X299 RAM Problem

[DogKnight]

Hello everyone,

 

I have a rather unusual problem that has got me stumped in regard to what's happening and how to fix it. 

This may be a little bit of a long post as I will advise the troubleshooting steps i've taken in order to eliminate more obvious options. 

 

Today I attempted to swap RAM kits around between two of my PC's. One is a X299 system and the other is a AM4 system. 

In the X299 system, I had Corsair LPX 64gb (3200mts, C16, 4 x 16gb sticks). 

In the AM4 system, I had Corsair Vengeance RGB Pro 32gb (3200mts, C16, 4 x 8gb sticks). 

The reason for the swap is not really important, but is was because the AM4 system has a Noctua NH-D15 on it. The Vengeance RAM is very tall and it forces me to mount the second CPU fan higher up on the heatsink in order to clear it. By putting in the LPX kit, it would allow me to lower the CPU fan. Might help with airflow, would definitely help with aesthetics. 

 

When I made the swap, the Ryzen system booted fine. Showed all four sticks of RAM and was running at the XMP profile of 3200 without any problem. 

However, when I put the Vengeance kit in the Intel system, it would not post. The fans spin up, some RGB lights up, then it resets. Not even able to get to the BIOS. Continues to boot loop indefinitely. 

Figured there must be some incompatibility, so I switched the RAM back to their original PC's.

Ryzen system booted just fine, correct RAM, XMP values, etc. Just as it was before.

But now the Intel system was boot looping with its original RAM in it. 

 

Now I have gone through a bunch of troubleshooting steps and I will list them all out below. Alas, none of this has worked and I am out of ideas. 

 

- Important to point out, whenever I changed RAM around in the X299 system, I powered the system off, removed the power cable, and cleared the CMOS by bridging jumpers on the motherboard. This was working as when I got the system to boot under other configurations, I would receive a pop up when entering the BIOS to advise it was completely reset. 

- Have also been referencing the motherboard manual to make sure I was using the correct RAM slots based off the amount of DIMMS installed. 

- At multiple stages I left the system boot looping for 5-10 minutes in case it was trying to identify the RAM. Made no difference. 

 

1) Tested to make sure RAM was seated correctly. Also used compressed air to clear out the RAM slots and on the RAM DIMMS as well. 

 

2) Tested a spare RAM kit. This was a Crucial 16gb kit (2133mts, C15, 2 x 8gb sticks). The system booted correctly. RAM recognized correctly also. 

 

4) Tested another spare RAM kit. This was a Crucial 16gb kit (2133mts, 4 x 4gb sticks). Booted correctly. RAM showed correctly. 

 

5) Tested with just 2 sticks of the original LPX RAM. This was boot looping still. 

 

6) To be completely sure the BIOS was being reset, disconnected everything and removed the CMOS battery. Left the PC alone for 15 minutes to make sure all power was drained. Also used the motherboard reset. Put in original kit, still boot looping. 

 

So I am running out of options. 

- I can confirm the RAM works in the Ryzen system

- I can confirm all the RAM slots work with an old kit of RAM

- I have made sure all previous settings were cleared

- The only thing I can think of is whether the memory controller is starting to fail, but I do not have a spare X299 motherboard or CPU to test this scenario. 

 

The specs of the Intel system with the problem are:

- Intel i9 10980xe

- Gigabyte X299 Aorus Ultra Gaming

- Corsair H150i Pro XT - 6 x 120mm fans. 

- EVGA 2070 Super

- Samsung 2TB NVME (boot drive) - 970 Evo I think

- Couple of SATA SSD's and one mechanical HDD. 

- Additional PCIE cards, TP Link Wireless / BT card, 10gbe NIC

- Corsair HX850i PSU

The motherboard I have had for a few years now (4 maybe?). The CPU and RAM kit have been in the PC for a bit over a year and a half and I haven't had any issues before now.

 

Sorry this is so long, but wanted to make sure I provided all the information I could. 

Please let me know if you have any questions or suggestions. 

Cheers. 

