2 Ram module makes PC restart

[amrgul1]

The problem I have currently is that PC restarts sometimes on its own. If I remove one of the ram module, then it does not. I can easily test this - when entering in the folder where I keep my videos about 4gb and set file view to large I get restarts everytime, but with only one memory module, it does not.
I tried to change memory modules and also put them separately in both slots and problem does not occur when there is only 1 ram, but if put both ram I have restarts.
I can also note that if I change the video card to my old one and both rams are on, then restart also does not appear. But I remember that restarts occurred even before I installed the newly bought video card but I didn't pay much attention then. About my video card issue you can read on my other post here: https://www.techpowerup.com/forums/threads/msi-armor-rx-570-8gb-oc-bios.304943/
Once when PC restarted Win10 gave the blue screen with error code: "irgl not less or equal". As it appears it is memory related problem (only which memory - ram or vram?!), I have dump file of this error if someone would need.
Also if I try to increase ram frequency in bios to 2800 Mhz (default is 2666Mhz), PC won't start at all, just black screen with "cable not connected" message.
Please, try to help me.
my specs:
Asus prime a320m-f
Ryzen 5 2600
Msi armor RX 570 oc
Patriot 2*8gb ddr4 ram
1st player 700 w 80+ psu

 

[Shrek]

Run MemTest

System Utilities Downloads | TechPowerUp

 

[amrgul1]

Run MemTest

System Utilities Downloads | TechPowerUp

Already did 1 pass with no errors.

 

[Harlow]

I usually run MemTest86 longer (at least three complete passes) when checking suspect RAM. One pass is usually enough to reveal gross memory faults, but minor errors may not appear until MemTest86 has gone through a number of full cycles. If you can, leave MemTest86 running overnight. If a fault is detected, re-run the test with each DIMM in turn in the first DIMM socket. Then, test each DIMM on its own in the second DIMM socket, thereby checking each DIMM socket (and memory channel) separately.

If you have a completely different set of RAM, it would help to prove whether or not your existing DIMMs are good or bad. Similarly, a different CPU would help in determining if the fault lies in the 2600 or the A320M-F. Substitute one component at a time until you find the culprit, or run out of suitable parts.

I have an older AMD Athlon X4 760K system where 2 of the 4 DIMM slots on the motherboard are unusable, but I haven't bothered to work out if it's the CPU or mobo at fault. The system runs fine with 2 x 8GB in Single Channel mode. It's not my main system so I'm happy with the slightly lower performance.

As for IRQL_NOT_LESS_OR_EQUAL errors, have you run 'DISM /Online /Cleanup-Image /CheckHealth' or 'sfc /scannow' to check your system files for errors. Have you recently installed a new driver. Both issues can result in an IRQL error.

 

[amrgul1]

I usually run MemTest86 longer (at least three complete passes) when checking suspect RAM. One pass is usually enough to reveal gross memory faults, but minor errors may not appear until MemTest86 has gone through a number of full cycles. If you can, leave MemTest86 running overnight. If a fault is detected, re-run the test with each DIMM in turn in the first DIMM socket. Then, test each DIMM on its own in the second DIMM socket, thereby checking each DIMM socket (and memory channel) separately.

All right I will leave memtest this night and post the result tomorrow.

If you have a completely different set of RAM, it would help to prove whether or not your existing DIMMs are good or bad. Similarly, a different CPU would help in determining if the fault lies in the 2600 or the A320M-F. Substitute one component at a time until you find the culprit, or run out of suitable parts.

I wish but unfortunately I don't have other parts except as I already mentioned in my first post - an old gpu and with it restarts dissapear when setting file's large viewing size. It feels like these 3 parts somehow don't like each other or maybe even power issue.

As for IRQL_NOT_LESS_OR_EQUAL errors, have you run 'DISM /Online /Cleanup-Image /CheckHealth' or 'sfc /scannow' to check your system files for errors. Have you recently installed a new driver. Both issues can result in an IRQL error.

I tested twice on a fresh install on windows 10 and windows 7, so commands like "sfc /scannow" shouldn't be needed imo but will try that too.
I installed Asus chipset drivers and latest video driver from Amd.

 

[Harlow]

I was going to suggest a clean OS install, but you've already tried that. If one of my computers crashes with a BSOD, I run "sfc /scannow" when the PC restarts. Sometimes the scan finds errors and it only takes a couple of minutes to fix. Any sudden shutdown can corrupt system files.

