System stability issues, possibly RAM?

[CrisR82]

Yesterday I grabbed a paid of extra RAM sticks for my desktop PC and last night, everything was actually running fine but this morning, I've been having some pretty severe freezing and outright instant rebooting.

Currently running an ASRock B650 PG Lightning (BIOS v2.10), AMD Ryzen 7700X, with x4 16GB sticks of Kingston Fury Beast DDR5-5600.

 

Previously with the 2 sticks (x2 16GB), I had them running with EXPO at their rated 5600 speeds. Now with the 4 sticks (x4 16GB), after the crashing started, I went back and switched the BIOS setting from "EXPO 5600" to "Auto" and I THINK it might be stable now (at least no further crashing at the time of writing).

The exact RAM: https://www.kingston.com/en/memory/gaming/kingston-fury-beast-ddr5-memory

The exact motherboard: https://pg.asrock.com/mb/AMD/B650 PG Lightning/

 

Got a suggestion to try using TestMem5 to check the stability but I'm not sure I'm using it correctly - when opening it, it auto-starts doing stuff, then I load a preset people suggested which closes the app, then I run it as an admin which gives me a bunch of identical messages that "I need to run it as an admin to use some instruction set(?)" (which is weird because I AM running it as an admin)...but weirdly enough, the first run (non-admin) caps out the RAM in task manager, while the 2nd run (as admin, supposedly with the custom config) caps out out around 20GB. Assuming I managed to do it correctly, while at the EXPO setting, the system DID crash after the app reported a few errors shortly (20-40 seconds), while later with the Auto setting on the RAM, it didn't crash for 5 minutes.

 

I did check ASRock's support page on the motherboard and the specific module set (KF556C36BBEK2-32) WAS listed as comparable in the Raphael memory validation, unless I somehow completely misunderstood what the listing means.

 

Anyone has any tips on how to approach this?

[TatamiMatt]

10 minutes ago, CrisR82 said:

Yesterday I grabbed a paid of extra RAM sticks for my desktop PC and last night, everything was actually running fine but this morning, I've been having some pretty severe freezing and outright instant rebooting.

Currently running an ASRock B650 PG Lightning (BIOS v2.10), AMD Ryzen 7700X, with x4 16GB sticks of Kingston Fury Beast DDR5-5600.

 

Previously with the 2 sticks (x2 16GB), I had them running with EXPO at their rated 5600 speeds. Now with the 4 sticks (x4 16GB), after the crashing started, I went back and switched the BIOS setting from "EXPO 5600" to "Auto" and I THINK it might be stable now (at least no further crashing at the time of writing).

The exact RAM: https://www.kingston.com/en/memory/gaming/kingston-fury-beast-ddr5-memory

The exact motherboard: https://pg.asrock.com/mb/AMD/B650 PG Lightning/

 

Got a suggestion to try using TestMem5 to check the stability but I'm not sure I'm using it correctly - when opening it, it auto-starts doing stuff, then I load a preset people suggested which closes the app, then I run it as an admin which gives me a bunch of identical messages that "I need to run it as an admin to use some instruction set(?)" (which is weird because I AM running it as an admin)...but weirdly enough, the first run (non-admin) caps out the RAM in task manager, while the 2nd run (as admin, supposedly with the custom config) caps out out around 20GB. Assuming I managed to do it correctly, while at the EXPO setting, the system DID crash after the app reported a few errors shortly (20-40 seconds), while later with the Auto setting on the RAM, it didn't crash for 5 minutes.

 

I did check ASRock's support page on the motherboard and the specific module set (KF556C36BBEK2-32) WAS listed as comparable in the Raphael memory validation, unless I somehow completely misunderstood what the listing means.

 

Anyone has any tips on how to approach this?

DDR5 runs noticably better with 2 sticks and can even be unstable with 4 sticks running overclocked

 

Especially with 2 seperate kits

[Alex Atkin UK]

2 hours ago, TatamiMatt said:

DDR5 runs noticably better with 2 sticks and can even be unstable with 4 sticks running overclocked

 

Especially with 2 seperate kits

Personally I went for 2x Corsair Vengeance 32 GB DDR5-6000 CL30-36-36-76 as I knew 4x sticks were known to be problematic.

