Monday, January 17th 2011

Socket Pin Burnout Returns to Haunt LGA1155?

Last year, credible reports of a design defect with LGA1156 sockets made by a certain component manufacturer in particular, made headlines. Reviewers found that extreme overclocking using increase voltages, in test cases, caused certain pins of the socket to burn out, damaging both the board and processor. The defect was found to be caused by shorting between the pins and the socket causing tiny electrical arcs. Motherboard reviewer from TechReaction.net discovered that his Gigabyte GA-P67A-UD4 and GA-P67A-UD7 samples showed signs of socket burn. The publication is yet to receive a reply from the board manufacturer.

LGA1155 and LGA1156 are very similar in physical pin layout, chip package, and retention clips, although the pin-maps are significantly different. TechReaction comments that while only "extreme conditions" cause such pin burnouts, 24/7-stable overclocked settings could, over a period of time, create similar conditions. Only time, and extensive testing will reveal if burnouts are a cause for concern to enthusiasts.

Source: TechReaction
Add your own comment

65 Comments on Socket Pin Burnout Returns to Haunt LGA1155?

#1
overclocking101
HillBeast said:

But this is the bad press and bad press sticks. I don't recall seeing anywhere near as much bad press about 1366 arcing as 1156 or 1155 have, let alone at all. I'm probably wrong on that one though because Google can be unreliable.
they see you trollin they hatein! 1366 has the OLD 775 style HOLD DOWN plates. 1156/1155 DO NOT these sockets were more new tech then 1366 was, sure with real new tech there are always problems for early adopters ALWAYS. it has nothing to do with money dip shit because quite literally for high end on both platforms it costs about the same within $100 of each other, except cpu's. so before you buy something just to look like you have all kinds of money to spend, you should really think about going mainstreem :nutkick:
Posted on Reply
#3
newtekie1
Semi-Retired Folder
miahallen said:
The board was sent to me directly from Gigabyte HQ in TW...but it appeared to have been used previously.
Ah, ok. That makes sense, either way it is probably whoever used it before you that jacked it up, and not an issue with the hold down.

And, IIRC, the issue with 1156 was that Intel changed the design specs closely before launch because they realized the issue, and Foxconn shipped some parts that followed the old spec sheet. I doubt that happened with 1155 since the hold down spec sheat was the same as 1156.

HillBeast said:
Pics or it didn't happen.
Ok, post some screenshots of your own rig doing some benchmarks at your 24/7 clocks, and I'll post mine with the same benchmarks, we'll see who gets the higher scores. Then we can compare prices and see who paid less.:toast:

HillBeast said:
http://www.techpowerup.com/forums/showthread.php?t=118081 <- Um yeah. 5GHz...
So you think one suicide run, that wasn't even stable beyond getting a GPUz screenshot, is a valid come back to fact that a lot of SandyBridge chips can run 5.0GHz 24/7 totally stable on air?

HillBeast said:
Socket Pin Burnout Returns to Haunt LGA1155? - Pa... <- You're looking at it.
Since LGA2011 is going to use the same hold downs as 1155/1156, the chances of having the same problem, by your logic, will be the same. Of course, as we already went over this, the board was used and came to the reviewer like that, so we have no idea how much voltage was pushed through the stocket or even if the stocket was damaged prior to the pins being burnt.

Also, the problem with 1156 was fixed before it was even released, it was crappy Foxconn hold downs that caused the issue, not the socket or hold downs design themselves.


HillBeast said:
I am going to assume you never learnt about electronics before? The more contacts, the less resistance something has. It's common sense. If you have two pipes feeding a house with water, it's not going to be as good as three pipes. Two pipes will need to feed more water in and it is more strain on the pipes. Do I need to pull up Wikipedia to prove this?
While that is a good example, the LGA1366 socket only really has about 25 more power contacts compare to LGA1156/1155. And those pins are needed because 1366 processors consume more power. So while 2 pipes will easily fead the water needs of a 2 bathroom house, 3 are needed to feed a 3 bathroom house, because the 3 bathroom house uses more water.


HillBeast said:
But this is the bad press and bad press sticks. I don't recall seeing anywhere near as much bad press about 1366 arcing as 1156 or 1155 have, let alone at all. I'm probably wrong on that one though because Google can be unreliable.
You are correct, but press is nothing more than word, it doesn't matter in the real world use. You can say OMG Toyota sucks all over the press, that doesn't mean they are actually bad, and it definitely doesn't mean I'm running to trade in my Camery with 200,000 miles on it that is still running perfectly(while my Chevy with 70,000 has done nothing but break down).