I'd recommend checking out eBay or a similar auction site and consider an inexpensive pair of used DDR4 DIMMs. They don't have to be anything fancy. 2 x 2GB would be sufficient to test the motherboard/AMD 2600/GPU. Of course there's no guarantee second-hand RAM will be in good working order, but I've not had many problems buying used DIMMs, even when they're not shipped in an ESD bag.

It might also be worth setting the memory clock speed in the UEFI/BIOS to the minimum possible, e.g. 2133MHz or 2400MHz (JEDEC defaults). I'd advise against overclocking your RAM with XMP settings until you've got a stable build. Marginal RAM can work fine at lower clock speeds. If you (or anyone else) have already "hammered" the system with a massive CPU overvolt or prolonged overclock (for several years), it may have suffered permanent damage. You're unlikely to notice much difference between 2400MHz and 2866MHz outside benchmark tests.

The A320-M appears to support a wide range of CPUs and you could pick up a low-spec CPU such as an A6-9500 for around $12/£10/E11 on eBay. I have an A6-9500 running stable with a mild overclock of 4.1GHz and a big heatsink. If the replacement CPU works, you'll know your 2600 is duff.

Do you have another computer you could "borrow" a PSU from? Have you set the UEFI/BIOS back to default speeds (no overclocking)?

As a last resort I'd try a replacement motherboard, but only after changing the RAM and the CPU. A super cheap second-hand board might fix the fault(s).

Unless you're prepared to spend (waste) some money on alternative parts, I can't think of anything else apart from a new system after winning the lottery.

Good luck.

 

[amrgul1]

I was going to suggest a clean OS install, but you've already tried that. If one of my computers crashes with a BSOD, I run "sfc /scannow" when the PC restarts. Sometimes the scan finds errors and it only takes a couple of minutes to fix. Any sudden shutdown can corrupt system files.

I'd recommend checking out eBay or a similar auction site and consider an inexpensive pair of used DDR4 DIMMs. They don't have to be anything fancy. 2 x 2GB would be sufficient to test the motherboard/AMD 2600/GPU. Of course there's no guarantee second-hand RAM will be in good working order, but I've not had many problems buying used DIMMs, even when they're not shipped in an ESD bag.

It might also be worth setting the memory clock speed in the UEFI/BIOS to the minimum possible, e.g. 2133MHz or 2400MHz (JEDEC defaults). I'd advise against overclocking your RAM with XMP settings until you've got a stable build. Marginal RAM can work fine at lower clock speeds. If you (or anyone else) have already "hammered" the system with a massive CPU overvolt or prolonged overclock (for several years), it may have suffered permanent damage. You're unlikely to notice much difference between 2400MHz and 2866MHz outside benchmark tests.

The A320-M appears to support a wide range of CPUs and you could pick up a low-spec CPU such as an A6-9500 for around $12/£10/E11 on eBay. I have an A6-9500 running stable with a mild overclock of 4.1GHz and a big heatsink. If the replacement CPU works, you'll know your 2600 is duff.

Do you have another computer you could "borrow" a PSU from? Have you set the UEFI/BIOS back to default speeds (no overclocking)?

As a last resort I'd try a replacement motherboard, but only after changing the RAM and the CPU. A super cheap second-hand board might fix the fault(s).

Unless you're prepared to spend (waste) some money on alternative parts, I can't think of anything else apart from a new system after winning the lottery.

Good luck.

Run memtest 7 pass on each module separately on each slots - no errors. I couldn't run with both of them together because one time pc restarted and second time I think it just froze but I read the message in the left window - "unexpected procedure from cpu1" or something like this.
I also tried downclocking to 2400mhz but still it restarted. That overclocking I did just to test it and pc became unbootable, I had to remove one memory stick to boot it again.
I agree, buying old model cpu\ram\motherboard may be a good idea to test even I can't be sure 100% that they also don't have problems.
Well. what else can I say, either I have to play with just 1 memory in the pc or get tired again to test all of this...

 

[Kodehawa]

one time pc restarted and second time I think it just froze but I read the message in the left window - "unexpected procedure from cpu1" or something like this.
I also tried downclocking to 2400mhz but still it restarted. That overclocking I did just to test it and pc became unbootable, I had to remove one memory stick to boot it again.

