DFI PSU Issues

[[Spectre]]

I also had very heavy instability overclocking my SLI-DR, corrupting two OS installs. While there is no guarantee that the PSU caused it, and that there was some funky BH-5 involved that I didn't use before, I do the same kind of overclocking all the time and didn't have that kind of supercrash anytime before or after using the Athena.

When using a DFI baord the first thing one should looks at for the problem should be the motherboard/RAM. DFI makes boards that violate spec....which is why overclockers like them....but the spec's exist for a reason....so this is also why you see so many problems with DFI boards and DFI boards not working with other spec compliant hardware.

 

[uOpt]

When using a DFI baord the first thing one should looks at for the problem should be the motherboard/RAM. DFI makes boards that violate spec....which is why overclockers like them....but the spec's exist for a reason....so this is also why you see so many problems with DFI boards and DFI boards not working with other spec compliant hardware.

I seem to have one of the best SLI-DR in existence, and I used it without any of the above problems with PSUs from Seasonic, Sparkle, Zalman (FSP made), topower OCZs, FSP-made OCZs, PCP&C 1 KW and probably more that I forget. I't didnt break either still worked fine after the Athena experience. But as I said, the RAM involved could be a culprit, too.

I have never seen a conclusion in the ongoing battle over coldstart issues between DFI and the PSU makers, we don't know who's fault it is.

 

[[Spectre]]

I have never seen a conclusion in the ongoing battle over coldstart issues between DFI and the PSU makers, we don't know who's fault it is.

If just one brand of board with one group of engineers making/designing boards that violate specs has issues with a number of different brand/manufacturers with a number of engineering teams making spec compliant PSU's the answer is rather obvious. People have some funny rose colored glasses when it comes to DFI. They quite simply make a very unstable product but because a small group of people get awesome overclocks out of them when the moon and stars align properly people think they are the second coming.

 

[thecoldanddarkone]

If just one brand of board with one group engineers making boards that violate specs has issues with a number of different brand/manufacturers with a number of engineering teams making spec compliant PSU's the answer is rather obvious. People have some funny rose colored glasses when it comes to DFI. they quite simply make a very unstable product beut because a small group of people get awesome overclocks out of them when the moon and stars align properly people think they are the second coming.

hahahaha

 

[[Spectre]]

Oi....some days my english really sucks.

 

[uOpt]

If just one brand of board with one group of engineers making/designing boards that violate specs has issues with a number of different brand/manufacturers with a number of engineering teams making spec compliant PSU's the answer is rather obvious. People have some funny rose colored glasses when it comes to DFI. They quite simply make a very unstable product but because a small group of people get awesome overclocks out of them when the moon and stars align properly people think they are the second coming.

As I said, my SLI-DR is not only my best overclocker but by far my most stable board in the socket 939 market.

The Asus A8N-* line is by far the worst for me.

 

[[Spectre]]

As I said, my SLI-DR is not only my best overclocker but by far my most stable board in the socket 939 market.

The Asus A8N-* line is by far the worst for me.

That is nice but it still doesn't refute the point that non-spec compliant DFI boards have a wide range of compatibilty issues with spec compliant hardware.

 

[uOpt]

That is nice but it still doesn't refute the point that non-spec compliant DFI boards have a wide range of compatibilty issues with spec compliant hardware.

It's not that I disgree entirely and I realize that I just have very good luck with my SLI-DR which does everything I want.

Do you happen to have more concrete info on what exactly the DFI boards do wrong?

Over at DFI-street people throw the ATX spec at each other without anybody ever quoting as much as a single number of how much is being pulled where. That way we'll never get to the ground of it.

 

[[Spectre]]

Do you happen to have more concrete info on what exactly the DFI boards do wrong?

Well there was the issue with +5vsb draw a while back. Also if you notice boards that have molex connectors on them then they are doing something "funny" for power distribution on the board and that isn't part of the spec. DFI trys to cover up a LOT of their design choices by using things like......requiring 24pin PSU's that are greater than 480w in order to offer support....whcih again has nothing to do with the spec. As for the rest of the fun issues...no not really since DFI refuses to come clean....which is why you don't see any numbers tossed around. How "picky" are DFI boards with RAM these days?

 

[uOpt]

Well there was the issue with +5vsb draw a while back. Also if you notice boards that have molex connectors on them then they are doing something "funny" for power distribution on the board and that isn't part of the spec. DFI trys to cover up a LOT of their design choices by using things like......requiring 24pin PSU's that are greater than 480w in order to offer support....whcih again has nothing to do with the spec. As for the rest of the fun issues...no not really since DFI refuses to come clean....which is why you don't see any numbers tossed around. How "picky" are DFI boards with RAM these days?

The RAM pickyness seems to vary by individual board.

My SLI-DR runs almost everything. I posted a thread at dfi-street running with 4 different modules, mixing Geil One, some ECC stuff and Ballistix.

The only RAM that ever refused to run was some obscure Mushkin which didn't work in my A8V-E SE either.

 

[[Spectre]]

The RAM pickyness seems to vary by individual board.

My SLI-DR runs almost everything. I posted a thread at dfi-street running with 4 different modules, mixing Geil One, some ECC stuff and Ballistix.

The only RAM that ever refused to run was some obscure Mushkin which didn't work in my A8V-E SE either.

