Slashdot Mirror
Memory Manufacturers Could be Cheating
Mark Brown writes "Tom's Hardware is live-testing DDR2 memory products in order to determine whether memory manufacturers submit cherry-picked products for reviews. 'GeIL DDR2-667 that was claimed to be purchased performed worse than the review samples they got: 471 MHz for the review samples vs. 421 MHz for the retail memory.'"
Oh, my, what ever will we do? Maybe the memory manufacturers should divorce and marry other companies.
In other words, set their affairs in proper order...
I never trust specs on any product I buy. For example, if I buy a hard drive the first thing I do is open it up, shake all the bits out it and count them. If they don't add up to exactly what is listed in the spec, I return it.
Oh dear lord, a company wants to make sure their product gets the best review possible and tests it before they send it.
I'm shocked!
... Job seekers have been putting ONLY their best accomplishments on their resumes
... Advertisers are STAGING their product photo shoots
... etc
No way, there can't be anyone making dishonest or cheap mem... PAGE_FAULT_IN_NONPAGED_AREA
"is 50 MHz really that large a discrepancy?"
Well it is more than 10%, that's pretty big.
There's a reason why Consumer Reports buys all of their products for testing through normal retail outlets.
Obviously companies will test any unit they send out to be reviewed to make sure it works as well as it can. The question is, how many other units did they test? If they only went out and bought one other unit, and the discrepancy was that large, it could be that the unit they bought was defective. They would need to buy several units from several retailers, preferrably in geographically dispersed areas, to get a real feel for how well these things will perform on average.
In order to evaluate this claim we need to know about the reliability of the test. What is the variance if the test is repeated many times on the same RAM? Without this piece of information we don't know if 50 MHz is a small or large difference, or if even if it is a real one.
Not surprising at all. Manufacturers do want to get the most positive reviews after all. Look at the hardware sites. Only rarely do you ever see reviews of value equipment rather than the latest top-end equipment. I do find it a bit of a shame though. 50 Mhz may not look like much, but for the enthusiasts that go after the best equipment regardless of price, or the person who needs the best possible specs for a system and is willing to pay, that 50 Mhz can mean the difference between a purchase and a pass-by. While I don't think Geil's doing anything different than any other manufacturer is likely to try, it does make me think twice about the claimed specs of their products.
Disclaimer: I didn't RTFA, and this is what I personally know for experience. correct me if i'm wrong.
Honesty may be the best policy, but by process of elimination, dishonesty is the second best policy.
I wonder if chips selected for reviews are overclocked first (just a bit), knowing full-well that it'll last long enough to go through the review process and the warantees wont be expensive to honor on just a small percentage of product.
Latewire
Memory is rated to perform within certain specifications. If it doesn't perform well within this range, that's a legitimate complaint.
Tom's is complaining about something totally different. They are seeing how well the memory will overclock. But the manufacturer makes no claims about how well it will overclock. They explicitly tell you that they cannot guarantee what will happen. This is a reasonable position on their part.
But what Tom's is asking is for all memory from a given manufacturer to overclock the same. This is crazy. The manufacturer has every right to switch production methods and to make other changes which could affect overclocking performance. The only question should be: does the memory perfom as specified.
If you overclock your memory and it works well, good for you. But you have no right to complain if overclocking doesn't work as well as you want!
Remember, the choke point in a running program is usually the memory. Once it's been read off the hard disk (which is a startup cost, but doesn't matter much after the first quarter second or two), and as long it isn't doing a lot of I/O, performance is highly tied to memory.
For the top of the line CPUs, if your memory isn't fast enough, you've wasted your money.
So with that in mind, I'd say an ~10% drop in performance is significant.
Most things seem crazy when analyzed as a percentage...must be a learned behavior....it just seems that in the "real world" this wouldn't matter a whole lot. I don't think even a gamer would notice unless s/he was running the benchmarks, in which case all that will happen is the numbers get used as a bragging figure in a forum somewhere....(yes I game, but I am relatively unconcerned with DDR2 as this point as I have RAMBUS in my system...maybe in my next purchase?)
Why go fast when you can go anywhere? O|||||||O
1) The article says that they bump the clock rate until the systems crash. ... just to carry on running with bad data.
... one for the review sample, and one for the retail purchased.
I'd be a little happier with running a memory test and running at progressively faster speeds until it detects an error. Some memory errors might not cause the system to crash
2) They have two "identical" systems
How do they know that all the components in the identical systems really have exactly the same specs? It would be more fair use just one system, or after the tests complete to swap the ram and re-run.
Taking a sample size of 1, not really. Their test leaves something to be desired. They really ought to be testing both memories in both systems, several times before jumping to conclusions. Slight variations in PCBs and silicon can build up to cause appreciable differences. Ultimately overclocking is taking entire designs well outside their specified operating limits. To do this reliably you need to test thoroughly on many samples.