 

I've never really paid much attention to the QVL for motherboards as they tend not to update them past launch day.

[CrisR82]

Ok, I think I managed to get the TestMem5 app to work correctly and ran some tests:
EXPO-5600 cl36 - crashes in less than a minute

Auto (defaults to 3600 according to Task Manager) - runs stable for over 1 hour and 15 minutes

EXPO-5200 cl40 - runs stable for over an hour

 

Given how the motherboard is supposedly validated for up to 6400 in 4-slot configuration and the 4 5600 modules I have are on the QVL list, my only guess is that this is a CPU issue? Checked the 7700X page and it does appear to be listed as validated up to 5200 in 4 module config.

 

Assuming that is correct, I guess the question now is - is this something a future BIOS update could solve, or this type of issue is only solved by upgrading the CPU itself?

[TatamiMatt]

28 minutes ago, CrisR82 said:

Ok, I think I managed to get the TestMem5 app to work correctly and ran some tests:
EXPO-5600 cl36 - crashes in less than a minute

Auto (defaults to 3600 according to Task Manager) - runs stable for over 1 hour and 15 minutes

EXPO-5200 cl40 - runs stable for over an hour

 

Given how the motherboard is supposedly validated for up to 6400 in 4-slot configuration and the 4 5600 modules I have are on the QVL list, my only guess is that this is a CPU issue? Checked the 7700X page and it does appear to be listed as validated up to 5200 in 4 module config.

 

Assuming that is correct, I guess the question now is - is this something a future BIOS update could solve, or this type of issue is only solved by upgrading the CPU itself?

Honestly, id say its just an AM5-DDR5 compliance thing, AM5 really benefits in performance and stability with dual channel over quad channel configurations, you can get it to run as youve found, but it takes a hit in performance to have that config working, are you able to RMA one of the kits? Id just stick to 32GB if thats all you need as youll be able to get a sizeable chunk more performance out of it, unless you need the 64GB then running it in 5200CL40 should be fine for now, you can maybe see about selling both kits and buying one 64gb kit instead, 5200 with a lower latency might also be stable and provide some extra performance also if you tweak the timings yourself instead of using a preset profile

[TatamiMatt]

@CrisR82 Added some more above that might benefit you

[Eigenvektor]

8 minutes ago, TatamiMatt said:

AM5 really benefits in performance and stability with dual channel over quad channel configuration

Minor nitpick: There is no quad channel on consumer boards. Four sticks is simply dual channel with two sticks per channel.

[TatamiMatt]

Just now, Eigenvektor said:

Minor nitpick: There is no quad channel on consumer boards. Four sticks is simply dual channel with two sticks per channel.

Yes thats what i meant, thanks for the clarification 

[CrisR82]

11 minutes ago, TatamiMatt said:

Honestly, id say its just an AM5-DDR5 compliance thing, AM5 really benefits in performance and stability with dual channel over quad channel configurations, you can get it to run as youve found, but it takes a hit in performance to have that config working, are you able to RMA one of the kits? Id just stick to 32GB if thats all you need as youll be able to get a sizeable chunk more performance out of it, unless you need the 64GB then running it in 5200CL40 should be fine for now, you can maybe see about selling both kits and buying one 64gb kit instead, 5200 with a lower latency might also be stable and provide some extra performance also if you tweak the timings yourself instead of using a preset profile

I suppose I'll live with a slightly lower clock on the RAM, at least I didn't over-pay for a DDR5-6000 set of sticks so...

The timings part I'd love to fix if that won't cause stability issues, but not really confident enough in my manual setting skills when it comes to overclocking stuff so I think I'll leave that for now (had a look in the BIOS, it has what looks to be the stick spec'ed values on the left and what it is currently set to on the right, but my guess is just matching the two is not the way you do this).

[TatamiMatt]

8 minutes ago, CrisR82 said:

I suppose I'll live with a slightly lower clock on the RAM, at least I didn't over-pay for a DDR5-6000 set of sticks so...

