Puzzling RAM or motherboard problem
A couple of days ago my PC rebooted itself twice in less than 24 hours, so I knew something was wrong. When looking at the logs in the Event Viewer, it seemed that they had been caused by RAM issues, so yesterday I bought Memtest86 Pro and started running it.
The first test showed something like 638 errors, all of them in slot A2, which in my motherboard is the slot that you put the DIMM if you only have one. In my case, I have four DIMMs from the same kit, the CORSAIR Vengeance DDR5 RAM 192GB (4x48GB) 5200MHz CL38 (CMK192GX5M4B5200C3
, which is the only kit in the HCL for my motherboard, the Asus ROG STRIX Z790-E GAMING WIFI II.
Today I took out the PC's cover, and it was extremely dirty, so I pulled it out of the room and blew the hell out of it with a real blower, not a can one. After that, I took out the DIMMs, blew the slots, and put the DIMMs back but at random places. Up to this point, I was sure that one of the DIMMs was defective, in fact I already started an RMA with Corsair. However, when I put back the DIMMs, making sure that I didn't put the same one in slot A2 (the second slot from the CPU). I ran Memtest86 Pro again, and this time it showed far less errors, depending on the settings that I would try in the BIOS, which were generally rather conservative, but applying the RAM's XMP profile.
However, this time the errors were all in slot A2 as well. That made it obvious that it wasn't the RAM, but rather the motherboard. It really pissed me off, because I've been building PCs for over 20 years for myself and others, and at least the ones I know of, work perfectly fine and don't have any RAM problems. Those had much cheaper motherboards, in some case really cheap ones, and still work fine. But when I built this machine in 2025 I wanted something really good, and after looking at what was available, I went for this one, which wasn't exactly cheap.
So it kept giving errors in the A2 slot, even when I swapped the DIMMs again and put a third one in that slot. So I decided to test the slot one DIMM at a time, and since A2 is the slot that according to the manual is the slot to use first when you have only one DIMM, I figured it was a good chance to test each and every DIMM to confirm that indeed it was the slot and not any one DIMM.
Well, all four DIMMs passed when tested one at a time over the course of hours. So now I'm really confused, because when all four DIMMs are in place, there's always some errors. In one test they were just 4, in another over 20. But there's always more than one error. Meaning, it fails. But when testing each DIMM in the slot that Memtest86 Pro showed me that it was the only one with errors, they all passed. Granted, it was just one run, but if the RAM is defective, one test is all it takes.
And if the slot is defective, then testing four different modules in that slot, should also show me an error or more. So this is very puzzling to me. Any suggestions as to what may be will be appreciated. I need to figure out very soon if I need to RMA the motherboard or the RAM, and right now I don't know which. I added another stick to slot B2, which is the one to use if you use two DIMMs according to the manual.
The first test showed something like 638 errors, all of them in slot A2, which in my motherboard is the slot that you put the DIMM if you only have one. In my case, I have four DIMMs from the same kit, the CORSAIR Vengeance DDR5 RAM 192GB (4x48GB) 5200MHz CL38 (CMK192GX5M4B5200C3
, which is the only kit in the HCL for my motherboard, the Asus ROG STRIX Z790-E GAMING WIFI II. Today I took out the PC's cover, and it was extremely dirty, so I pulled it out of the room and blew the hell out of it with a real blower, not a can one. After that, I took out the DIMMs, blew the slots, and put the DIMMs back but at random places. Up to this point, I was sure that one of the DIMMs was defective, in fact I already started an RMA with Corsair. However, when I put back the DIMMs, making sure that I didn't put the same one in slot A2 (the second slot from the CPU). I ran Memtest86 Pro again, and this time it showed far less errors, depending on the settings that I would try in the BIOS, which were generally rather conservative, but applying the RAM's XMP profile.
However, this time the errors were all in slot A2 as well. That made it obvious that it wasn't the RAM, but rather the motherboard. It really pissed me off, because I've been building PCs for over 20 years for myself and others, and at least the ones I know of, work perfectly fine and don't have any RAM problems. Those had much cheaper motherboards, in some case really cheap ones, and still work fine. But when I built this machine in 2025 I wanted something really good, and after looking at what was available, I went for this one, which wasn't exactly cheap.
So it kept giving errors in the A2 slot, even when I swapped the DIMMs again and put a third one in that slot. So I decided to test the slot one DIMM at a time, and since A2 is the slot that according to the manual is the slot to use first when you have only one DIMM, I figured it was a good chance to test each and every DIMM to confirm that indeed it was the slot and not any one DIMM.
Well, all four DIMMs passed when tested one at a time over the course of hours. So now I'm really confused, because when all four DIMMs are in place, there's always some errors. In one test they were just 4, in another over 20. But there's always more than one error. Meaning, it fails. But when testing each DIMM in the slot that Memtest86 Pro showed me that it was the only one with errors, they all passed. Granted, it was just one run, but if the RAM is defective, one test is all it takes.
And if the slot is defective, then testing four different modules in that slot, should also show me an error or more. So this is very puzzling to me. Any suggestions as to what may be will be appreciated. I need to figure out very soon if I need to RMA the motherboard or the RAM, and right now I don't know which. I added another stick to slot B2, which is the one to use if you use two DIMMs according to the manual.
Comment