[Shimejii]

Sounds like a classic x99 / x299 board issue. My x99 motherboard simply died after a plugged in a USB Wifi card after about 4 years of use. Was quite frustrating honestly, i spent 7 hours trying pretty much everything, and nothing fixed it. 

 

Honestly just sounds like if you can get into the bios, maybe do a full bios update, then try the ram again. If you are on the latest version, maybe just reflash it for shits and giggles.

 

HEDT platforms tend to be pretty sparse for these kinds of bugs and issues. Its pretty much you have to have spare parts on hand just to see, and at least you have extra ram kits that do work.

[DogKnight]

Thanks for the suggestions. Will try a BIOS reflash to see if it helps.

 

I hope it isn't starting to fail. My options would be to use 2133 RAM with a 10980xe, or replace the motherboard. And X299 boards are more expensive now than they used to be, 

 

 

[DogKnight]

I have now done some further troubleshooting and testing.

 

7) I reflashed the BIOS with the same version in case it was corrupt. No change. 

8) I flashed the newest BIOS. It was only released in Dec to address security concerns. No difference. 

9) Have tried another RAM kit. Corsair Vengeance LPX 2666 64gb (4 x 16gb). I only have tried two of the DIMMS currently, but it boots fine.

 

I looked further into the differences of the RAM sticks. The only detail I can find that may be a cause is the voltages.

The two 3200 kits use 1.35v. Whereas all three of the other kits (both kits of 2133 and the 2666 kit) run at 1.2v.

 

Now with the 3200 kits, they are only listed as requiring 1.35v when XMP is enabled. They are  supposed to only require 1.2v at jedec base spec (2133). It seem's though XMP must be auto-enabling itself somehow as even after a CMOS reset, the same problem is occuring.

 

I don't know if the voltage is definitely the cause of the issue, but its the only theory I have. I am not sure whether I can save higher voltage settings in the BIOS, the shutdown and put the other RAM in?

 

If there is anyone in Melbourne, Australia that has a spare X299 board I can borrow for testing with, please reach out. 

 

Any other suggestions are definitely welcome. 

[DogKnight]

Additional testing.

 

10) Manually set the RAM voltage in BIOS to 1.35v as required by the 3200 kits. Changed this for both regular run levels and memory training levels. No changes.

 

For now, have put my 2666 kit in, which runs an XMP profile just fine. 
 

Does anyone know if this issue is likely to be the motherboard or the CPU? 

I assume its the memory controller. But outside of it being called a memory controller and its task of controlling memory, I know very little about it. 

My understanding is that it is part of the CPU. Just trying to work out which parts I should try get my hands on for testing. 

[djbuddha]

I’m having an identical issue on almost identical hardware with my ASUS X299 Sage/10G

 

This machine ran flawless before, but in there I’m running OEM Samsung DDR4 at 2666 and it’s stable.

 

I have 8x16 sticks of Corsair Vengeance LPX which ran fine with XMP and all the sticks installed.  One day, it started boot looping.  Then on an unrelated matter- it stopped boot looping but was spitting code 62 and the VGA light was on (turns out my Vega 64 is having VBIOS issues so literally last night, I got a 6900XT and that solved that issue)

 

So if I insert any of the RAM in any combination/ slot - the result varies in boot looping or loading the OS but crashing as soon as I hit it with a load.  I’m wondering if my memory controller is gone or if the board is done.  Luckily, the 10980XE I’m using and the board still have warranty.

 

With the Samsung RAM it’s flawless right now minus the lack of RAM I’m using and obviously RAM speed decrease which I do notice on my video exports.  The Corsair like yourself, ran fine for 2+ years.