The part of it that convinced me that they're right anyhow is the memory supply voltages. "Normal" on the cherry picked Gigabyte board was ~2.2V, normal on the storebought was ~1.83V (FVI 1.8V is the DDR2 spec supply voltage). You'll have to take my word for it, but THAT variation is huge. People who build computers do not tolerate voltage discrepancies like that, it's out of spec for the devices which usually allow 5% variation (1.71V-1.89V). You can verify this by going to Hynix/Micron/Infineon and pulling down a DDR2 component datasheet.
The headline is beyond wrong though, it's probably actually criminal. GeIL does not control the memory supply voltage (they make the DIMM), Gigabyte does (they make the mobo). GIGABYTE is cheating.
It's very easy to figure out if memory makers are cheating: take the heatsink off, look at the device part numbers and look them up. There's not a whole lot to tweak that doesn't involve a complete redesign of the DIMM. If they cheat it's almost always because they used a DDR2-400 device but branded their DIMM as DDR2-something_higher.
The ram was rated as DDR2 667 even the retail at 421 MHZ. That comes out to DDR2-842 doesn't it?
The ram met and far exceeded it's rated clock speed. Sure the give good stuff to reviewers. If the review sites want to do valid tests of which brand of ram is the best for over clocking they would have to purchase multiple samples of each brand from the retail channel.
When overclocking the truth is your results may very. If you are pushing past specs then some will work and some will not. Heck even different production batches will give different averages.
See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
Horsepower in cars rarely meets up with the numbers. Fuel efficiency, either. Carb content in food is labeled, but most people don't read the serving size, so that is advertising funk, too.
Why should this be different? When a company ships a product to be reviewed and tested, they'll ship the best. When they test their own, they'll test the best. You should NEVER accept that specs are factual, and you should spend some time confirming what you bought.
This is the great thing about specs -- if they're lies, just return the product. If a company lies enough, the customers will go elsewhere.
It is really all common sense.
http://www.hardocp.com/reviews.html?cat=MjUsRGVza3 RvcCBDb21wdXRlcnMsaGNvbnN1bWVyLCws
What they are doing is having other people buying systems and then reviewing those systems. They will only review systems where they have an agreement with the manufacturer that the computer can be returned at the end of the review. The key is that the manufacturer never knows who is getting a system which may be subject to review.
It actually works well for both parties. Some manufacturers are proactive in the forums and even acted on complaints received, strengthing their processes.
* Winners compare their achievements to their goals, losers compare theirs to that of others.
Hmmm... If I read this right, it looks like the motherboard that came w/ the memory had its voltage increased to induce higher speeds. This would skew any test - not just overclocking, unless you knew to reset it.
Well, the article says the RAM is DDR2-667 which (I'm pretty sure) implies a clock rate of 333MHz (somebody correct me if I wrong). So gamers are still getting more than they paid for...
What I want to know is where do Tom's Hardware get off thinking this is statiscally significant? Basically their saying "We took one part from the suppliers, and one part from retail sources. The retail parts performed worse. OH MY GOD, that must mean they're cheating!!!" Compare numerous examples from each source and then I'll be more easily swayed to their argument...
To me, the increased voltage on Gigabyte's motherboard is far more interesting...
TWW
"Encyclopedia" is to "Wikipedia" what "Library" is to "Some people at a bus stop"
That's news...so I went back and actually read the WHOLE article...that board voltage is a huge difference, no wonder there was a performance hit...with that in mind, what is Tom's actually testing anyway....so much for identical systems
Why go fast when you can go anywhere? O|||||||O
Correct me if I'm wrong, but isn't DDR2-667 only certified to run at 333MHz? Either way, 471 MHz and 421 MHz are both well above that... It's not as if they're claiming it runs at 471 and it actually runs at 421... they're only guaranteeing it to run at 333... right?
From TFA:
"Its DDR2-667 memory......"
"maximum clock speed of 471 MHz, which corresponds to DDR2-942"
vs
"a memory clock of 421 MHz (DDR2-842)"
So its more than 20% faster than what it is rated at... Whats the big deal? Everyone knows there are certain processors/memory modules from the same exact part# that outperform others. This has been the case since before the Celeron 300a even. If the memory performed below its rating, then there would be a problem
If all of the manufacturers cheat, then none of them are cheating.
Recall the hubub from as recently as a half-decade ago, when video card manufacturers were rigging their drivers (or the cards themselves) to recognize when they were being asked to draw the same patterns over and over again (like, say, 10,000 colored boxes, or circles... like benchmark programs do) and would silently decide to perform only a fraction of them to jack the benchmark numbers up?
Never, ever trust the results from an item that the company sent you when they knew you were a reviewer. You should just go out and buy one off the shelf in a store. If you can't afford to do that, buy one from a store and ask the company for a review sample, return the sample to the store and test the, now free, one that you got "in the wild", as it were.
Hardly.
A DDR-667 chip (or more specifically, a PC2-5300 stick) is supposed to run at at 333 MHz. So one runs at 421 MHz and the other runs at 471 MHz. To me, it looks like both of those sticks are performing way faster than the specification requires.
