Dead Intel Haswell CPU?

[dilacerated]

I need help from those of you whom are intimately familiar with Intel processors/motherboards (I've been all Cyrix/AMD since the early/mid-90's).

 

Earlier this year I upgraded my UTM/firewall appliance from an AMD SoC board w/an A6 processor to an Intel Core i5-4570T.  The motherboard I picked up for such is Biostar's H81MHV3 (Ver. 7.0).  The reason for this upgrade in 2019 was performance based.  Going the Haswell route was born out of the fact that my UTM is using the LTSS kernel (3.12.74...) for SLES 12 SP1...  Unfortunately there's no means to move to a newer kernel thus full CPU support past Haswell is a glaring issue.

 

So this weekend I disabled SpeedStep and C States in the BIOS of this Biostar board and upon saving the changes made/exiting the BIOS the system powers on/off repeatedly.  The culprit is anything in the PCIe 16x slot...  I've re-seated the CPU, tried different RAM, two PSU's, two PCIe 4x NIC's and a GPU in the 16x slot.  Anything in the slot = problem.  Nothing in the slot and the system posts fine.

 

That all being said is it possible for the System Agent to fry in a way where only the PCIe 16x slot would prohibit the system from posting?  I've never had a PCIe slot fail and have typically had 2+ motherboards running in my house since PCIe's inception.  I've had dozens of personal builds and have custom built several dozen servers and never once have I seen a PCIe slot go bad.  Motherboards from Tyan, Gigabyte, ECS, ASUS, ASRock, Biostar, MSI, DFI. and Supermicro.

 

I'm asking as since my UTM also acts as my families DHCP Server I ran out and got a Dell Optiplex with an i7-4790 from a guy off Craigslist and holy **** have OEM systems become even worse.  XMP not supported...  Proprietary PSU, fan headers, front panel connections, etc...  What a POS.  It covers my need for the moment but I'd rather transplant the CPU into my Biostar board if it is conceivable that the System Agent in the i5-4570T could meltdown in such a way where only the PCIe 16x slot is lost.

[Skiiwee29]

Since you were playing around in the BIOS, have you tried doing a clear CMOS? There should be some jumpers on the board you short out or you can remove the battery from the board for 30s to clear it as well. 

[dilacerated]

8 hours ago, Skiiwee29 said:

Since you were playing around in the BIOS, have you tried doing a clear CMOS? There should be some jumpers on the board you short out or you can remove the battery from the board for 30s to clear it as well. 

Sorry didn't mention it but yes I cleared the CMOS via the jumper.  After that didn't work I pulled the battery for 60 seconds.  That is troubleshooting 101.

[Allan B]

Way to call out someone with over 10K+ posts on this form.   I am fairly new to this forum but if you go on any forums you can't assume people have tried anything, he offered a suggestion that you didn't mention that you did..   Troubleshooting 101.. lol..

 

It looks like you have narrowed down the problem, so i am not sure why you are asking, but sure it is possible some PCIE lanes to stop working or even a memory channel (if CPU is dropped for example)..   It could be the PCIE slot itself (any damage or stuff in it), the communication to the CPU (CPU socket pins damaged?) or even the CPU itself, common sense 101..   Trying the other CPU is really your only next step

[dilacerated]

41 minutes ago, Allan B said:

Way to call out someone with over 10K+ posts on this form.   I am fairly new to this forum but if you go on any forums you can't assume people have tried anything, he offered a suggestion that you didn't mention that you did..   Troubleshooting 101.. lol..

 

It looks like you have narrowed down the problem, so i am not sure why you are asking, but sure it is possible some PCIE lanes to stop working or even a memory channel (if CPU is dropped for example)..   It could be the PCIE slot itself (any damage or stuff in it), the communication to the CPU (CPU socket pins damaged?) or even the CPU itself, common sense 101..   Trying the other CPU is really your only next step

Wasn't calling him out --> was just saying that's a 101 step and yes I know of it.

 

I get that most people that show up in forums are the layman but I am here talking about a custom built UTM/FW appliance running an enterprise Linux OS at the core.  Discussing a part of the CPU die that may have failed in a way that I personally don't have any experience with having been away from Intel CPU's for about 20 years.  Dropped that I have built countless systems from Desktops to Servers.  Based on those details being given I thought it would be expected that I would have tried the first thing anyone would try after making a BIOS change resulting in the system not posting.  It wasn't which didn't bother me --> I just simply pointed out that I have done 101.