The timings part I'd love to fix if that won't cause stability issues, but not really confident enough in my manual setting skills when it comes to overclocking stuff so I think I'll leave that for now (had a look in the BIOS, it has what looks to be the stick spec'ed values on the left and what it is currently set to on the right, but my guess is just matching the two is not the way you do this).

Nah, thats bang on, just match the 2 and for the "main" CAS latency drop it to 38, run stability test, if that works try 36, rinse, repeat, should net you some small performance gains as AM5 quite likes low latency RAM

[CrisR82]

3 minutes ago, TatamiMatt said:

Nah, thats bang on, just match the 2 and for the "main" CAS latency drop it to 38, run stability test, if that works try 36, rinse, repeat, should net you some small performance gains as AM5 quite likes low latency RAM

Oh, well, things HAVE changed for the better in some regards since my last PC-tinkering adventures period, thanks, I'll give it a go  

[TatamiMatt]

2 hours ago, CrisR82 said:

Oh, well, things HAVE changed for the better in some regards since my last PC-tinkering adventures period, thanks, I'll give it a go  

You can also drop all the other timings further by tinkering but i am in no way knowledgable enough to be able to tell you what to do in that regard 

 

I know there are a few guys here who are quite knowledgable about RAM tuning, you could post into the memory sub thread about tuning your memory timings and possibly get advice from some of them

 

i dont like always tagging you for RAM tuning @RONOTHAN## but youre the smartest guy i know on the subject

 

@Eigenvektor Do you know your RAM? I have an inkling of a memory of you helping out with RAM tuning at some point but i could be wrong

[RONOTHAN##]

4 hours ago, TatamiMatt said:

dont like always tagging you for RAM tuning @RONOTHAN## but youre the smartest guy i know on the subject

Na, I don't really mind at all. 

 

 

OP, if you really want to run 4x16GB, there's a couple things you should try to do. First things first is to make sure that you're sticks are all actually the same, since if you're running a RAM config with mixed memory dies, tuning this will be an absolute nightmare and it's just not worth the migraine. There's a couple ways of doing that, though the easiest is to just figure out the max frequency of each stick. Set the memory to voltages to 1.4V and the primary timings to 40-50-50-128, then start increasing the memory frequency until it stops booting (this will take quite a while to boot, so be prepared for 2-3 minutes per boot). If it does 6400 at the max, it's likely Samsung, if it does ~6800-7000 at the max it's likely Hynix 16Gb M die, and if it does 7600+ it's Hynix 16Gb A die. There are other ways to figure it out by looking at the label (with Kingston kits, this can only confirm if its Samsung, not the Hynix revision), checking tRFC (this more works on Intel/APUs though for complicated reasons), or using a program called Thaiphoon Burner (this doesn't always work and can give some weird issues on DDR5/Windows 11, so keep that in mind), so going for the max frequency is probably the most consistent option, albeit the most time consuming. 

 

After you figure out if everything is actually the same die revision, try messing around with a couple different BIOS revisions. Over the past ~year, AMD has more or less alternated between AGESA versions that have had 4x16GB working well and being completely broken. Newer does not always mean better for the BIOS, and if you find the right BIOS revision you might be able to get even 6000+ to work with 4 DIMMs. I should probably mention that the method I'm giving more or less ignores the XMP/EXPO profile on those DIMMs as almost any kit rated at 5600 CL36 should do at least DDR5 6000, so you might as well get that extra frequency to work if you're gonna have to tune this slightly anyway

 

Once you have the BIOS that you're running, it is somewhat time to get the max frequency to work with 4 DIMMs. Keep using those timings from before (you can lower them later), then start raising the memory frequency until you have instability in a stress test called Y-Cruncher VT3 (this is arguably the best test for finding memory instability due to high frequency). Auto settings should get up until 6000 to work, though once you go to 6200 you might need to mess with the SOC voltage (higher is better for maximum frequency, though don't go above 1.3V) and you'll need to manually set it to MEMCLK == UCLK. After the max frequency is discovered, it's time to start lowering the primary timings until you have stability issues. These will vary greatly on what memory chip you have, though if you have one of the Hynix variance, it'll be something between 30-36-32-30 and 32-40-40-40 and if you have Samsung it'll almost certainly be 36-36-36-76. If you want to just figure it out yourself, lower them one at a time until it starts crashing in memory stress tests (TestMem5 is a good free one, though admittedly it's a little hard to use). 