Isn't this just the price the user pays for being too stingy to pay for a memory stick which is actually rated to run at 400 MHz in the first place?
Karma: It's all a bunch of tree-huggin' hippy crap!
Corporations are ripping off its customers with rigged tests... I'm truly shocked.
They aren't necessarily rigging anything -- chip production runs always produce a range of qualities, and they're submitting the best they have. To not do so, especially when everyone else does, would be to sabotage your own reviews. There are no "unbiased" samples.
The only practical way to fix this is to establish a standard for what companies should send in -- preferably something like five to ten random chips that have passed basic testing.
It's tragic. Laugh.
how is this different than the automotive market, where manufacturers routinely send wringer cars to magizenes to test. I think Car & Driver actually did a teardown on one, that had a Formula 1 quality engine :) I guess computer manufacturers are tryin the same thin
-TubaMan / ThE_DoOmSmItH
it is if its below the advertised speed. If they are just giving samples that overclock more, then I don't see how anyone can complain about that. Thats why people use random samples for statistics. But then you would have to pay for them wouldn't you.
Unprofessional benchmarks and overclocking?
If Tom's Hardware has a problem with this perhaps they should stick to real-world benchmarks and purchase all the equipment they test for review, instead of trusting manufacturers to "help" them. Its very unprofessional of them to work so closely with the businesses they are supposed to be reviewing..
Let's see - the GeIL memory is rated at DDR2-533. The module from the vendor ran at DDR2-942. The module from the store ran at DDR2-842. Now, Tom makes this out to be some big controversy, but it seems to me that a module running 36% faster than specified is no small thing, particularly at that high of a data rate.
I'm an engineer who designs memory modules. In most cases, our modules are overclockable, at least to some degree - some go faster than others. At the sort of speed that Tom's Hardware is running, I'm not really surprised that there's more than a 2 or 3% variation in performance, espeically if the chips on those modules came from different manufacturing lots. At the outer limits of memory speed performance, the tiniest changes in parasitic capacitance can be death to performance - and those values change from lot to lot, even from wafer to wafer.
When manufacturers specify that 2% to 3% tolerance, they're referring to the module's performance at its rated speed, and that makes sense. Plug two modules into a system and they will run in virtual lockstep - at their rated speed. There are a million analogies that I could use, but the bottom line is that there are assumptions and statements in Tom's article that just aren't right.
Maybe the module was cherry-picked and maybe it wasn't, but, if nothing else, a sample of two doesn't make for much of a study. After all, if the retail module had been DOA, a pedantic person could say that GeIL cherry-picked the evaluation samples and sends all the defective modules to retail.
-h-
Get the writer loaded and laid.
Seriously. Many years ago, I worked as a technician for a (now defunct) major audio equipment manufacturer. When a writer from "Stereo Review" or "Audio" magazine came to visit, we'd play with the equipment a little, my Engineering boss would hand him some specs, and they'd go out on the town (leaving me to work the rest of the day {grumble, grumble}). A few months later, we'd see those exact specs printed in the magazine, along with some well-placed ads. I never believe a review I read in a trade publication.
Consumer Reports lacks technical expertise in many areas, but at least their approach has some level of integrity.
Research shows that 67% of those who use the term "research shows", are just making shit up.
Actually, it's innate human nature to think of things that way. Put a one pound weight in one hand, and a two pound weight in the other - virtually everybody will be able to tell the difference between the two. Now put a forty pound weight in one hand, and a forty-one pound weight in the other - very few people will be able to tell the difference, despite the fact that it's a difference of one pound in both cases.
The reason we perceive the two cases differently is that, in the first case, "B" is twice as heavy as "A", whereas in the second case, "B" is only 2.5% heavier than "A". Or if you don't have heavy objects handy, get a three-way lightbulb and a lamp to match. Notice how the jump from 50 to 100 watts seems like a bigger jump in brightness than the jump from 100 watts to 150 watts. That's because, in percentage terms, it is a bigger jump. It's how we're wired to see the world, in terms of percentage differences.
ABSURDITY, n.: A statement or belief manifestly inconsistent with one's own opinion.
See what I mean, it didn't weigh 1024g like it should have had it been measured computer standards... Show me who you are you Coward! Kneel before ZOD!
So because other people in other industries do it, that makes it okay? (In either industry, for that matter?)
Random and weird software I've written.
The major difference is that these hardware sites are running the product through quantitative benchmarks to compare products. This throws doubt onto that whole entire notion of comparison.
If you could say.. foster that doubt sufficiently, you might be able to make a business out of buying and benchmarking hardware, hand picking the good stuff and selling it at a boosted price as "guaranteed best."
Then throw a "credited rating system" around it, and you could potentially have a nice little middleman racket like what card shops sort of have.
