Bad News For LGA-1156 Foxconn Socket Users

[Zertz]

Gigabyte P55A boards available (UD4P,UD6), from the pictures they are using the Lotes socket

Yes, they have indeed switched to Lotes.

[RJR]

Well they will get this straightened out one day, but it's getting funny to me now. Lots of people are freaking out because they THINK their cpu is not making contact with the socket. They are noticing the very sharp round holes indented in the cpu pads that are not on every pad and stating their cpu is not making contact, when in fact the socket doesn't have any pins to make marks like this. The scratches in the center of the pins are where the socket is making contact and can be seen from these pics, http://www.xtremesystems.org/forums/showpo...p;postcount=362 that even at the location with no "pin marks" there is socket contact.

 

I guess i'll have to take out my processor and take a few pics and then mark the pads with a marker or something, reinstall it and see what the contact looks like after that. Should be interesting and put my mind at ease if in fact this is the case.

[Waco]

Well they will get this straightened out one day, but it's getting funny to me now. Lots of people are freaking out because they THINK their cpu is not making contact with the socket. They are noticing the very sharp round holes indented in the cpu pads that are not on every pad and stating their cpu is not making contact, when in fact the socket doesn't have any pins to make marks like this. The scratches in the center of the pins are where the socket is making contact and can be seen from these pics, http://www.xtremesystems.org/forums/showpo...p;postcount=362 that even at the location with no "pin marks" there is socket contact.

 

I guess i'll have to take out my processor and take a few pics and then mark the pads with a marker or something, reinstall it and see what the contact looks like after that. Should be interesting and put my mind at ease if in fact this is the case.

The point is that the socket is not putting proper pressure on the connections and that WILL cause problems over time (especially on power/ground pads). It's worse on the 1156 platform because there are fewer power and ground pins so the current through each is a lot higher...

Edited November 11, 2009 by Waco

[RJR]

The point is that the socket is not putting proper pressure on the connections and that WILL cause problems over time (especially on power/ground pads). It's worse on the 1156 platform because there are fewer power and ground pins so the current through each is a lot higher...

 

Well, my old 4 megapixel camera just won't give me a picture good enough to post, oh well. No visible problems yet

 

Waco, is your statement fact or your opinion? I haven't come across any viable source that states all Intel processors running in the Foxconn 1156 socket WILL have problems over time. The motherboard manufactures and Intel would be looking at one hell of a lot of returns if that did happen.

[Waco]

Waco, is your statement fact or your opinion? I haven't come across any viable source that states all Intel processors running in the Foxconn 1156 socket WILL have problems over time. The motherboard manufactures and Intel would be looking at one hell of a lot of returns if that did happen.

A bad contact will worsen over time or under heavy load - as that happens I would put VERY good money on the fact that we'll see a lot of overclocked or heavily loaded 1156 CPUs/mobos die in the next few years because of it.

 

You won't find a source saying that it'll happen to all of them (especially not from the manufacturers). Look at it this way - if you install a power supply that can't make it's rated load properly it might work for a while (especially if you don't load it down) but as soon as you stress it for a period of time it will pop. The same thing is going to end up happening to these sockets that don't put the proper pressure on the pads (thus raising resistance and by proxy, heat).

 

If it were improper pressure on data pins only I don't think I'd care too horribly much (though it would definitely limit overclocking) but having loose connections for power and ground is inexcusable.

Edited November 11, 2009 by Waco

[RJR]

The same thing is going to end up happening to these sockets that don't put the proper pressure on the pads (thus raising resistance and by proxy, heat).

 

OK, so if your theory is correct, and heat will be the killer, a simple visual check of the suspect socket/cpu over time will give you advanced warning (darkening pads) so you can hurry up and sell the computer as long as your not doing anything too extreme with it that is. I guess i'll just have to take it apart once in a while and have a look to make sure it's still OK. It's back running at 4.0 GHz, doing it's vmware/smp and gpu folding after todays teardown.

 

Any idea of how many vss/vcc pads are required for normal operation? I know that even some of the burned processors that lost some of their pads were still able to function, just not an electrical engineer so I don't know if Intel designs their chips with any safe guards like that. I guess the more important question would be what is the maximum amperage per pad that the cpu can handle? Geeze I love this stuff, too bad I'm too old to go back to school and become an engineer.

Edited November 11, 2009 by RJR

[Waco]

I have no idea on the amperage limit...I think the stock pads for the 1366 platform are rated for 1 amp...no clue on the 1156 socket.

[RJR]

I have no idea on the amperage limit...I think the stock pads for the 1366 platform are rated for 1 amp...no clue on the 1156 socket.

 

Another crazy question, before I jump to any conclusions of my own here, are all the vcc (and vss on their side) pads ganged (or whatever the proper terminology is) together in the cpu creating a single source or are their multiple end uses for these? It would seem to me that if all these pads are supplying a single source or even multiple depending on demand and the remaining pads have more than adequate amperage capacity to run the cpu at the desired frequency, even losing a couple wouldn't mean that much. Once again, NOT an electrical engineer just spit balling here.

[Waco]

Another crazy question, before I jump to any conclusions of my own here, are all the vcc (and vss on their side) pads ganged (or whatever the proper terminology is) together in the cpu creating a single source or are their multiple end uses for these? It would seem to me that if all these pads are supplying a single source or even multiple depending on demand and the remaining pads have more than adequate amperage capacity to run the cpu at the desired frequency, even losing a couple wouldn't mean that much. Once again, NOT an electrical engineer just spit balling here.

I honestly don't know for sure but I certainly wouldn't expect them to be completely tied together simply because of space, leakage, and heat concerns. I'd be extremely surprised if they did...

 

The loss of a certain pad (or pads) may very well cripple the entire processor if those pads were the only sources of power for that section of the chip.

Edited November 11, 2009 by Waco

[RJR]

I honestly don't know for sure but I certainly wouldn't expect them to be completely tied together simply because of space, leakage, and heat concerns. I'd be extremely surprised if they did...

 

The loss of a certain pad (or pads) may very well cripple the entire processor if those pads were the only sources of power for that section of the chip.

 

Thanks for all the info Waco I do appreciate it.

 

It seems that all the extreme OCing burns have been in one general area of the chip/socket:

and since the Intel spec sheets are very vague:

any idea on how a person would go about determining the exact destination of any given pad(s)?

I know i'm getting carried away with this, but it's fun for me.

[Zertz]

any idea on how a person would go about determining the exact destination of any given pad(s)?

If it's not in the whitepaper you linked, I don't think it's possible to figure out what pad feeds what part. However, I think it's safe to assume that the burns come from the core. The uncore (IMC, cache), PCI-Express controller and other logic most likely don't draw that much power.

[RJR]

If it's not in the whitepaper you linked, I don't think it's possible to figure out what pad feeds what part. However, I think it's safe to assume that the burns come from the core. The uncore (IMC, cache), PCI-Express controller and other logic most likely don't draw that much power.

 

Thanks Zertz, agree with you on that.

 

I just finished an e-mail to Intel to see if they would shed any light on my quest here, I would be shocked if I even receive a response but I'll have to wait and see.

 

Also, love your signature, if you knew how many times I've wanted to..............I better stop that train of thought, shes my wife. Thanks again.