 

You can then try tuning the sub-timings afterwards, as those have a quite a bit of a performance uplift. Those you can find manually, though it'll probably be best to just find a Buildzoid timings video with your specific IC and just copy the timings from it. 

 

LMK if you have any questions, almost certainly something in there was confusing and I'll need to clarify it. 

[Eigenvektor]

5 hours ago, TatamiMatt said:

Do you know your RAM? I have an inkling of a memory of you helping out with RAM tuning at some point but i could be wrong

I'm afraid not. I can answer general technical question when it comes to RAM, but nothing specific when it comes to tuning, especially about DDR5. But I can see you're in good hands already

[TatamiMatt]

15 minutes ago, RONOTHAN## said:

Na, I don't really mind at all. 

 

 

OP, if you really want to run 4x16GB, there's a couple things you should try to do. First things first is to make sure that you're sticks are all actually the same, since if you're running a RAM config with mixed memory dies, tuning this will be an absolute nightmare and it's just not worth the migraine. There's a couple ways of doing that, though the easiest is to just figure out the max frequency of each stick. Set the memory to voltages to 1.4V and the primary timings to 40-50-50-128, then start increasing the memory frequency until it stops booting (this will take quite a while to boot, so be prepared for 2-3 minutes per boot). If it does 6400 at the max, it's likely Samsung, if it does ~6800-7000 at the max it's likely Hynix 16Gb M die, and if it does 7600+ it's Hynix 16Gb A die. There are other ways to figure it out by looking at the label (with Kingston kits, this can only confirm if its Samsung, not the Hynix revision), checking tRFC (this more works on Intel/APUs though for complicated reasons), or using a program called Thaiphoon Burner (this doesn't always work and can give some weird issues on DDR5/Windows 11, so keep that in mind), so going for the max frequency is probably the most consistent option, albeit the most time consuming. 

 

After you figure out if everything is actually the same die revision, try messing around with a couple different BIOS revisions. Over the past ~year, AMD has more or less alternated between AGESA versions that have had 4x16GB working well and being completely broken. Newer does not always mean better for the BIOS, and if you find the right BIOS revision you might be able to get even 6000+ to work with 4 DIMMs. I should probably mention that the method I'm giving more or less ignores the XMP/EXPO profile on those DIMMs as almost any kit rated at 5600 CL36 should do at least DDR5 6000, so you might as well get that extra frequency to work if you're gonna have to tune this slightly anyway

 

Once you have the BIOS that you're running, it is somewhat time to get the max frequency to work with 4 DIMMs. Keep using those timings from before (you can lower them later), then start raising the memory frequency until you have instability in a stress test called Y-Cruncher VT3 (this is arguably the best test for finding memory instability due to high frequency). Auto settings should get up until 6000 to work, though once you go to 6200 you might need to mess with the SOC voltage (higher is better for maximum frequency, though don't go above 1.3V) and you'll need to manually set it to MEMCLK == UCLK. After the max frequency is discovered, it's time to start lowering the primary timings until you have stability issues. These will vary greatly on what memory chip you have, though if you have one of the Hynix variance, it'll be something between 30-36-32-30 and 32-40-40-40 and if you have Samsung it'll almost certainly be 36-36-36-76. If you want to just figure it out yourself, lower them one at a time until it starts crashing in memory stress tests (TestMem5 is a good free one, though admittedly it's a little hard to use). 

 

You can then try tuning the sub-timings afterwards, as those have a quite a bit of a performance uplift. Those you can find manually, though it'll probably be best to just find a Buildzoid timings video with your specific IC and just copy the timings from it. 

 

LMK if you have any questions, almost certainly something in there was confusing and I'll need to clarify it. 

Your knowledge on RAM never cease to amaze me

[CrisR82]

2 hours ago, RONOTHAN## said:

Na, I don't really mind at all. 