Smart people please tell me how many bits in a byte, how many bytes in a Kb, how many Kb in a Mb
Ok, there are eight bits in a byte, one thousand twenty-four bytes in a Kb and one thousand twenty-four Kb in a Mb. Glad to be of service.
Oh, and I'm terribly bad at reading sarcasm in written communications. So in case you ever decide to use sarcasm, you might want to include some sarcasm tags.
0*0
00*
***
In computers, specs are very, very important. With RAM, the spec tells you the maximum frequency at which it is rated to run. So if the memory is DDR2 667, it is rated to operate at a maximum frequency of 333MHz (DDR values are doubled). You can try running it faster, it may work (that's what Tom was doing) but no gaurentees. However it is gaurenteed to operate properly, withoug stability problems at 333MHz or below.
Thus, if your system requires DDR2 667, you need to make sure you buy it, otherwise your system may crash, corrupt data, or simply fail to POST.
Computer specs are generally like engineering data: They are maximum safe ratings. The company gaurentees that the product will work up to and including this level, but not more. It may go higher, but you do so at your own risk.
As for taking the heat spreader off and looking at the PN's on the memory chips, there's no reason GeIL can't brand a 'DDR2-400' chip as DDR2-533 if it runs stable at that speed. Ditto for branding a 'DDR2-667' capable chip as DDR2-533.
[Fuck Beta]
o0t!
Tuesday, April 10th? Wednesday, April 11th?
So if you want the best stuff, convince them you're a review site and just wait for them to ship you the cream de la creme.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
i work at hp, and i'm sure this is standard practice across every industry... all review units go through a series of stringent screening process to determine the absolutely best units.
seriously think about it.... if you had a hot date, would you show up in a yellow wife-beater, messy hair and bad breathe, and ask her to pay the cab that's been waiting for them for the last 30 minutes?
start naming programs that don't do alot of I/O tasks. see any games in that list? i'll take 1GB pc100 over 512 DDR2 any day of the week.
"Prediction: within 10 years, Windows will be a Linux distribution." Me, 7-6-2016
So you got me on the labelling, GeIL does relabel their memory components. That right there is an indicator they're probably dishonest, but not proof. I think there are 5 DRAM mfg's (in order of my experienced trustworthiness): Micron, Infineon, Samsung, Hynix and one other I can't remember. Anyhow there's nothing secret about DRAM, so the only purpose in relabelling is to misinform. Not a problem on the systems I work on lately, but I can see why there is a deserved distrust.
But the rest you're wrong, or at least not proveably right. Page 10 shows that there was a measured discrepancy between the "normal" setting on the Gigabyte provided mobo vs. the storebought one. They set both to "Normal" (which implies 1.8V, but is different from MEASURING 1.8V). It turns out the "normal" setting on one was not 1.8V. The article did not explain when this was discovered, if anything was done to compensate, and if this could have affected the results. What I did notice is GeIL was the only memory actually tested, they hit the panic button before they ran the other results. Experiments with a sample size of 1 are bunk.
Even their clock margining scheme for determining overclockworthiness leaves something to be desired. Silicon speed grade sorting is not precise to any number of decimal points. All it says is a given device fits into a certain speed category. The only GUARANTEE is that a device will run within spec, once you're out of spec, all bets are off. To get an accurate reporting you need to get several devices from several die lots and re-run. The best way to try this is to buy several DIMMs from stores across the US at different times (assuming DIMM vendors who relabel their devices, otherwise you can read the lot code from the chip).
The results about GeIL are inconclusive, but I'd say they have Gigabyte red-handed.
AFAIK, it is fine to say any of the following: 1 KB=1000 bytes, 1 MB = 1000 KB, 1 Kb = 1000 bits, 1MB = 1000 Kb, 1 KiB = 1024 bytes, 1 MiB = 1024 kibibytes, 1 kibibit = 1024 bits, 1 mebibit = 1024 kibibits, and so on...
There are two types of people in the world: those who divide people into two types and those who don't.
You're having problems with the megaHURTS?
She: Hey, are you a traitor? Me: No, I'm atheist.
Actually you're right twice, they did correct the problem, although the motherboards are clearly not identical. Still shame on Gigabyte, they are definitely in the wrong.
Tom's Hardware and other reviewers may not be able to buy their tested items from retailers, but I can think of a great way to get retail items without any cost to them. When they receive a "cherry picked" piece of hardware they can post it on their website and ask for users to register to purchase a matching retail item to trade. The "winning" user can then get a retail part, ship it to the reviewer, and receive the primo hardware in return. This way the reviewer gets to test both parts, and the user has a good chance of getting a hand-picked piece of hardware. Win-win. Just an idea.
Accepting merchandise-even to use-opens one to the bribery effect. Accepting merchandise-even just to "review" -means that yes indeed you can get a cherry picked tweaked pristine example of the article in question.
It is just common sense bad mojo to accept stuff directly from the manufacturers for this sort of work.
Image manipulation usually requires a lot of beefy memory, i.e. the faster, the better.
True confidence comes not from realising you are as good as your peers, but that your peers are as bad as you are.
