RedLine Death

[Duonger]

maybe DFI should examine their voltage delivery systems, would a line filtering UPS help?

I have sent this link to the DFI Product manager and will speak with him hopefully tomorrow or the next day when he is back from Taipei.

Duonger

[felinusz]

You guys crack me up, there has been plenty of people with "old" Winbond and even TCCD going bad on DFI board, yet everytime this issue comes up there is always someone claiming that this only effects UTT. Go check 2 threads on this issue in AMD sections if you care.

I'm glad that you find the honest pursuit of answers humourous


Count how many UTT deaths we have on our hands. Count how many BH-5/BH-6 deaths we have on our hands.

Which number is overwhelmingly larger?

We must always expect some memory to die, regardless of what IC type it uses. But when a very large amount of memory that uses a specific IC type starts dying, and a trend starts to form, it is logical to blame the specific IC type.

[Duonger]

while these UTT memory do die in DFI boards, i highly doubt its a DFI problem. it's not a memory maker problem either. it's the IC's. there's a reason why they're so cheap. it's untested so actual quality is unknown. if they do run for a long period of time, be happy. if they die, dont be surprised or angry.

overclockers love it because its cheap and does what we want. but if you want stability and reliability, UTT is at the bottom of the list. i personally stick to micron or samsung based memory. they overclock and they actually live

Poiuy223

Once we buy these Untested Chips it is US who do the testing and binning for specific yeild rates. So there is testing involve at the TSOP level.

Duonger

[bachus_anonym]

Poiuy223

Once we buy these Untested Chips it is US who do the testing and binning for specific yeild rates. So there is testing involve at the TSOP level.

Duonger

But can you honestly say that you guys test them same way that ICs fabs are doing it? I mean, testing them in the 70C ovens, at different voltages for EXTENDED periods of time?

I just think that testing standards that fabs use vs modules mfrs are much different. I would be more than happy if I saw any rep (OCZ, Mushkin, Gskill, etc) rep tell us what is the actual testing procedure that they're using and how different it is and efficient in finding bad ICs.
I auusme that when they're saying hand-tested, that means they use those memory testers, like RAMCHECK or sth similar. If so, then this is not exactly same standard as fabs use. At least, from what I know... I will stand corrcted if anyone can provide better insight

[ben805]

I guess these chips just aren't capable of running high voltages for extended periods of time the way BH-5 were.

It seems to be the case, but cowpuppy fed them only 3.3v....and the recommended vdimm is up to 3.5v so I don't think high vdimm causes it.

sounds good to me, jeez you guys at mushkin dont mess around.

agreed, mushkin have excellent customer supports!!

But can you honestly say that you guys test them same way that ICs fabs are doing it? I mean, testing them in the 70C ovens, at different voltages for EXTENDED periods of time?

I don't know if Mushkin test these UTT chips the same way as BH5 at the lab but one thing I know for sure is, if your sticks died they'll ship a new pair to you....warranty is good as long as you didn't take off the heatspreader and didn't overvolt the hell out of them and got them fried due to user error. Hell, it's a 'limited' lifetime warranty so if my sticks died again, just send me another pair! lol

[Duonger]

But can you honestly say that you guys test them same way that ICs fabs are doing it? I mean, testing them in the 70C ovens, at different voltages for EXTENDED periods of time?

I just think that testing standards that fabs use vs modules mfrs are much different. I would be more than happy if I saw any rep (OCZ, Mushkin, Gskill, etc) rep tell us what is the actual testing procedure that they're using and how different it is and efficient in finding bad ICs.
I auusme that when they're saying hand-tested, that means they use those memory testers, like RAMCHECK or sth similar. If so, then this is not exactly same standard as fabs use. At least, from what I know... I will stand corrcted if anyone can provide better insight

Actually we do do test these modules beyond industry standards. They are tested at ranges of -40 degress celsuis to +85 degrees celcius. We also screen our modules in environmental and accelerated life tests as well as software analysis tests.

I can go into more but there is a lot of testing that we do to ensuer quality control on our components as well as modules.

[ben805]

Actually we do do test these modules beyond industry standards. They are tested at ranges of -40 degress celsuis to +85 degrees celcius. We also screen our modules in environmental and accelerated life tests as well as software analysis tests.