 

 

OP, if you really want to run 4x16GB, there's a couple things you should try to do. First things first is to make sure that you're sticks are all actually the same, since if you're running a RAM config with mixed memory dies, tuning this will be an absolute nightmare and it's just not worth the migraine. There's a couple ways of doing that, though the easiest is to just figure out the max frequency of each stick. Set the memory to voltages to 1.4V and the primary timings to 40-50-50-128, then start increasing the memory frequency until it stops booting (this will take quite a while to boot, so be prepared for 2-3 minutes per boot). If it does 6400 at the max, it's likely Samsung, if it does ~6800-7000 at the max it's likely Hynix 16Gb M die, and if it does 7600+ it's Hynix 16Gb A die. There are other ways to figure it out by looking at the label (with Kingston kits, this can only confirm if its Samsung, not the Hynix revision), checking tRFC (this more works on Intel/APUs though for complicated reasons), or using a program called Thaiphoon Burner (this doesn't always work and can give some weird issues on DDR5/Windows 11, so keep that in mind), so going for the max frequency is probably the most consistent option, albeit the most time consuming. 

 

After you figure out if everything is actually the same die revision, try messing around with a couple different BIOS revisions. Over the past ~year, AMD has more or less alternated between AGESA versions that have had 4x16GB working well and being completely broken. Newer does not always mean better for the BIOS, and if you find the right BIOS revision you might be able to get even 6000+ to work with 4 DIMMs. I should probably mention that the method I'm giving more or less ignores the XMP/EXPO profile on those DIMMs as almost any kit rated at 5600 CL36 should do at least DDR5 6000, so you might as well get that extra frequency to work if you're gonna have to tune this slightly anyway

 

Once you have the BIOS that you're running, it is somewhat time to get the max frequency to work with 4 DIMMs. Keep using those timings from before (you can lower them later), then start raising the memory frequency until you have instability in a stress test called Y-Cruncher VT3 (this is arguably the best test for finding memory instability due to high frequency). Auto settings should get up until 6000 to work, though once you go to 6200 you might need to mess with the SOC voltage (higher is better for maximum frequency, though don't go above 1.3V) and you'll need to manually set it to MEMCLK == UCLK. After the max frequency is discovered, it's time to start lowering the primary timings until you have stability issues. These will vary greatly on what memory chip you have, though if you have one of the Hynix variance, it'll be something between 30-36-32-30 and 32-40-40-40 and if you have Samsung it'll almost certainly be 36-36-36-76. If you want to just figure it out yourself, lower them one at a time until it starts crashing in memory stress tests (TestMem5 is a good free one, though admittedly it's a little hard to use). 

 

You can then try tuning the sub-timings afterwards, as those have a quite a bit of a performance uplift. Those you can find manually, though it'll probably be best to just find a Buildzoid timings video with your specific IC and just copy the timings from it. 

 

LMK if you have any questions, almost certainly something in there was confusing and I'll need to clarify it. 

Thanks for the detailed info!

Was messing around after work for a while and was able to get it running a bit better - switched it to DDR5-5400 and the latency to CL36 using the same values from the EXPO profile for DDR5-5600 (36-38-38-80-118-85-827-449-365)...though weirdly enough, one of them - tRC - refused to accept what I typed - the preset has it at 85, but whenever I typed it, it auto-changes it to 84...no clue what that's about but I can't seem to change it so I guess that's alright?

 

The sticks themselves, I think I might've gotten very lucky - according to HWiNFO, they are identical other than the serial number and manufacturing month (one set is 2023 week 43, the other is 2023 week 51, both sets with SK Hynix chips)

 

Running TestMem5 again didn't give me any errors after 30 minutes (5600 frequency showed errors in the first minute, sometimes hard-resetting the whole PC) so maybe this is as far as I can get without switching BIOS versions.