What about the reviewers? To be accepting free samples (aka bribes) seems asking to get tricked like this. I doubt the review companies/reviewers care, as they kind of like their perks. A little like lobbyists and politicians.
The only magazine I know of that buys their test samples retail is Consumer Reports, and they do it for this reason (as well as to avoid any conflict of interest).
the products behave well within their specifications. IMO overclocking is lots of hype over small differences. what i care for is a product that works well. and 90% (or more) of the customers feel exactly the same way (or are too ignorasnt to understand what overclocking is)
god, i wish sometimes that geeks were a tad bit more pragmatic, and would put themselves in the position of manufacturars (or anyone "regular" for that matter)
(no i don't deal in popular opinions. it's not my style)
Just as a reminder, it was proved by different benchmarks (gaming bench as well) that DDR2 gained around 7% real performance over DDR... You have to remember that memory size is more important than pure speed, along with bus speed and what amout of that memory is held on the video card (and not in system memory). There are soooo many factors, and I personnally find that the extra half fps is not worth upgrading your 500mhz memory to 667... I prefer to spend on video cards.
Of Code And Men
Fuck that. IEEE cave to MARKETING PRESSURES thus adding more confusion to the industry is why I am no longer a member.
The Kruger Dunning explains most post on
It's how we're wired to see the world, in terms of percentage differences.
Yes and no, generally speaking human senses work on a logorithmic scale.
Think about how sound loudness is measured. The decibel scale is logorithmic, as the sound gets louder it requires much more force to produce a result we can physically tell is "louder". Same for the other senses.
The ratio of people to cake is too big
This is a very big problem overall. One of the points of these reviews is to give the consumer a better idea of how a product overclocks (the point of whether or not this is above rated spec is irrelevant nowadays). The memory manufacturers have all caught on to this, which is why some memory manufacturers even give warranties of "you won't void your memory's warranty as long as you don't go above 3.1V," even when you only need
Furthermore, if you look at it from an overclocking perspective, it's not just "50MHz" anymore. The memory's spec is actually 333MHz, so the difference is 421 - 333 = 88MHz versus 471 - 333 = 138MHz, an approximately 57% increase (this is mainly directed at what others posters are calling "10%"). As you can see, when a product is being advertised (since this sample was sent to Tom's for review in the first place) as "enthusiast" or "performance" part, this then is classified as deceptive practice on the memory manufacturers' end. Moreover, from reported user experience (just keep up with the user forums on overclocking at sites like Tom's, Anand's, or [H]ardOCP), the differences between the review and what people buy off the shelves should not be this large.
Argh, didn't notice the parsing problem. Meant to say: "...even when you only need less than 3V to run it at spec."
Erm... your math is dodgy. If one overclocked to 334MHz and the other overclocked to 335MHz, then sure, one overclocked by 100% more than the other, but they're still less than 1% away from each other in raw performance.
--JoeProgram Intellivision!
Credit where its due: http://en.wikipedia.org/wiki/Weber-Fechner_law
Comment removed based on user account deletion
Humans also perceive sound in a similar manner.
When the frequency of a sound is doubled, it's perceived one octave higher. (The frequency of a note f(n) = F 2^(n/12) where n is the note, 0 signifying the A above middle C, 1 signifying A#, 2 signifying B etc. and typically F = 440 Hz, i.e. the "concert pitch", or the frequency of the A above middle C.)
The loudness of a sound is also perceived in a logarithmic manner. The level of a sound must increase with an accelerating rate in order for us to perceive a linear change in loudness. (The difference in sound pressure level is often measured in decibels. The function is 10 log (p_2^2/p_1^2) dB = 20 log (p_2/p_1) dB where p_1 is the reference sound pressure and p_2 is the measured sound pressure.)
Please excuse any mistakes. It's late, i'm very tired and i don't seem to be able to concentrate at all. :-)
Comment removed based on user account deletion
Comment removed based on user account deletion
There is more to this than meets the eye. Most modern processors are set up to run at a fixed multiple of the memory bus speed. If you are trying to squeeze extra performance out of your system by overclocking the cpu, you are also overclocking your memory. A popular strategy for building a system that is fast and cheap is to buy a value processor with a good reputation for overclockability. However, one risk is that you may find that your overclock is severely limited by the speed that the memory will operate reliably at. Consequently, to improve your odds, you might spring for "premium" memory that is reported to also be highly overclockable. In this case, a niche market memory manufacturer can benefit a great deal from some positive reviews on hardware web sites for their supposedly premium memory.
They say AMD stresses other components. It has been my experience for a couple years that if it says PC2700, for example, I better get PC3200 if I don't want lockups.
This would clarify a few things.
Memory should run at a rated speed. Any speed over that only matters if you overclock, I would think.
On a slightly more close to home note, ever notice how time seems to go faster the older you get.
Well, in the same sense that the difference between a pound and two is more significant than forty and forty one pounds, each year of your life is compared in your mind to the whole of your life.
