3900x, Ubuntu & 3600mhz ram.

/dev/null

[H]F Junkie
Joined
Mar 31, 2001
Messages
15,190
So I've had 2x16GB (single rank) ddr4-3000 working fine on my 3900x/crosshair VI for a while.

I've replaced this with 2x8G ddr4-3600 cas16 and about once every day or two a program crashes (seems to usually be Xorg) and I get this in the logs:

Code:
[22729.576327] mce: [Hardware Error]: Machine check events logged
[22729.576331] [Hardware Error]: Corrected error, no action required.
[22729.576336] [Hardware Error]: CPU:0 (17:71:0) MC27_STATUS[-|CE|MiscV|-|-|-|SyndV|-|-|-]: 0x982000000002080b
[22729.576339] [Hardware Error]: IPID: 0x0001002e00000500, Syndrome: 0x000000005a020001
[22729.576341] [Hardware Error]: Power, Interrupts, etc. Ext. Error Code: 2, Link Error.
[22729.576342] [Hardware Error]: cache level: L3/GEN, mem/io: IO, mem-tx: GEN, part-proc: SRC (no timeout)

So is it
1) Bad ram
2) Bad cpu
3) CPU just can't handle 3600mhz cas16?

Motherboard is an X370 board (crosshair VI). This has the T-topology layout so maybe 3600 w/ 2 sticks just doesn't work on this box? Possible?
I could swap these 2 sticks to my windows i5-8400 box I suppose....
 

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
58,053
You might want to increase your memory voltage. It's doubtful that anything's bad. These boards and CPU's are simply fickle when it comes to compatibility. What RAM is it?
 

/dev/null

[H]F Junkie
Joined
Mar 31, 2001
Messages
15,190
You might want to increase your memory voltage. It's doubtful that anything's bad. These boards and CPU's are simply fickle when it comes to compatibility. What RAM is it?

In the short term I've moved this to my i5-8400 box to see if anything pops up...If nothing pops up in a week or so of gaming/playing/load I am going to move this back to my 3900x box.
I was thinking maybe with t-top I'm being too optimistic on the speed I can reach...

Edit: Looks like there is a new bios out as well. I'm on 7704...Maybe the new AGESA code helps?

Ram is:
Crucial Ballistix 3600 MHz DDR4 DRAM Desktop Gaming Memory Kit 16GB (8GBx2) CL16 BL2K8G36C16U4W (White)
 
Last edited:

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
58,053
In the short term I've moved this to my i5-8400 box to see if anything pops up...If nothing pops up in a week or so of gaming/playing/load I am going to move this back to my 3900x box.
I was thinking maybe with t-top I'm being too optimistic on the speed I can reach...

Edit: Looks like there is a new bios out as well. I'm on 7704...Maybe the new AGESA code helps?

Ram is:
Crucial Ballistix 3600 MHz DDR4 DRAM Desktop Gaming Memory Kit 16GB (8GBx2) CL16 BL2K8G36C16U4W (White)

The new AGESA code might help you. It might also make things worse.
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
28,124
I'd definitely double check with memtest to make sure your memory is stable IN THAT SYSTEM.

I was thinking that your motherboard might be holding you back with the higher memory speed, but maybe not. That's one board I haven't used. I'd almost wager that all three parts, the motherboard, CPU, and RAM, are all good independently, but for some reason don't like working well together. The T-topology might have something to do with it.
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
28,124
Welcome to AMD Memory Compatibility Bingo(tm)

[may the odds be ever in your favor]

Lol...It's not THAT bad. It's not like it doesn't work at all. I bet with a 500 series motherboard it would work fine ;).

There must be something with the t-topology holding memory back because just about everyone switched to daisy chain in the 400/500 boards. I'm not an engineer though just a pragmatist.
 

IdiotInCharge

NVIDIA SHILL
Joined
Jun 13, 2003
Messages
14,710
OR he could just use ram that's listed in the mobo compatibility table. It's not rocket science.
You seen what memory availability looks like?

It's pretty shitty over here in the 'States. Very difficult to get your 'preferred' kit. I've been looking for 2x32GB 3600C16 kits to drop into my Intel system that would be probably kosher if I dropped them into an AMD system later, and well, they usually go for ~US$350+.
 

B00nie

[H]F Junkie
Joined
Nov 1, 2012
Messages
9,000
You seen what memory availability looks like?

