Lost a ram channel

[mackerel]

Thought I'd test the GTX 260 in my 5820k system as it was nearest, and noticed it had HT off from something earlier. No problem, went in bios to turn it back on, when I noticed only 3 channels were detected. Weird. In short, reset to basic speeds, tried reseating, and swapping modules around. One of the ram slots seems to be not operational currently. Took off waterblock, jiggled CPU a bit, didn't make any difference. Guess I'll have to pull the mobo out to see if there's anything else going on.

Oh, this was working fine until I did a case transplant recently. This is annoying... I also intend to use this system for a 3x crossfire run at some point, and I can't do it in case anyway as there's not enough room for the 3rd card. So I would have had to extract it anyway. Still annoying. I'll leave that for later, for now testing the '260 in another system which does seem to be working normally.

 

[Bill Dimwit]

I had similar problems with x58 systems. Etch time it was a bad cpu. And 3 of the 4 systems had a burnt contact on the cpu.
Try a second cpu if you have one.

 

[Johan45]

Hopefully, you get it sorted out but that was a fairly common issue on X99, try CMOS, battery out new BIOS flash, reseat CPU etc.. if that doesn't help then it's likely the board

 

[mackerel]

I had similar problems with x58 systems. Etch time it was a bad cpu. And 3 of the 4 systems had a burnt contract on the cpu.
Try a second cpu if you have one.

I have a Xeon spare that I got originally for the system. No OC fun there, hence the 5820k now they're more affordable.

Hopefully, you get it sorted out but that was a fairly common issue on X99, try CMOS, battery out new BIOS flash, reseat CPU etc.. if that doesn't help then it's likely the board

All on the to do list... no cheap X99 boards left around though Could use a 2nd anyway.

 

[Dr. McCoy]

Have the same problem with my 8320, it lost a RAM channel sometime back and confirmed it by switching out CPU's.
I'd try a different CPU first (If you have another one of that socket) since that's the simplest way to determine if it's the chip or the board.

 

[wingman99]

+1 I have seen this happen with other folks in the forums and it was the CPU.

 

[DaveB]

Could be the socket too. Check for a bent pin.

 

[Woomack]

On X58/79/99 and Z8x/9x/100 chipsets it's usually CPU socket issue. I had couple of motherboards in RMA for that. ASUS was replacing whole sockets in their ROG boards and even when pins were burned they were not asking any questions. Sometimes memory slots lose contact because of dust or something else but since you were checking memory in all slots and replacing them then I doubt it's the case.
Can check on Xeon and just look for any burned pins or pads on the CPU. The only issue related to CPU that I had was because it had burned pad. After cleaning it with high% alcohol it was working fine.

 

[mackerel]

Iv'e extracted the mobo and will test it shortly. Need to get it working if I'm going to attempt the 3-way crossfire for a last minute entry into the monthly challenge.

 

[mackerel]

Took CPU out, no visible damage to pins, no visible dirt or whatever on CPU pads. Put it back together, powered it up open bench... all channels appeared first time. So, either something came loose giving a bad contact, that the reseat fixed, or there's something about the case I put it in that cause it not to work. Like, either a short (nothing obvious on 1st look) or maybe due to tolerances it was bending the mobo in some way.

Right, time for GPU benching...

 

[Woomack]

I guess it was some bad contact. If it was a short in some place then it would behave in a different way. I hope all will be fine now. I remember a lot of issues with memory channels on older boards. So far not even one issue on X299.

 

[UltraTaco]

I would keep booting with 1-2 lost channels if I overclock memory part too much(triple channel x58). Lowering overclock brings everything back to normal of course.

 

[WhitehawkEQ]

I had similar problems with x58 systems. Etch time it was a bad cpu. And 3 of the 4 systems had a burnt contract on the cpu.
Try a second cpu if you have one.

I just have to say something about this, CPU's have CONTRACTS? or CONTACTS? I know what you meant

 

[freeagent]

A tight fitting cpu cooler helps. I used to see that quite a bit with my old h100, I have a Thermalright Le Grand Macho RT on my X5690 and I haven't seen a channel drop since. Although I did see some discoloration on some of the pads, it wiped off with isopropyl alcohol. I suppose the real test would be to populate all six dimms, but I'm not looking to stress myself out just yet. But to be fair, when I ran a Gulftown I had no problems with six dims, with my Xeon.. it gets cranky. It is an E.S. so maybe its just old, used, and abused.

 

[trents]

Yes, but if the HSF is too tight it can cause exactly the same issue, especially on some of the newer generations of Intel CPUs with that thin PCB.

 

[freeagent]

Yes, but if the HSF is too tight it can cause exactly the same issue, especially on some of the newer generation of Intel CPU with that thin PCB.

I hear you, its just my h100 wasn't as tight as it could have been.. as for newer cpus, the cooler comes with a shim so you don't crush your ihs

Edit:

The TS140P also comes with a shim