Ok....and that is acceptable? A spec means that every piece will work with every other piece every time barring defects in material or workmanship.

That is my point. With DFI that does not occur.

 

[uOpt]

Ok....and that is acceptable? A spec means that every piece will work with every other piece every time barring defects in material or workmanship.

That is my point. With DFI that does not occur.

???

The RAM did work on neither bord. It almost certainly was the RAM's fault.

 

[madmat]

His point is how hit-or-miss DFI is from board to board. Yes, yours has no real issues being nit-picky with ram while other guys with the same board have issues with ram. The same thing goes with power supplies. DFI is horribly beyond spec so one board could be fine with brand X PSU while another board of the same model/config/bios revision/etc could totally bork with it.

Get the idea?

 

[[Spectre]]

???

The RAM did work on neither bord. It almost certainly was the RAM's fault.

I was referring to the part I bolded as madmat said.

 

[uOpt]

Well, while some random DFI board may work or not work with a given pair of RAM modules, you have any given Seasonic work or not with with the SLI-DR.

My point is, I have seen no actual measurements of the PSU loads and currents for the turnon problems or for the overload protection kicking in problems. Nobody submitted even measurements, much less an interpretation of these measurements with regards to the ATX specification.

I certainly agree that the behavior of the guys over at DFI-street is not inclined to make you rule in their favor, but that doesn't change the fact that we have no hard info either way. Just because some support people close threads when they become uncomfortable doesn't prove anything.

 

[[Spectre]]

Well, while some random DFI board may work or not work with a given pair of RAM modules, you have any given Seasonic work or not with with the SLI-DR.

My point is, I have seen no actual measurements of the PSU loads and currents for the turnon problems or for the overload protection kicking in problems. Nobody submitted even measurements, much less an interpretation of these measurements with regards to the ATX specification.

I certainly agree that the behavior of the guys over at DFI-street is not inclined to make you rule in their favor, but that doesn't change the fact that we have no hard info either way. Just because some support people close threads when they become uncomfortable doesn't prove anything.

I still don't understand how you defend DFI. DFI already has a track record of making design choices that violate spec in regards to the +5vsb so why would one rationally believe the issue is with spec compliant parts of compaines that have not done so? There is some fundamental problem with DFI boards that is causing incompatibility with spec compliant PSU's/RAM etc. If DFI refuses to say anything about the situation (which they are) there isn't much we can do to get to the bottom of it. With Antec/Asus they wen't rounds to figure out what the problem was after they got reports. When DFI get's reports what do they do? You already said as much...they sweep it under the rug. Without any input from DFI there isn't much we can do other than not use their product. All the measurements in the world (which most users are incapable of getting) won't get us anywhere without DFI working with customers.

When proven non-spec compliant DFI products in combination with spec compliant PSU's etc lead to stability issues you have to understand the non-spec compliant part is the issue....either that OR DFI simply has horrendous quality control because as you say it varies from board ot board. So either DFI has a design issue (which is the case in my estimation) or they have a QC issue (which seems to be what you have pointed towards...but not said so correct me if I am wrong)....either way there is a reason DFI baords NEVER go into systems I build for clients.

 

[uOpt]

I still don't understand how you defend DFI. DFI already has a track record of making design choices that violate spec in regards to the +5vsb so why would one rationally believe the issue is with spec compliant parts of compaines that have not done so? There is some fundamental problem with DFI boards that is causing incompatibility with spec compliant PSU's/RAM etc. If DFI refuses to say anything about the situation (which they are) there isn't much we can do to get to the bottom of it. With Antec/Asus they wen't rounds to figure out what the problem was after they got reports. When DFI get's reports what do they do? You already said as much...they sweep it under the rug. Without any input from DFI there isn't much we can do other than not use their product. All the measurements in the world (which most users are incapable of getting) won't get us anywhere without DFI working with customers.

I'm not defending DFI per se.

What I am doing is that if the PSU manufacturers say it's DFI's fault they should present measurements.

I don't understand what is so hard about it. All they got to do is plug in a couple amperemeters and show us where DFI's draw is below or above the spec.

It is the absense of such measurements that make me suspicious that the situation is more complex than it seem.

Also, in some of the closed threads there are quotes from mainboard manufacturer's tech support that are entirely bogus, and DFI has the right to show those to undermine the competence of their adversaries. That it is stupid of them to close the threads is a different, non-technical matter.

 

[[Spectre]]

I'm not defending DFI per se.

?

What I am doing is that if the PSU manufacturers say it's DFI's fault they should present measurements.

I don't understand what is so hard about it. All they got to do is plug in a couple amperemeters and show us where DFI's draw is below or above the spec.

It is the absense of such measurements that make me suspicious that the situation is more complex than it seem.

No that is not all they have to do. If the issue isn't draw in this case but is DFI's onboard power regulation, routing, sense etc......it would be missed. Yes the situation could be much more complex but it is still a DFI issue since the spec compliant parts work fine with other spec compliant parts. If they don't work with non-spec compliant parts it is a non-issue for the PSU manufacturer unless they want it to be...it is up to DFI to make their parts conform to spec which they have chosen not to.

I can only say that so many times...and I believe I have hit that limit.

 

[dandragonrage]

My Seasonic works fine with my Ultra D. Maybe I just got lucky?