It's pretty shitty over here in the 'States. Very difficult to get your 'preferred' kit. I've been looking for 2x32GB 3600C16 kits to drop into my Intel system that would be probably kosher if I dropped them into an AMD system later, and well, they usually go for ~US$350+.
That's hardly the point though. You have to follow the compatibility table or take responsibility for the problems. It's not AMDs problem if you can't use the same kit with a competitions product...
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
28,124
That's hardly the point though. You have to follow the compatibility table or take responsibility for the problems. It's not AMDs problem if you can't use the same kit with a competitions product...

It's not AMD's problem either way. It's on the board manufacturer. It doesn't take a genius to realize that they can't possibly test every set of memory that could possibly be available over the life of the board. But with essentially only three manufacturers of RAM ICCs, you can make pretty good inferences as to whether or not something will work based on which ICCs are in it.

I can't remember the last time I actually looked at the QVL. I never had a serious enough problem to warrant it.
 
Last edited:

B00nie

[H]F Junkie
Joined
Nov 1, 2012
Messages
9,000
It's not AMD's problem either way. It's on the board manufacturer. It doesn't take a genius to realize that they can't possibly test every set of memory that could possibly be available over the life of the board. But with essentially only three manufacturers of RAM ICCs, you can make pretty good inferences as to whether or not something will work based on which ICCs are in it.

I can't remember the last time I actually looked at the QVL. I never had a serious enough problem to warrant it.
Yet the topic is now...
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
28,124
Yet the topic is now...

Be my guest if you want to hunt down 3-year-old SKUs just so you can find something on the QVL list, especially for the OPs board. I just looked it up. The QVL has not been updated since 2017.

Do you also only buy SSDs off of the QVL or power supplies that have been tested with that motherboard? If you look on the CH6 support page, you can find the devices that ASUS notes are compatible. Once again though, a list from 2017.
 

B00nie

[H]F Junkie
Joined
Nov 1, 2012
Messages
9,000
Be my guest if you want to hunt down 3-year-old SKUs just so you can find something on the QVL list, especially for the OPs board. I just looked it up. The QVL has not been updated since 2017.

Do you also only buy SSDs off of the QVL or power supplies that have been tested with that motherboard? If you look on the CH6 support page, you can find the devices that ASUS notes are compatible. Once again though, a list from 2017.
Nice straw man there. I'm not the one having problems. Just buy junk if you want... But don't blame others if it doesn't work. It's not hard at all to find 3 year old components normally. Hardware development has become quite stagnant.
 

/dev/null

[H]F Junkie
Joined
Mar 31, 2001
Messages
15,190
So I've replaced my windows box ( intel i5-8400/Z370) box with a R5-3600 & an X570 board & moved the ram there. Will see if I get different results. I'm fine with the current box @ 4 * 8G ddr4-3200 cas16. Don't think it's on the QVL but haven't seen a single crash since I swapped.

I think QVLs are useful if you buy a board within say 6 mo. after it's released. Not really useful after that. I"m not blaming the board or amd or anything. Just trying to see if I can get it working. Technically no DDR4-3600 is "within spec" for ryzen AFAIK, but AMD still had it in some slides at one point that it was a best bang for buck.
 

/dev/null

[H]F Junkie
Joined
Mar 31, 2001
Messages
15,190
Lol...It's not THAT bad. It's not like it doesn't work at all. I bet with a 500 series motherboard it would work fine ;).

There must be something with the t-topology holding memory back because just about everyone switched to daisy chain in the 400/500 boards. I'm not an engineer though just a pragmatist.
These were my thoughts as well. I was actualy reading many t-top boards have trouble with hittig much more than 3433 or so...so it might just not be possible with that combo.
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
28,124
These were my thoughts as well. I was actualy reading many t-top boards have trouble with hittig much more than 3433 or so...so it might just not be possible with that combo.

Tighten the timings then. Should be pretty close in performance if you can get 3200 with Cas 14.
 