 

My specs:

 

Asus X299 Sage/10G

128 GB DDR 4 Corsair Vengeance LPX (not installed)

(16 GB DDR 4 Samsung OEM running at 2666 w/ auto settings installed currently)

Sapphire Toxic Air Cooled AMD Radeon 6900XT

Intel 10980XE

Corsair h150i Pro w 3 Noctua fans

2 x 2 TB Samsung 970 Pro NVME

2 x 8 TB WD Red

Gigabyte Thunderbolt 3 card (firmware modded w jumper for Hackintosh- yes I’m running Monterey on this rig)

Fenvi PCIE BT/Wi-Fi adapter (for Airdrop / Continuity etc)

Seasonic 1300W Platinum

I have a 9900X I’m going to drop in over the weekend to determine if it’s board or processor.  I removed it back in 2020 to install the 10980XE so I know it works correctly.  This situation is driving me insane though.

[DogKnight]

On 5/5/2022 at 3:07 AM, djbuddha said:

I have a 9900X I’m going to drop in over the weekend to determine if it’s board or processor.  I removed it back in 2020 to install the 10980XE so I know it works correctly.  This situation is driving me insane though.

Will be interested to hear the outcome of this. I got rid of my previous processor (7820x), so have not been able to test that aspect. 

 

At this point i've essentially hit a dead end. I have the 64gb of Corsair LPX 2666 running in that PC just fine at the moment. Not ideal, but it works at least. 
 

I suspect it is down to the memory controller and the voltage required for higher clocked RAM kits. But will never know unless I invest in or gain access to alternate motherboards or CPU's. 

 

Please share your results. 

[djbuddha]

On 5/5/2022 at 11:30 PM, DogKnight said:

Will be interested to hear the outcome of this. I got rid of my previous processor (7820x), so have not been able to test that aspect. 

 

At this point i've essentially hit a dead end. I have the 64gb of Corsair LPX 2666 running in that PC just fine at the moment. Not ideal, but it works at least. 
 

I suspect it is down to the memory controller and the voltage required for higher clocked RAM kits. But will never know unless I invest in or gain access to alternate motherboards or CPU's. 

 

Please share your results. 

Last night I dropped in my 9900X.

 

No meaningful change sadly.  My ASUS x299 Sage/10G seems to work fine with the Samsung RAM but I have warranty until October.

 

I have a feeling a pin on the board is bent.  It’s LGA 2066 and I felt like my eyes are playing tricks on me- because something looked wrong at first but then I stared at it for an hour and it looked perfectly fine.  I put the 10980XE back in and with my Corsair RAM (all 8 sticks at 128 GB) w/ XMP enabled would not post (stuck on B7).  At auto and 2133 it gets into the OS- but opening any window caused a reboot.  It wasn’t even getting that far before.  Again, this RAM worked for 2 years and even trying a whole brand new set of Kingston RAM and it behaved the same as my Corsair which rules out RAM being the cause vs motherboard.

 

Put the generic Samsung RAM back in and everything works at auto 2666). Exported a 1 hour 4K/23.98 timeline in Final Cut without pre rendering and it did the job in about 11 minutes.  I re-ran it about 10 times and it was fine.  Going to RMA it this week (ASUS is close to me- hopefully they let me drive it over as they have with other friends because I don’t trust the shipping company here at all).  If it is a socket issue and they discover a bent / broken pin, I’ll happily pay the fee to replace the socket as they won’t cover that under warranty (the board was just over $1000 CAD and I’m committed to keeping this machine alive for at least another 5 years minimum)

[DogKnight]

Very frustrating. Surprised the CPU swap didn't change anything. It does point to the motherboard though. I have had the same board for a bit over 4 years, so well outside of any warranty. 

Hopefully you have luck with the Asus RMA. Keen to see if they find anything. 

[djbuddha]

2 hours ago, DogKnight said:

Very frustrating. Surprised the CPU swap didn't change anything. It does point to the motherboard though. I have had the same board for a bit over 4 years, so well outside of any warranty. 

Hopefully you have luck with the Asus RMA. Keen to see if they find anything. 

I’m sending the board in tomorrow.  If there is a damaged pin on the socket like I said, I’ll pay the fee to fix that.  The rest of the board is covered under warranty- I just hope if they find something else when they fix / replace the board, they can tell me what else was wrong with it.  If I needed to use better VRM cooling for example; if that’s another cause of the issues I’m having.