[Skiiwee29]

3 hours ago, Allan B said:

Way to call out someone with over 10K+ posts on this form.   I am fairly new to this forum but if you go on any forums you can't assume people have tried anything, he offered a suggestion that you didn't mention that you did..   Troubleshooting 101.. lol..

 

It looks like you have narrowed down the problem, so i am not sure why you are asking, but sure it is possible some PCIE lanes to stop working or even a memory channel (if CPU is dropped for example)..   It could be the PCIE slot itself (any damage or stuff in it), the communication to the CPU (CPU socket pins damaged?) or even the CPU itself, common sense 101..   Trying the other CPU is really your only next step

I dont take it as a call out or anything. No need to assume something of a post. 

 

OP, have you checked to make sure none of your pins are bent in the socket?

[Mister Woof]

Edit nm reread it

[dilacerated]

37 minutes ago, Skiiwee29 said:

I dont take it as a call out or anything. No need to assume something of a post. 

 

OP, have you checked to make sure none of your pins are bent in the socket?

When I got the board (used via Amazon) there was one pin that looked off but I put the board under a desktop magnifying lens at my work and my coworkers and I felt it was OK.

 

I assembled the system and it has ran fine for 6-7 months until now.

[Skiiwee29]

1 minute ago, dilacerated said:

When I got the board (used via Amazon) there was one pin that looked off but I put the board under a desktop magnifying lens at my work and my coworkers and I felt it was OK.

 

I assembled the system and it has ran fine for 6-7 months until now.

So the 4790 you have in the Optiplex should work in that board in the h81 board. Have you thought about pulling that and putting it into the Biostar board and see if the behavior is replicated? I would think this would give definitive answers where the problem lies if the issue translates to a new CPU or not. 

[Mister Woof]

19 minutes ago, Skiiwee29 said:

So the 4790 you have in the Optiplex should work in that board in the h81 board. Have you thought about pulling that and putting it into the Biostar board and see if the behavior is replicated? I would think this would give definitive answers where the problem lies if the issue translates to a new CPU or not. 

The CPU is probably fine it's probably the motherboard has taken a shit.

[dilacerated]

52 minutes ago, Skiiwee29 said:

So the 4790 you have in the Optiplex should work in that board in the h81 board. Have you thought about pulling that and putting it into the Biostar board and see if the behavior is replicated? I would think this would give definitive answers where the problem lies if the issue translates to a new CPU or not. 

Yeah I know that CPU is compatible with this Biostar board.  Problem of course is with children when do I have the time so I was here seeking opinion(s) before I either take a day off on a school day or pull an all nighter to do so.

[dilacerated]

34 minutes ago, Plutosaurus said:

The CPU is probably fine it's probably the motherboard has taken a shit.

So the SA (System Agent) controls the PCIe 16x slot.  Only reason I turned up here is the knowledge that the SA is part of the CPU hence my bet that's the malfunctioning component.

 

If it weren't for the host of off putting issues with this Dell Optiplex I'd be fine with it but I am going to roll the dice and see what happens with the i7-4790 in the Biostar board as I would much rather my RAM run at its timings AND actually be able to use a real PSU than the tin can Dell calls a Power Supply.

[Skiiwee29]

22 minutes ago, dilacerated said:

Yeah I know that CPU is compatible with this Biostar board.  Problem of course is with children when do I have the time so I was here seeking opinion(s) before I either take a day off on a school day or pull an all nighter to do so.

I agree downtime can suck for sure, but sadly sometimes its unavoidable and necessary, and I feel in this instance, its necessary to get definitive answers as to where the problem lay. 

[dilacerated]

33 minutes ago, Skiiwee29 said:

I agree downtime can suck for sure, but sadly sometimes its unavoidable and necessary, and I feel in this instance, its necessary to get definitive answers as to where the problem lay. 

Agreed.  Will roll the dice as this Dell system makes me want to voluntarily initiate a protein spill.

[dilacerated]

Picked up an i7-4770K for a fair price.

 

PROBLEM SOURCE AS SUSPECTED: the i5-4570T processor.  With the i7-4770K the system posts and the PCIe 16x slot is perfectly fine (tested with an LSI controller and a GPU).

 

Going to transplant the i7-4790 into the Biostar board, the i7-4770K into the Dell, and get this OEM POS the hell out of my house.

 

Thanks Skiiwee29 and Plutosaurus for chiming in.

[dilacerated]

Welp turns out it was the Motherboard after all.  Upon powering up with the i7-4790 in the system the same issue occurred.  If left off for 24+ hours things are fine until the next power cycle...  Got a new board but will post about the situation with such in a new thread.