 

One thing I wasn't able to figure out is the Y-Cruncher test, I assume I need to start the benchmark, then select an option for RAM, but the highest it goes is 44GB and at one point (less than a minute I think) it just said it ran out of decimals and stopped...no clue if that's supposed to happen, but it didn't seem to say anything about a pass/fail or give me any indication if that was good or not o_o

[RONOTHAN##]

1 hour ago, CrisR82 said:

one of them - tRC - refused to accept what I typed - the preset has it at 85, but whenever I typed it, it auto-changes it to 84...

Pretty sure the timing that's set to 85 by default isn't actually tRC, but tWR. On AMD CPUs, WR can only be set to multiples of 6, so hence why it always went to 84 (Intel CPUs can set it to any arbitrary value, so it's probably just a carryover from there). tRC should be the 118 value, as tRC is almost always equal to tRP (the 2nd 38) plus tRAS (the 80). 

 

If your curious, that entire string should be in the order tCL-tRCD-tRP-tRAS-tRC-tWR-tRFC-tRFC2-tRFCpb. 

 

1 hour ago, CrisR82 said:

both sets with SK Hynix chips)

Well, there are three different SK Hynix chips that exist: 16Gb A die, 16Gb M die, and 24Gb M die. You probably have one of the two 16Gb varieties, though the only real way to tell is to test what the max frequency they'll boot is. 16Gb M die doesn't like to do over 6800 on AMD, while A die will do 8000+ with ease if the board and CPU can support it (they should do 7600 on your board without too much trouble). If I were to take a guess, they're probably both 16Gb M die, but it is still worth checking just to be sure. 

 

1 hour ago, CrisR82 said:

One thing I wasn't able to figure out is the Y-Cruncher test, I assume I need to start the benchmark, then select an option for RAM, but the highest it goes is 44GB and at one point (less than a minute I think) it just said it ran out of decimals and stopped...no clue if that's supposed to happen, but it didn't seem to say anything about a pass/fail or give me any indication if that was good or not o_o

OK, so Y-Cruncher is a bit weird to use, I probably should've specified how to use it. There is a way to get to it through the GUI (launch the .exe, hit 1 for stress, hit 8 to clear all tests, type the number associated with VT3 (I think 17, but don't feel like checking), then hit 0 to start), but the way I tend to do it is just through the command line. Launch CMD and cd into the directory of Y-Cruncher. Run the command "y-cruncher stress VT3" then let it run for two hours to confirm stability. 

 

You can make that into a batch file as well, and you can also set a time limit by making the batch file "y-cruncher stress VT3 -TL:7200" where 7200 is the number of seconds you want it to run for. 

[CrisR82]

On 4/10/2024 at 12:54 AM, RONOTHAN## said:

Pretty sure the timing that's set to 85 by default isn't actually tRC, but tWR. On AMD CPUs, WR can only be set to multiples of 6, so hence why it always went to 84 (Intel CPUs can set it to any arbitrary value, so it's probably just a carryover from there). tRC should be the 118 value, as tRC is almost always equal to tRP (the 2nd 38) plus tRAS (the 80). 

 

If your curious, that entire string should be in the order tCL-tRCD-tRP-tRAS-tRC-tWR-tRFC-tRFC2-tRFCpb. 

 

Well, there are three different SK Hynix chips that exist: 16Gb A die, 16Gb M die, and 24Gb M die. You probably have one of the two 16Gb varieties, though the only real way to tell is to test what the max frequency they'll boot is. 16Gb M die doesn't like to do over 6800 on AMD, while A die will do 8000+ with ease if the board and CPU can support it (they should do 7600 on your board without too much trouble). If I were to take a guess, they're probably both 16Gb M die, but it is still worth checking just to be sure. 

 

OK, so Y-Cruncher is a bit weird to use, I probably should've specified how to use it. There is a way to get to it through the GUI (launch the .exe, hit 1 for stress, hit 8 to clear all tests, type the number associated with VT3 (I think 17, but don't feel like checking), then hit 0 to start), but the way I tend to do it is just through the command line. Launch CMD and cd into the directory of Y-Cruncher. Run the command "y-cruncher stress VT3" then let it run for two hours to confirm stability. 

 

You can make that into a batch file as well, and you can also set a time limit by making the batch file "y-cruncher stress VT3 -TL:7200" where 7200 is the number of seconds you want it to run for. 