HillBeast said:
Or people could stop being cheapskates, pay more and not whinge when it craps out because it is much higher quality.
I don't see many people whining about it crapping out. The only people that whined were the extreme overclockers that were effected by the arcing problem, it didn't happen in normal enthusiast usage, and those people likely got the hardware for free. So it is pretty hard to cheap out when you are getting the stuff for free. And of course quality isn't what you are paying more for on 1366, there are some pretty shitty 1366 boards out there that are far below the quality of some 1156 boards, you are paying for the artifical sense of extra performance that x16/x16 and Tri-Channel RAM gives you.
Posted on Reply
#4
[H]@RD5TUFF
Tartaros said:
Sometimes this things happens and everyone can fail. Quality controls are not perfect, that doesn't mean gb is a bad brand.
My personal experience of GB is that their QC is the worst I have ever found, as I had to purchase no less than 6 motherboards to find one that worked straight out of the box (4 were DOA, and 1 had a faulty cmos that wouldn't keep bios settings).

If these were samples this is people making a mountain out of mole hill, if these were legit consumer boards, then GB may be at fault. But 2 boards for 1 person a well known fatal flaw does not make.
Posted on Reply
#5
miahallen
[H]@RD5TUFF said:
If these were samples this is people making a mountain out of mole hill, if these were legit consumer boards, then GB may be at fault. But 2 boards for 1 person a well known fatal flaw does not make.
TPU mis-quoted me...it was only one board, not two....go to the source :rolleyes:
Posted on Reply
#6
cadaveca
My name is Dave
newtekie1 said:
So you think one suicide run, that wasn't even stable beyond getting a GPUz screenshot, is a valid come back to fact that a lot of SandyBridge chips can run 5.0GHz 24/7 totally stable on air?
5.0ghz on a single core only...keep that in mind. SandyBridge clocking is quite the interesting thing.


:toast:
Posted on Reply
#8
LAN_deRf_HA
HillBeast said:
LGA2011. Durr.

EDIT: Also you're bashing 1366 when your system has a 1366 CPU.
Hurg-derpa derrr derr. How is "LGA2011" a response to what I said? Never mind, I'm sure the logic will kill kittens.

For the record, I'm moving to 1155. It overclocks better and performs better clock for clock with nicer features and lower power consumption. Your whole mainstream el-cheapo shtick is a looney lie you tell yourself just so you don't have to feel inadequate about your rig for another year. As of right now there is no more mainstream vs highend sockets for intel. Just old and new and the less old but more unwanted (1156).

And here's your preferred comparison. Not that you want to see it in this case. http://www.anandtech.com/bench/Product/287?vs=142

Wins 67% of the time, and hits 5 GHz without spoiling like old milk. Going "mainstream" isn't "cheap", it's what you do if money doesn't grow on trees. Not that 1155 is really any cheaper than your normal 1366 rig to begin with so that entire aspect of your rage fit didn't make sense to begin with. Same with the durability crap, if anything SB can take more voltage than it's 1366 32nm counterparts.
Posted on Reply
#9
Wile E
Power User
[H]@RD5TUFF said:
My personal experience of GB is that their QC is the worst I have ever found, as I had to purchase no less than 6 motherboards to find one that worked straight out of the box (4 were DOA, and 1 had a faulty cmos that wouldn't keep bios settings).

If these were samples this is people making a mountain out of mole hill, if these were legit consumer boards, then GB may be at fault. But 2 boards for 1 person a well known fatal flaw does not make.
I have the exact opposite with GB. At least a dozen between myself and client computers, and not a single DOA or failure. I have your luck with Asus products, however.

I do agree this is mountain out of molehill material.
Posted on Reply
#10
1BadMoJoe
buggalugs said:
We already established the reviewer didnt even use the board. Gigabyte sent it like that.

Ya but because they were sample boards i'm guessing Gigabyte sent a couple of boards they were using and testing with and didnt realise the socket was burnt. I doubt a new board would have burnt sockets. Haha looks bad for Gigabyte though.
WELL...I haven't received a sample product from a manufacture...YET, I can only imagine if I did receive one I would treat it like a Ferrari, built to be driven HARD and FAST.
I am sure the fortunate few whom do receive an "engineering sample" have developed the skills necessary to create the identifiers to problems that should be addressed to revisions necessary prior to production.
Going from 1366 pin to 1156...well and trying to get more energy saving features and trying to interface video with higher performance than 1366 is a challenge.