I can go into more but there is a lot of testing that we do to ensuer quality control on our components as well as modules.

Hey Tom, glad to know that they were being tested beyond the industry standards!! but can you please tell us more as to what else you guys put them through in order to passed the mushkin standard?

but geewizz...tested at +85'C!!

[bachus_anonym]

@Duonger,

That's great to hear, Tom.

But don't you think that UTT ICs-based sticks seem to have a higher failure rate than BH-5 or TCCD. I've had various BH-5 sticks, had some UTTs, still have some TCCD; and I gotta tell ya - I've never seen sooo many dead sticks, until UTT showed up.
I know that may be a good business-model to buy cheaper UTT, be able to replace them in case the go bad and still be "above-the-red". But that looks pretty bad from our point of view. I know that in that case warranty and customer service has to be beefed up, because I tell you what I think: Mushkin and other companies are going to need it, probably more than before

[Jaco]

while these UTT memory do die in DFI boards, i highly doubt its a DFI problem. it's not a memory maker problem either. it's the IC's. there's a reason why they're so cheap. it's untested so actual quality is unknown. if they do run for a long period of time, be happy. if they die, dont be surprised or angry.

overclockers love it because its cheap and does what we want. but if you want stability and reliability, UTT is at the bottom of the list. i personally stick to micron or samsung based memory. they overclock and they actually live

That's my conclusion also... Bottom of the barrel...
No garantee UTT will be stable/reliable . It remains to be seen how long these sticks are gonna hold up.
Lots of horrorstories regarding UTT that's for sure.

I'm gonna try some Ballistix if I can find them here.

[hovo73]

I'm glad that you find the honest pursuit of answers humourous


Count how many UTT deaths we have on our hands. Count how many BH-5/BH-6 deaths we have on our hands.

Which number is overwhelmingly larger?

We must always expect some memory to die, regardless of what IC type it uses. But when a very large amount of memory that uses a specific IC type starts dying, and a trend starts to form, it is logical to blame the specific IC type.

Well obviously there are going to be more UTT cases, because that is what selling lately. "Old" Winbond chips haven't been selling for a while, thus the name "old" . And most of the people who owned them had 256MB sticks, so they probably don't want to use them nowdays anyway.

The point is, it is not only UTT that is dying on these boards, yet everytime we start discussing this people start claiming that it is.

[Duonger]

can you guys point me to most of these cases. I havent heard of them till this post. Can you direct me to other posts?

We have been testing these UTT for 3 months now. why do u think we took so long to release our parts.

Please enlighten me as i will try to test and find out if it is chips, boards, booster or combination of both or all three.

Duonger

[tictac]

can you guys point me to most of these cases. I havent heard of them till this post. Can you direct me to other posts?

We have been testing these UTT for 3 months now. why do u think we took so long to release our parts.

Please enlighten me as i will try to test and find out if it is chips, boards, booster or combination of both or all three.

Duonger

Very important, read it before you kill your RAM
http://www.xtremesystems.org/forums/...ad.php?t=59414

My DFI has flatlined my redline....
http://www.xtremesystems.org/forums/...ad.php?t=64720

DFI NF4 the UTT/VX/BH5 KILLER READ IF YOU HAVE DFI
http://www.xtremesystems.org/forums/...ad.php?t=62804

DFI Admits 5Volt Vdimm Problem
http://www.xtremesystems.org/forums/...ad.php?t=63932

Cold boot issues on DFI, i need testers
http://www.xtremesystems.org/forums/...ad.php?t=63715

[Jaco]

On GOT (dutch forum) i've read several horror stories.
http://gathering.tweakers.net/forum/...s/1039273/last

It's in Dutch , but maybe some of the guys wil reply here

[poiuy223]

You guys crack me up, there has been plenty of people with "old" Winbond and even TCCD going bad on DFI board, yet everytime this issue comes up there is always someone claiming that this only effects UTT. Go check 2 threads on this issue in AMD sections if you care.