The older you are, the more you have to compare against, so the less significant the last few weeks seem.
The key to always feeling like you have a lot of time is obviously forgetfulness!!!
I wouldn't consider the mad hatter mad. Just reality impaired. He sure can make a mean cup of tea.
So what you're basically saying is, someone is using the product outside of the product's operating specifications, and then bitching because some other guy was able to use it *further* outside of the operating specifications.
I still don't see the problem here, except perhaps the problem that overclockers are a little too enthusiastic about saving those extra few dollars.
Karma: It's all a bunch of tree-huggin' hippy crap!
>>It's how we're wired to see the world, in terms of percentage differences.
>Yes and no, generally speaking human senses work on a logorithmic scale.
your saying the same thing.
I hope you click on the wiki Weber-Fechner law of the other reply to understand that measuring perception that is a percentage change, requires a log scale to measure it, really the best explained math I have read in a while. (he didn't point out you were saying the same thing as the parent.)
>>It's how we're wired to see the world, in terms of percentage differences.
>Yes and no, generally speaking human senses work on a logorithmic scale.
lol, and you aren't wired mathematically.
It has been statistically shown that helmets increase the risk of head injury.
Exactly. It isn't. They should have seen this from the start. If you made the memory, wouldn't you only send the best to be reviewed?
Once again, correct me if i'm wrong.
Honesty may be the best policy, but by process of elimination, dishonesty is the second best policy.
Consumer Reports lacks technical expertise in many areas, but at least their approach has some level of integrity.
CR takes an approach that is valuable to the very largest number of people possible. Their computer reviews are probably uninteresting to a computer expert, and their auto reviews to a mechanic. But they provide useful high-level information that has one terribly useful characteristic -- it can be trusted.
It is *unbelivably* difficult to get information that can be trusted when you have whole industries built around not just feeding misleading information to the consumer, a la advertising and marketing, but around figuring out how to mislead the people that feed information to consumers. Buy off or influence reviewers, get desired songs played on the radio, etc. I'm sure that advertising agencies will start addressing Wikipedia soon enough.
I agree that more specialization might be nice, but it's difficult to have a broad enough appeal to be completely subscription-funded and thus remain neutral if you become too specialized.
Any program relying on (nontrivial) preemptive multithreading will be buggy.
My god, who can even read the article, that website is such a piece of utter shit. The article is divided up into 10 pages with only a couple of paragraphs on each page and a ton of flashing adverts.
Screw them.
I'm a slashdot geek and i am thinking "get a life"
sure this is one guy's experience with DDR2, but it's good stuff
crucial ddr2-667 doesn't perform well.
ocz gold ddr2-800 just doesn't work.
corsair ddr2-675 performs well above specs. and it's half the price.
in my abit AW8-MAX board with it's i955x chipset, I'm running the ram at 667MHz with timings of 4-4-4-12
I can clock the RAM to 220@3:5 or 733MHz!
as far as I can tell, the crucial ballistix I bought was stable at jdec specs.
the ocz gold didn't even post.
They're using their grammar skills there.
Exactly. It isn't. They should have seen this from the start. If you made the memory, wouldn't you only send the best to be reviewed?
Depends on if my main focus is to sell products with a hint of honesty, or just spread falsities and FUD.
I reject you're (sic) reality and substitute my own. Smartypants.
The ratio of people to cake is too big
Well, that is why I said "Yes and no". Maybe other people don't understand the relationship. "Logorithmic scale" provides a certain clarity in meaning. Yeah, that's it.
The ratio of people to cake is too big
Given that lack of information, I may still be able to explain a bit... Though I no longer work for a memory manufacturer, I do work for a semiconductor manufacturer, and the failure mechanisms of chips are still the same (DRAM or not)...
It almost goes without saying that ECC memory is more expensive due to the extra chips involved, but I'll mention it for completeness.
It also (almost) goes without saying that part of what you are paying for is the brand... which is as much perception as reality... note there is value for you as well: A valuable brand often means the company will make replacements or reparations at low effort to you in order to protect the brand.
Outside of that, there are still a significant number of things that could differ between two seemingly identical DRAM modules.
Part of the difference may be testing. My last post on Slashdot actually talked about testing relative to consumer semiconductor chips. Memory works the same way, in that memory sold under longer warranty periods or designed for higher level systems (mainframe, server-class products, etc.) will have longer and more stringent testing cycles.
This testing actually takes multiple forms, both static (DC) and dynamic (AC) testing. At speed testing is also possible, but adds additional cost. There are even more testing options pre-packaging, as you can actually do process tests (testing resistance in specialized devices on the chip, etc.) in order to detect how centered the process was on a particular device. Often, there are additional I/O that are never pinned out in a package (but are therefore exposed on an unpackaged die) that can serve some of these specialized testing purposes. There are also tests that can detect how much the device varies across the chip - you'll see this latter item occasionally referred to as ACLV [across chip linewidth variation] - the lower the better, of course. Chips are now sensitive enough that significant ACLV can actually make a processor completely fail to operate.
