipset page 8.