Anandtech Motherboard reliability database?

[Soccerman]

why not?

this would be similar to the Storage Review hard drive reliability database (I'd like to see how their filtering engine works though before totally trusting it on it's results)! it's a great idea, why not extend it to what Anandtech does best?

what do you guys think?

I would probably just put out a model number of every board (a board with RAID as an option would all be one board, unless people have proof of RAID boards failing more often than their counterparts).

the question would be similar to Storage Review: is the board running? does it have 'unique' problems (which might be a defective board, where you could probably RMA it to obtain a fully functional board) that haven't been resolved through BIOS updates, driver updates, or the like?

I guess what I want to see most is whether or not ABIT has that many boards that are DOA (obviously requiring an RMA). it also appears that a fair amount of ECS boards shipped that weren't DOA, but that did have problems that couldn't be resolved (again, if a Power Supply wasn't on the AMD approved list but upgraded to an approved one, and that fixed the problem, than I probably wouldn't consider that a partly defective mobo, even though there might be ones that work fine with power supplies of lesser quality).

 

[LXi]

Soccerman has my 100% support. I was thinking about that myself.

 

[RSI]

Soccerman also has RSI's support. Excellent idea!

-RSI

 

[Noriaki]

Ahhh the motherboard, the most often overlooked component in building a good system. And one of the most important!

And excellent idea in my opinion!

 

[AndyHui]

A very good idea.

 

[Soccerman]

hmm.. ok then, it looks like a couple long days are coming ahead..

/me wonders what I've just dug myself into

btw, keep the comments coming.. tell me what you think I should include and what I shouldn't. next I gotta see if I can talk to The Boss before I go much further..

 

[AndyHui]

Muahahah....A lifetime of slavery at AT....just like the rest of the staff...

 

[Shalmanese]

I can see one problem with it though.

People with working mobos arent likely to submit wheras people with defective mobos are so you will end up getting the most popular mobo in there the most all other things being equal so it will discourage people from buying that mobo. I can really see a way to do this properly

 

[AGodspeed]

I think this is a great idea. There should be standard benchmarking tests that us ATer's can run on our computers as reliability tests.

Good idea Soccerman.

 

[LXi]

<<People with working mobos arent likely to submit wheras people with defective mobos are so you will end up getting the most popular mobo in there the most all other things being equal so it will discourage people from buying that mobo. I can really see a way to do this properly>>

That's true, but we can ask the surveyors to submit results of all of their motherboards, every person should at least have one working mobo, right? We can also encourage everyone to survey, regardless of whether you have a dead mobo or not, we should also allow them to survey RMA'ed replacements, the more results the better.

 

[Taylorm]

nice idea

 

[Mitzi]

Certainly a features I'd like to see!

 

[Ausm]

I think that would be cool...


Ausm

 

[GTaudiophile]

I wonder if it could be integrated into the AT Rigs section. Like, when you add a new rig to your profile, beside each entry, you also have the ability to rate the product's reliability. That would cut down on the negative skew you would find if only people with defective products submitted their review/rating.

 

[whoya]

Got my vote. I have been wondering why know one has done this?

 

[ucla88]

this is a great idea. you could also extend it to other hardware components at some point in the future.
i made a rambling post awhile back, similar to this, but your idea is actually practical.
hardware reliability
i kind of like the idea of randomly tracking hardware that is purchased, to get around this selection bias, because i think shalmanese is kind of right. (see the other thread-i was sort of cranky is some of my replies though. oh well, it was probably a bad day)

 

[slicksilver]

would be great for the newbies..............just hope the boss checks this thread

Raj

 

[DSTA]

@Soccerman

Like the idea, but please only for real show stopping defects, not "buggy" products. I think one would have to be strict on that, or at least post the rest of the system specs. I want to know about what memory and PSU are used before I'd judge a Mobo as defective-

@ucla88

I kind of like the idea of randomly tracking hardware that is purchased, to get around this selection bias, because i think shalmanese is kind of right.

That's a brilliant idea IMO and might actually produce valuable data one could sell someone ;-). Wonder if market researchers do something like that already.


(see the other thread-i was sort of cranky is some of my replies though. oh well, it was probably a bad day)

PMS?

;-)

SCNR (and I did not think you sounded cranky).

 

[Soccerman]

ok, here's what I've figured out so far. the VERY MINIMUM it appears that you could report, is the name of a motherboard (selected from a list, not typed in), and if it is working (ie, no problems), or not working at all. I would also probably want a purchase date (in quarters, not the specific date) cause people might like to see how well older boards from a manufacturer stood the test of time. Also, the power supply would be required (I'd like to try to make a drop down list for everything, cause I don't want this to end up like the overclocking database where people didn't know the numbers on their chip cause it was already in the computer). of course every entry could have a comment, so if the board is a little weird, you could put down what happened to u. it might just happen to be a common occurence.

the results could be sorted by chipset, by purchase date, or by motherboard model..

now, as for this other stuff.. we might have to rely on FAQs to help achieve some filtering.. for example, if we want to include people who have partly working boards (ie it works with one video card but not another), they need to have tried everything we can think of (drivers, BIOS updates, OS updates) in order to get the motherboard running normally. typically the forums are where this happens, but not everyone who will visit this website will use the forums..

I like the idea of eventually moving off to other peripherals, like video cards etc, but we'll stick with motherboards for now, as its such a problematic part of the computer.

also, this so far is going into whether the motherbaord works as advertised or not. adding stability into this would be a hole new can of worms.. some people might say that their computer is stable even if it crashes on them once a day, whereas others would say it's stable only if it's been up for 30 days and nights without a crash.

so we can't just say, is your computer stable? yes/no. it would probably have to be a metric. 'on average, how often does your computer crash a day?' or getting REAALLY into it, a testing program There should be standard benchmarking tests that us ATer's can run on our computers as reliability tests. wow, that would be something else! download a program to test the CPU and mem (probably use something like prim95), run a few 3D tests to be sure OpenGL and DX work, do some DX testing of audio? then report results back? hmm.. we'd need a liscence agreement!

DSTA I totally understand what you mean. I was thinking of having it so you select your mobo, then select one of 3 levels of reliability: works, doesn't work, and partly works (then you would HAVE to list pretty much every system component so we could try to pinpoint the problem).

ucla, after looking breifly over that thread, I noticed near the bottom that people are worried about selection bias.. ie, these numbers only hold true to people like us (gamers, who treat their hardware differently then your office dude in a cubicle), they might even test durability (if they did, no doubt the guy reporting probably has a comment in the SR database saying he dropped it and his heart was pounding). that's not a problem, it just means that you can't use these numbers quite as well for blanket statements about a company, which is fine anyway, becuase there are always good and bad products to come out of a company. besides, would we really want our data to be used by MS? I like that idea about a card being handed out at newegg though