 

They said it should be 15 days including shipping time (I guess with COVID protocol they can’t accept walk-in customers at their facility) so I’ll take my chances for 2 weeks.

[djbuddha]

On 5/8/2022 at 9:12 PM, DogKnight said:

Very frustrating. Surprised the CPU swap didn't change anything. It does point to the motherboard though. I have had the same board for a bit over 4 years, so well outside of any warranty. 

Hopefully you have luck with the Asus RMA. Keen to see if they find anything. 

Update- ASUS elected to replace the board so I should have it back and running tomorrow / Thursday.  Hopefully everything just works when I get it installed.  Wish me luck.

[DogKnight]

9 hours ago, djbuddha said:

Update- ASUS elected to replace the board so I should have it back and running tomorrow / Thursday.  Hopefully everything just works when I get it installed.  Wish me luck.

Great that they're replacing the board. Hope it fixes the issue. 
If it does, then I got to weigh up whether its worth paying for a new X299 board for the sake of running faster memory. I am leaning towards no though.

[djbuddha]

10 hours ago, DogKnight said:

Great that they're replacing the board. Hope it fixes the issue. 
If it does, then I got to weigh up whether its worth paying for a new X299 board for the sake of running faster memory. I am leaning towards no though.

After months of running the OEM Samsung RAM at 2666 with only 16 GB - having 128 GB at 2993 as per intel’s spec (possibly with XMP enabled) will be blissful

 

I mean I will attempt 3200 w/XMP - but I just want the RAM to work at all.

[DogKnight]

13 hours ago, djbuddha said:

having 128 GB at 2993 as per intel’s spec (possibly with XMP enabled) will be blissful

 

Absolutely. 64gb of 2666 won't be too bad. A few hundred $$ for a new motherboard to potentially push it to 2933 is a tough sell. If I knew I could buy a 3800 kit of 128gb that would work, it would be more tempting. But there is just no way to know without sinking the funds first. 

[djbuddha]

Ok so a couple of updates.

 

This issue I had with the random reboots started back in 2020 - but it didn't get bad until recently.

 

So - I install the board, and all my RAM - everything worked immediately.  So that's the positive.

 

The negative, it rebooted when I went to render a file (again, running a Hackintosh) - I realize that it was my PSU triggering some protection.  So I started pulling all my PCIE devices (including a USB card I have installed because heaven forbid the ASUS X299 Sage/10G have a USB 2.0 header on board) - and then..... everything just worked.

 

So now, it got me thinking.  Over a 2 year period, I think this card's power issue over a long span destroyed my board.  Getting replacement stuff from the time the pandemic started was a rough go.  But I can almost say .... it's working flawlessly now.  I exported about 20 projects today from FCPX with no issues, and I ran Unigine Heaven & Cinebench for 12 hours last night while streaming a screen capture of it from OBS to Twitch at 4K/60 to a test account I have.  Flawless and stable.

 

On a sidenote, can you have Stockholm syndrome because of computer issues?  I suddenly feel ‘incomplete’

[DogKnight]

Awesome news that it is working with the replacement board. Frustrating that it was a USB card in the end that was potentially causing the issue all this time. 

 

If you're feeling incomplete without an issue, upgrading the version of OSX you're running should do the trick pretty easily. 

I've run a Hackintosh in the past and it wasn't quite as easy as it had become these days. Any little update would usually break some kind of functionality. 

 

 

[djbuddha]

14 hours ago, DogKnight said:

Awesome news that it is working with the replacement board. Frustrating that it was a USB card in the end that was potentially causing the issue all this time. 

 

If you're feeling incomplete without an issue, upgrading the version of OSX you're running should do the trick pretty easily. 

I've run a Hackintosh in the past and it wasn't quite as easy as it had become these days. Any little update would usually break some kind of functionality. 

 

 

I’ll put it this way……. Carbon copy cloner, external SSDs, EFI folder juggling and Dropbox for super important things (like kexts, serial numbers etc) is clutch to keep things alive.  I actually have to do an update but I’m waiting LOL