PS. Does anyone know how which rail does Abit AN8 Fatality SLI board is using to power DIMMs?

i cant speak for everyone of course, but from all the memory reviews i've done over at insanetek, NONE of the TCCD have died on the dfi board. i've already had 2 kits of utt poop out. one from ocz and one from mushkin. duonger, im getting in touch with lee tomorrow. i understand that memory makers do test the chips before they use it. of course they have to. otherwise they wont be able to keep up with rmas. thing is, that kind or level of testing is not the same as the IC maker. its pinpointing down to better chips, but not totally wiping out the bad ones that may just accidentally slide by.

i personally just dont like utt. i LOVE the performance but the death rate is high, not to mention the need for voltage. benching, WRs, overclocking, i'd use it in a split second. 24/7, dont even think about it.

i believe the abit sli should use 5v. i'm finding out.

[hovo73]

i cant speak for everyone of course, but from all the memory reviews i've done over at insanetek, NONE of the TCCD have died on the dfi board. i've already had 2 kits of utt poop out. one from ocz and one from mushkin.

Check the threads in tictac's links (two messages up), especially 3-rd and 4-th.

[poiuy223]

Check the threads in tictac's links (two messages up), especially 3-rd and 4-th.

well yes of course tccd's die, but like someone else mentioned in this thread, which IC dies more? tccd has been out for a long time now. utt has been out for less than a year. weigh the two. there is a reason behind it.

[hovo73]

tccd has been out for a long time now. utt has been out for less than a year. weigh the two. there is a reason behind it.

Hmmm .... DFI has been out less than a year too ...

I am not arguing about percentages, I am simply pointing out that the claim that "only UTT dies" is not accurate and in fact counter productive in trying to pin point the problem.

[robberbaron]

I contend that the nForce4 board was a rushed product and a witches brew of bad hardware here and there. My nForce3 DFI board has been running UTT at 3.2v like no tomorrow without any issues. Nobody is reporting the 250GB-UT killing UTT

[felinusz]

Hmmm .... DFI has been out less than a year too ...

I am not arguing about percentages, I am simply pointing out that the claim that "only UTT dies" is not accurate and in fact counter productive in trying to pin point the problem.

Fair enough. And indeed, it is **NOT** just UTT that is dying.


One way to look at it, is that it doesn't matter. Run your sticks under 3.5V, and they're under warranty.

If they die, return them for a fresh set.

So, in one sense it isn't even an issue because of excellent warranty coverage.

[madpete]

[QUOTE=Duonger,

In the case of the thread starter, as well as one of the cases that I am personally aware of the common thread was that in both cases the boards were shut down, then later restarted. You've just stated that you guys are running one 24/7 without issues.

Is there any chance that the problem could be caused by the sudden surge of high voltage when the boards are restarted?

Maybe as part of your testing you might want to consider having one rig that gets shut down and restarted, with timed intervals of run and down time simulating a non 24/7 useage pattern.[/QUOTE]

This is exactly how mine died

[jiff]

I contend that the nForce4 board was a rushed product and a witches brew of bad hardware here and there. My nForce3 DFI board has been running UTT at 3.2v like no tomorrow without any issues. Nobody is reporting the 250GB-UT killing UTT

Erm yes they are I have had a stick die, I also know of 2 others

[poiuy223]

any kind of memory will be found die-able (is that a word???) in any motherboard. it just so happens that dfi actually supplies the voltage. im just saying that UTT dies more often than others because there's no assurance at the IC fab level

[hovo73]

Are there any users with Abit AN8 SLI Fatality? It would be really interesting to see what is their experience with UTT with up to 3.55 VDIMM.

[poiuy223]

at this point, im tempted to buy one.

[felinusz]

I have a feeling that if everyone used DDR Boosters, we'd have an epidemic of high voltage memory death "because of the boosters" as well. 8-/

To an extent, voltage is voltage. If there aren't any spikes (and voltage spikes would likely kill the A64 memory controller before memory anyhow, as someone wisely pointed out), then the voltage supply isn't the issue, it's the high voltage itself (current density, heat as a by-product of high-overvolts, potential immediate damage to transistors, etc.).


I hope some more detailed information presents itself soon. It is excellent to hear that Mushkin and other memory companys thoroughly test their UTT ICs, and even better to hear that the industry is persuing these problems .