Note also some of the higher priced memory may also be memory that was perfect yield off of the manufacturing line (didn't require blowing a fuse in order to get the memory to operate). [If you didn't know it, a 64Mb DRAM chip might actually have 72Mb of cells, the redundancy is to increase net chip yield] Using a fuse on a DRAM isn't a big deal if the defect was a spot defect, but other defect mechanisms might be more problematic.
Cluster defects are one example. These can be caused by a piece of dirt skidding across the wafer as it is spun, leaving a trail of destruction which can be seen as a large arc across the wafer. Cluster defects often include both large defects that will cause measurable failures that require fusing as well as small (often undetectable) defects that degrade performance. These smaller defects often show up as soft error rates that appear above the intended design point of the memory. Since this requires significant at-speed testing to detect, they can slip through testing. However, many manufacturing lines do significant (if not complete) wafer inspection at each process stop, so the company may know which chips were subject to cluster defects and avoid packaging those as high-end memory. (I don't know that any companies do this, I'm just saying that the existing technology makes such a process possible - it might be too expensive to make it practical, however)
Additionally, a timing problem due to an overly thin layer somewhere in the process could demonstrate systemic defects due to electromigration (or other lifetime effects) during the latter part of the life of the s
Can anyone even tell me if DDR2 is an official standard yet? Because last I heard, it wasn't - and that was AMD's reason for never supporting it. AFAIK, DDR2 is basically just a rogue project for overclockers. Like anything else designed for overclocking (Motherboards, Graphics Cards), the OC editions are never exact anyways.
I do, its not about that, your missing the point.
The point is, it doesn't matter what the review was about, or what the user was doing. The reviewer of a product is taking the product, using it, or performing tests on it, and then using the information to inform the public, the consumers.
To really do that, they have to have sample of the product that is representative of what you can buy in the store. Its not like you could go out and buy a new memory stick from a retail store that was tested out and deemed to be the best of the best. You buy one that tested out to be "within spec".
The fact is, if... IF they are right, and the RAM manafacturer is testing the ram to a higher standard, and then sending them to reviewers, then they are essentially asking a reviewer to review a product thats not actually available to consumers, and passing it off as one that is (that is ram, tested to a lower standard)
That means they are not only being dishonest, but tricking an independant third party into shilling for them.
-Steve
"I opened my eyes, and everything went dark again"
Yet, the fact remains that the stick of RAM available in stores is MUCH faster than it's supposed to be, faster in fact, than this speed deficit that everyone's bitching about. People here really need to get some perspective. What's the bigger difference, 471-417 or 417-333?
Karma: It's all a bunch of tree-huggin' hippy crap!
But what does this have to do with memory? you ask. Ok, technically you didn't ask, but I've never let that stop me before. In your desktop, a 2.5% difference in memory speed is imperceptible. If you have a server farm with enough machines, that could be the equivalent of an extra server or two.
That said, my problems with TFA are:
1. The size of their statistical universe, and
B) Of course a manufacturer is going to send you their best. A responsible lab will do their own random sampling.
Some see the vessel as half full; others see it as half-empty; We pour it out on the floor and laugh
No. The manufacturers are all supposed to provide a sample that is representative of the product that a consumer would purchase off the shelf. They are supplying this sample to a third party in hopes of a favorable review. The companies have been caught cheating.
It wouldn't surprise me, in fact I think some explicitly admit it, that many of these hardware sites were set up to review hardware as a precursor to ganking free gear.
Fair enough, I'll stop feeding the trolls now.
Karma: It's all a bunch of tree-huggin' hippy crap!
Since the variations in the sticks are random there's no way of knowing whether the stick was really pre selected or just a lucky draw. The comparison stick bought from the store might have been extraordinarily bad or just average. The real problem this exposes is the testimg method of picking one stick and testing it. What if you got a dud instead and told everyone that stick can't be overclocked at all? Wouldn't be the average experience, either.
Justice is the sheep getting arrested while an impartial judge declares the vote void.
True. Put a four hundred pound weight in one hand and a four hundred and one pound weight in the other and the person wont even bother trying to figure out which is heavier.
Fast machines, powerfull AI, impulsive invention,... All I lack is a good espresso machine!
Given F1-quality engines usually explode after about 620km that seems like a self-defeating plan.
I guess today is a passable day to die.
To buy something from the store with the intention of returning it is fraud. To have the intention to return something else as if it were what you bought is at least as bad (legally).
Please don't think you can use stores as lending libraries.
I don't know if I want to get my reviews from individuals who would defraud stores out of money (handling/restocking fees, turning their stock into non-new stock which doesn't fetch the same price).
http://lkml.org/lkml/2005/8/20/95
Ask Ford or Mazda (again Ford) about this. Both companies doled out lots of money because they sold cars that didn't make the HP they claimed (Ford Mustang Cobra and Miata, respectively). Infiniti also had to spend a lot of money on their Q45 customers trying to make up what seemed to be a HP deficit (although I don't know if it was ever proven) in the Q45.