DEEM0N

n00b
Joined
Apr 28, 2015
Messages
16
Sorry for necroing, but what was the problem then finally? And solution?
Getting the same problem in log on my brand new build (few times a day only so far).
MSI Tomahawk X570 (with newest non-beta BIOS -- 7C84v14; there is newer beta also which should enable SAM, but ... don't risk with beta);
Ryzen 5950X;
4x8 3600MHz G.Skill 18CL (so nothing fancy; but with "Compatible with Ryzen" sticker; F4-3600C18D-16GTZRX; 2 of such kits then; with the stock A.XMP conf);
Linux Zen 5.10.0-1-MANJARO #1 SMP PREEMPT Wed Nov 18 07:17:08 UTC 2020 x86_64 GNU/Linux
linux-firmware 20201120.r1785.bc9cd0b-1

Code:
nov   24 01:23:57 Zen kernel: [Hardware Error]: Corrected error, no action required.
nov   24 01:23:57 Zen kernel: [Hardware Error]: CPU:0 (19:21:0) MC27_STATUS[-|CE|MiscV|-|-|-|SyndV|-|-|-]: 0x982000000002080b
nov   24 01:23:57 Zen kernel: [Hardware Error]: IPID: 0x0001002e00000500, Syndrome: 0x000000005a020001
nov   24 01:23:57 Zen kernel: [Hardware Error]: Power, Interrupts, etc. Ext. Error Code: 2, Link Error.
nov   24 01:23:57 Zen kernel: [Hardware Error]: cache level: L3/GEN, mem/io: IO, mem-tx: GEN, part-proc: SRC (no timeout)
1 time in first 48h of operation.
Panicing a little, googling, found this thread here and another for EPYC processors.

Reading here suggestion to rise mem voltage, upped from 1.37 (auto setting; I assume from 1.35 real results slightly bigger) to 1.40 (but bios shows real voltage 1.42) and got almost immiadetly in next hour 2 more those errors (and second one is marginally different also... there is one "Over" instead of previous two "-" in status and the last long number there is different also:
Code:
nov   24 05:01:00 Zen kernel: [Hardware Error]: Corrected error, no action required.
nov   24 05:01:00 Zen kernel: [Hardware Error]: CPU:0 (19:21:0) MC27_STATUS[-|CE|MiscV|-|-|-|SyndV|-|-|-]: 0x982000000002080b
nov   24 05:01:00 Zen kernel: [Hardware Error]: IPID: 0x0001002e00000500, Syndrome: 0x000000005a020001
nov   24 05:01:00 Zen kernel: [Hardware Error]: Power, Interrupts, etc. Ext. Error Code: 2, Link Error.
nov   24 05:01:00 Zen kernel: [Hardware Error]: cache level: L3/GEN, mem/io: IO, mem-tx: GEN, part-proc: SRC (no timeout)
nov   24 05:37:38 Zen kernel: [Hardware Error]: Corrected error, no action required.
nov   24 05:37:38 Zen kernel: [Hardware Error]: CPU:0 (19:21:0) MC27_STATUS[Over|CE|MiscV|-|-|-|SyndV|-|-|-]: 0xd82000000002080b
nov   24 05:37:38 Zen kernel: [Hardware Error]: IPID: 0x0001002e00000500, Syndrome: 0x000000005a020001
nov   24 05:37:38 Zen kernel: [Hardware Error]: Power, Interrupts, etc. Ext. Error Code: 2, Link Error.
nov   24 05:37:38 Zen kernel: [Hardware Error]: cache level: L3/GEN, mem/io: IO, mem-tx: GEN, part-proc: SRC (no timeout)
but then overnight additional errors didn't happen. Anyway now downed the ram voltage to 1.33V just in case (read somewhere that those newer memories sometimes behave better with lower voltage and upping voltage might even cause more problems), so far first two hours errorless, but we'll see what happens in few days.
 
Last edited:

/dev/null

[H]F Junkie
Joined
Mar 31, 2001
Messages
15,190
Sorry for necroing, but what was the problem then finally? And solution?
Getting the same problem in log on my brand new build (few times a day only so far).
MSI Tomahawk X570 (with newest non-beta BIOS -- 7C84v14; there is newer beta also which should enable SAM, but ... don't risk with beta);
Ryzen 5950X;
4x8 3600MHz G.Skill 18CL (so nothing fancy; but with "Compatible with Ryzen" sticker; F4-3600C18D-16GTZRX; 2 of such kits then; with the stock A.XMP conf);
Linux Zen 5.10.0-1-MANJARO #1 SMP PREEMPT Wed Nov 18 07:17:08 UTC 2020 x86_64 GNU/Linux
linux-firmware 20201120.r1785.bc9cd0b-1