Last time I checked, pin count always risen in count with new architecture not the other way around, looks like another Intel dead end product line.
Posted on Reply
#11
Wile E
Power User
1BadMoJoe said:
WELL...I haven't received a sample product from a manufacture...YET, I can only imagine if I did receive one I would treat it like a Ferrari, built to be driven HARD and FAST.
I am sure the fortunate few whom do receive an "engineering sample" have developed the skills necessary to create the identifiers to problems that should be addressed to revisions necessary prior to production.
Going from 1366 pin to 1156...well and trying to get more energy saving features and trying to interface video with higher performance than 1366 is a challenge.

Last time I checked, pin count always risen in count with new architecture not the other way around, looks like another Intel dead end product line.
Based on a single bad board?

Little extreme, don't you think?
Posted on Reply
#12
LAN_deRf_HA
For all we know gigabyte was using it to try and kill chips, dumping asinine volts into it. Could have just been your run of the mill suicide run. And even after killing a chip your first impulse may not be to go grab a magnifying glass and look at the socket pins. Hell maybe they had a hundred chips in there at 1.7-1.9 volts trying to find the limits. I'd bet they cracked 6 GHz for sure. Maybe they even went out of their way to screw up a socket and sent it out by mistake!

It's just absurd for anyone to assume this has any bearing on P67 durability.
Posted on Reply
#13
1BadMoJoe
Tick in the tack Intel release cycle, Sandy Bridge has excellent architecture features to build upon further product developments, I'm just not excited to hop in an soon to be unsupported 11xx CPU architecture migrating to 20xx pin.

How many years did socket seven and 775 last? 1366 didn't come out to long ago, now 1155/1156, soon to introduced 20xx pin?

Intel Dead End products lines based on more than a single board architecture, so disposable money minded end-users can upgrade continuously in the wash cycle.

miahallen said:
The board was sent to me directly from Gigabyte HQ in TW...but it appeared to have been used previously.
I'll try to stay on topic and go to the source, the lucky few whom have the skills to identify problems that need to be addressed prior to mainstream production. As it may appear that the Foxconn supplier may have had a new socket product interface error than may needed to looked into and revised for what ever reasoning dreamed necessary.

I do not see a fewer than 1155/1156 pin count in the next pin cycle and I also dought the majority of big business that still using reliable proven 775 hardware migrating to 1366 last year to 1156 this month only to repurchase next month to 1155 to anticipate for-site upon revamping to 20xx next year realizing the dead end architecture so to be abandoned?
Posted on Reply
#14
Wile E
Power User
1BadMoJoe said:
Tick in the tack Intel release cycle, Sandy Bridge has excellent architecture features to build upon further product developments, I'm just not excited to hop in an soon to be unsupported 11xx CPU architecture migrating to 20xx pin.

How many years did socket seven and 775 last? 1366 didn't come out to long ago, now 1155/1156, soon to introduced 20xx pin?

Intel Dead End products lines based on more than a single board architecture, so disposable money minded end-users can upgrade continuously in the wash cycle.



I'll try to stay on topic and go to the source, the lucky few whom have the skills to identify problems that need to be addressed prior to mainstream production. As it may appear that the Foxconn supplier may have had a new socket product interface error than may needed to looked into and revised for what ever reasoning dreamed necessary.

I do not see a fewer than 1155/1156 pin count in the next pin cycle and I also dought the majority of big business that still using reliable proven 775 hardware migrating to 1366 last year to 1156 this month only to repurchase next month to 1155 to anticipate for-site upon revamping to 20xx next year realizing the dead end architecture so to be abandoned?
And what does any of that have to do with a single faulty board, not even a faulty model, just a faulty board?

PS: I'm fine with buying both a new board and cpu if it provides the advances I'm looking for. If I have the money for LGA2011, and it does what I want, I'll buy it. I'll not limit myself over some false sense of wrong doing on Intel's part. Nobody FORCES you to upgrade at all. I simply buy the best performing product for the money I am willing to spend.
Posted on Reply
#15
R_1
I had similar experience with GA965 board. After removing CPU from the socket, few pins in different locations don't rise themselfs to the upper position and stay bended, but there was no burnout and can be mechanically fixed. What I think is that pins were made from bad high-temperature copper alloy, that looses spring properties at high temperatures. It is easy to imagine how sparks can appear when lots of amps went through this bad electrical connection.
Posted on Reply
Add your own comment