Car companies do test their cars, and the HP at the shaft (BHP) is supposed to be at least what is advertised, or else. So usually it is. There was a good story in the Detroit News recently about how companies were retesting their cars under a more standard set of rules now, and the ones that didn't match up well (either over or under). In this case, the companies aren't liable, because it is assumed their cars made the rated HP, but under different testing conditions. This loophole is now closed and cars going foward must be tested independently under these rules.
http://lkml.org/lkml/2005/8/20/95
Well sorta....
The fact is that the same process produced both. Both will be in the pool.
Its not so much about being 471 or 417, its about being capable of being run that fast. Some can handle it, some can't. Its like CPUs, the processors come off the assemply line, and are tested at various speeds.
Some can run at the highest speeds, some don't, but run fine at a lower clock rate... so they test em, mark em, and toss em into their respective bins based on testing.
Thats one reason overclocking has worked so well, sure you can push the chips faster than they were tested for, but at some point they become unstable, so you throttle back to where it works.
Basically overclockers are redoing the cpu frequency testing that the vendor did and determining their own clock freq. Albeit, with more granularity to the test, but.... probably alot less sophistication.
Anyway, even if the samples were random, these same ratings could and would show up. Thats the real problem with this test that toms hardware is doing. Any random chip could work at 471 or 417 or even 400 or 500 (feel free to substitute those for actually possible numbers, i am no overclocker).
The only way to fairly "catch" this would be to take say... 100 store bought, and 100 from the review process. If the chips bought from the store range from say 390-480, but not a single chip from the reviewer fell lower than 450.... then its pretty obvious.
N=1 studies are fundamentally flawed. Sad, it would be nice if someone really did do this properly. Any one of the store bought chips could be at the high end or low end of the spectrum. Without doing a seires of tests on different chips, there is no way.
Also... main boards and CPUs? They should be swapping which module is in which "identical machine" and redoing each test too... that way they can cancel out all of these exact same quality differences between manafactured chips that exist in every chip in the main board and in the CPU.
Do the complete set of tests twice, once in each machine... then repeat that for 10 different store bought and 10 different ones sent to review... thats 40 tests per chip model per manafacturer.... THEN they will have something worth saying.
-Steve
"I opened my eyes, and everything went dark again"
Yeah, because we all know that marketers should obey the non-standards... Seriously though, there wasn't a standard for binary prefixes, there was a need for one, and now there is one. What kind of problem can you have with that?
There are two types of people in the world: those who divide people into two types and those who don't.
Comment removed based on user account deletion
Many car reviews are done in pre-production cards that other auto journalists have already beaten the heck out of, and are in worse shape that one you'd drive off the lot. The better reviewers mention shortcomings in their reviews to do you the favor of letting you know the weak spots to look for to see if they've kept up on their promises to the reviewer.
While I agree that it's true that people should expect that nothing is guaranteed if you overclock the memory beyond its specification, you have to remember that manufacturers are submitting their memory to THG for testing specifically KNOWING it's going to be OCed! They want to look as good as possible. Sure, a FooCo-533 may perform just as good as a BarCo-533, but if the FooCo-533 fails when overclocked 566 and the BarCo-533 doesn't, BarCo comes out looking like they have a higher-quality product. And even if the consumer doesn't intent to overclock his memory, might he not feel SAFER with the BarCo memory? After all, if the FooCo memory can't handle a litte overclocking, how many FooCo-533 modules are unreliable even at 533? Such a conclusion is not logical, and not supported by the evidence, but it is how some consumers think.
But perhaps more to the point, pointing out how a manufacturer's memory sample may not be representative of retail quality isn't so much as to say "Gotcha!" to the manufacturer, but rather to help the hardcore computer hobbyists KNOW which memory they can trust to overclock. If Tom's doesn't test retail-quality memory, then the consumers can't necessarily trust the results. By providing a comparison study like the current one, it actually does help them evaluate products better, regardless of the fact that they're using these products out-of-spec.
Bruce
If you have no intention of keeping it when you bought it, then you only pretended to buy it. You had one thing in mind ("borrowing" the item), and you did another (pretended to purchase the item) and it cost someone else money (the company), so you committed fraud. It's that simple.
If a store has to raise their prices to cover someone else's acts of fraud, it bothers me, because it hurts me. And if you owned the company, I'm sure you'd be concerned too.
Crimes aren't okay just because you don't happen to know personally the person you are hurting.
http://lkml.org/lkml/2005/8/20/95
really with a sample size of 2. freaking 2 they conclude that mem makers are shipping slower memory to retail. TOMS HARDWARE needs to attend high school. such a small smaple size can not prove anything. now if they had compared some 50/100 mem sticks from diffrent retailers all the country /world then i'd accepted that maybe their conclusion holds some weight.