Code:
nov   24 01:23:57 Zen kernel: [Hardware Error]: Corrected error, no action required.
nov   24 01:23:57 Zen kernel: [Hardware Error]: CPU:0 (19:21:0) MC27_STATUS[-|CE|MiscV|-|-|-|SyndV|-|-|-]: 0x982000000002080b
nov   24 01:23:57 Zen kernel: [Hardware Error]: IPID: 0x0001002e00000500, Syndrome: 0x000000005a020001
nov   24 01:23:57 Zen kernel: [Hardware Error]: Power, Interrupts, etc. Ext. Error Code: 2, Link Error.
nov   24 01:23:57 Zen kernel: [Hardware Error]: cache level: L3/GEN, mem/io: IO, mem-tx: GEN, part-proc: SRC (no timeout)
1 time in first 48h of operation.
Panicing a little, googling, found this thread here and another for EPYC processors.

Reading here suggestion to rise mem voltage, upped from 1.37 (auto setting; I assume from 1.35 real results slightly bigger) to 1.40 (but bios shows real voltage 1.42) and got almost immiadetly in next hour 2 more those errors (and second one is marginally different also... there is one "Over" instead of previous two "-" in status and the last long number there is different also:
Code:
nov   24 05:01:00 Zen kernel: [Hardware Error]: Corrected error, no action required.
nov   24 05:01:00 Zen kernel: [Hardware Error]: CPU:0 (19:21:0) MC27_STATUS[-|CE|MiscV|-|-|-|SyndV|-|-|-]: 0x982000000002080b
nov   24 05:01:00 Zen kernel: [Hardware Error]: IPID: 0x0001002e00000500, Syndrome: 0x000000005a020001
nov   24 05:01:00 Zen kernel: [Hardware Error]: Power, Interrupts, etc. Ext. Error Code: 2, Link Error.
nov   24 05:01:00 Zen kernel: [Hardware Error]: cache level: L3/GEN, mem/io: IO, mem-tx: GEN, part-proc: SRC (no timeout)
nov   24 05:37:38 Zen kernel: [Hardware Error]: Corrected error, no action required.
nov   24 05:37:38 Zen kernel: [Hardware Error]: CPU:0 (19:21:0) MC27_STATUS[Over|CE|MiscV|-|-|-|SyndV|-|-|-]: 0xd82000000002080b
nov   24 05:37:38 Zen kernel: [Hardware Error]: IPID: 0x0001002e00000500, Syndrome: 0x000000005a020001
nov   24 05:37:38 Zen kernel: [Hardware Error]: Power, Interrupts, etc. Ext. Error Code: 2, Link Error.
nov   24 05:37:38 Zen kernel: [Hardware Error]: cache level: L3/GEN, mem/io: IO, mem-tx: GEN, part-proc: SRC (no timeout)
but then overnight additional errors didn't happen. Anyway now downed the ram voltage to 1.33V just in case (read somewhere that those newer memories sometimes behave better with lower voltage and upping voltage might even cause more problems), so far first two hours errorless, but we'll see what happens in few days.
Moving My 3600mhz ram to my x570 board+Ryzen 5 3600 works great. I'm using it there. 0 crashes. My x370+3900x hs 4 * 8GB @ 3200mhz and 0 crashes as well. I'm guessing it's a t-top problem.
 

DEEM0N

n00b
Joined
Apr 28, 2015
Messages
16
Moving My 3600mhz ram to my x570 board+Ryzen 5 3600 works great. I'm using it there. 0 crashes. My x370+3900x hs 4 * 8GB @ 3200mhz and 0 crashes as well. I'm guessing it's a t-top problem.

No luck with the "error" messages - they still happen when mostly gaming, however there has been 0 crashes for me (so far they have been 100% corrected then I assume?).
Anyway, just in case ran memtest86 for past 9-10 hours. 0 errors. Also I am using MSI motherboard... not sure it has this "T-topology" also, as it's kind of ASUS thing?

got in MSI forum such a reply - "I have this error too. Wait for agesa 1.1.8.0 in december. Its agesa bug or cpu defect." and when you say it went away with swapping RAM, then it can't exactly be cpu defect (because cpu defects don't "go away" with swapping ram?)? So hopefully it's agesa bug and will be fixed soon :) 🤞
 
Last edited:
Top