Finally managed to spare some time to mess around a bit more, man this is a hectic week...anyway, I got around to running stress tests on this and so far I'm getting the following:

 

DDR5-5400 CL36 (manually set):

Y-Cruncher -> error on run 9

TestMem5 -> 1 error after about 35 minutes

 

DDR5-5200 CL36 (manually set):

Y-Cruncher -> error on run 46

TestMem5 -> error after 1 hour and 35 minutes

 

Attached a couple of screenshots of what the exact errors are since they seem to be actually different.

Also, you were right, I miswrote there, 85 was set on tWR, but I wasn't wrong on the preset having it on 85 on EXPO, adding a picture (sorry for the poop quality) of the preset that I used to manually set timings (EXPO 0).

 

I was running DDR5-5400 for a couple of days and it seemed to be working fine, but yesterday after a Skype call I unplugged a USB mic from my PC and the whole PC immediately restarted, not sure if this is a RAM-thing but I haven't seen it happen before so it that is the case, I'll have to keep lowering settings I guess =/

 

Probably not going to result in anything tangible, but also emailed ASRock earlier today to see if they have any recommendations on this (I mean there must've been someone else with this type of issue so they might have more exact recommended settings?).

[RONOTHAN##]

2 hours ago, CrisR82 said:

I was running DDR5-5400 for a couple of days and it seemed to be working fine, but yesterday after a Skype call I unplugged a USB mic from my PC and the whole PC immediately restarted, not sure if this is a RAM-thing but I haven't seen it happen before so it that is the case, I'll have to keep lowering settings I guess =/

Given that you were still getting crashes in Y-Cruncher and TM5, I'd say that's pretty likely to be memory that's the issue still. If your setup was stable, it should run indefinitely at whatever memory speed you want. 

 

2 hours ago, CrisR82 said:

Also, you were right, I miswrote there, 85 was set on tWR, but I wasn't wrong on the preset having it on 85 on EXPO, adding a picture (sorry for the poop quality) of the preset that I used to manually set timings (EXPO 0).

Yeah, I never doubted that, it's just that they copied the XMP profile's timings and since Intel lets you set tWR arbitrarily, 85 works on there but needs to be rounded to 84 on AMD. Hynix memory very consistently does 48 on AM5 systems, so rounding down one tick from the EXPO is more than fine for stability. 

 

It will be worth checking a few other BIOS revisions though, there are AM5 BIOSes where 4 DIMMs is sketchy and others where it's about as good as 2x32. You seem to be on one that's more towards the sketchy side. 

[CrisR82]

On 4/11/2024 at 7:18 PM, RONOTHAN## said:

Given that you were still getting crashes in Y-Cruncher and TM5, I'd say that's pretty likely to be memory that's the issue still. If your setup was stable, it should run indefinitely at whatever memory speed you want. 

 

Yeah, I never doubted that, it's just that they copied the XMP profile's timings and since Intel lets you set tWR arbitrarily, 85 works on there but needs to be rounded to 84 on AMD. Hynix memory very consistently does 48 on AM5 systems, so rounding down one tick from the EXPO is more than fine for stability. 

 

It will be worth checking a few other BIOS revisions though, there are AM5 BIOSes where 4 DIMMs is sketchy and others where it's about as good as 2x32. You seem to be on one that's more towards the sketchy side. 

Will probably keep experimenting the following days to get it as stable as I can.

Thank you and everyone else in this thread, learned quite a lot of new things I didn't even think about before! ^-^

 

EDIT for anyone that happens to have the same question and comes across this: tried 4 other BIOS revisions going back from 2.10, all seem to have the same results. Went back to 2.10 and ran further testing, the only frequency that was fully stable was DDR5-3800 with manual CL36 set timings.

I managed to get in contact with ASRock's support team, they claim that 4 sticks WILL run but only if they are part of a 4-stick kit rather than x2 2-stick kits. I have no way to test that obviously so cannot comment any more on that. Since the performance impact of a frequency this low seems to be pretty substantial, I decided to return the RAM kit and just spend the money on something else.