Bad CPU contact, bent pins that connect to RAM traces, bad motherboard.

Those are the options remaining in my table, but its definitely better to test with another motherboard if you can. Maybe one of your friend's.

 

[Harlow]

You seem to have shown that both DIMMs work OK when a single stick is inserted into either socket. I infer from this that both memory channels in the CPU are working (in isolation) and the PCB traces from the CPU to both DIMM sockets are OK. However, the computer becomes unstable when the system switches from Single Channel to Dual Channel mode with both DIMMs installed.

I've seen a note saying the Ryzen 2600 can only reach 2933MT/s on a mobo with a high quality 6-layer PCB (and only 2 DIMMs) but i suspect your motherboard may be only 4-layer with fewer screening planes. You've already shown the system remains unstable with 2 DIMMs at 2400MT/s, but a 4-layer PCB might explain why the PC doesn't boot at 2800MT/s.

A quick check on the web reveals that early Ryzen chips were fussy about RAM. Do you know if your memory on the QVL (Qualified Vendors List) for the Asus Prime A320M-F? I don't usually worry when my RAM isn't mentioned in the motherboard's QVL and have been lucky so far. I've just pulled one of the two DIMMs from my working Gigabyte GA-A320M-H mobo with A6-9500 CPU and it's Apacer AU04GGB26CQWBGH, 4GB DDR4 2666 CL19.

Do you have the latest BIOS installed, Version 6042 or Version 6061 Beta? I'm not sure if I'd use the Beta version, but if your BIOS is earlier than 6042, it might be worth updating to see if it improves stability.

I notice you've had problems with a second-hand RV570 which may have been used in crypto mining. I get the impression the system is still unstable when you use another (non-gaming?) graphics card, but it's slightly more stable than the RV570. I'd be inclined to continue testing without the RV570, even if you've successfully flashed its mining BIOS back to standard. It's possible the RV570 may have been stressed/damaged by mining and is now affecting the stability of the PSU and hence the CPU's supply. This is only a guess on my part. I'm starting to "clutch at straws". I don't know the 1st Player 700W PSU. Is it a high quality device?

 

[amrgul1]

You seem to have shown that both DIMMs work OK when a single stick is inserted into either socket. I infer from this that both memory channels in the CPU are working (in isolation) and the PCB traces from the CPU to both DIMM sockets are OK. However, the computer becomes unstable when the system switches from Single Channel to Dual Channel mode with both DIMMs installed.

I've seen a note saying the Ryzen 2600 can only reach 2933MT/s on a mobo with a high quality 6-layer PCB (and only 2 DIMMs) but i suspect your motherboard may be only 4-layer with fewer screening planes. You've already shown the system remains unstable with 2 DIMMs at 2400MT/s, but a 4-layer PCB might explain why the PC doesn't boot at 2800MT/s.

I don't know what pcb layers my mobo has, how to check it...

A quick check on the web reveals that early Ryzen chips were fussy about RAM. Do you know if your memory on the QVL (Qualified Vendors List) for the Asus Prime A320M-F? I don't usually worry when my RAM isn't mentioned in the motherboard's QVL and have been lucky so far.

I looked at QVL list and there is not support for the exact my ram's 2666mhz versions. In details, my ram part.no is PSD48G266681, it is not in the list. There is similar with 2400mhz version which supports 2 socket dimm. So in theory (and practice) this maybe the reason why dual channel doesn't work well!? I'm attaching the list pdf you can also view. Btw, if I buy new rams, which brand would you advise for higher frequencies (which is supported by my mobo and cpu)?

Do you have the latest BIOS installed, Version 6042 or Version 6061 Beta? I'm not sure if I'd use the Beta version, but if your BIOS is earlier than 6042, it might be worth updating to see if it improves stability.

Yes, I have 6042 bios and also don't want to upgrade to beta (if at all).

I notice you've had problems with a second-hand RV570 which may have been used in crypto mining. I get the impression the system is still unstable when you use another (non-gaming?) graphics card, but it's slightly more stable than the RV570. I'd be inclined to continue testing without the RV570, even if you've successfully flashed its mining BIOS back to standard. It's possible the RV570 may have been stressed/damaged by mining and is now affecting the stability of the PSU and hence the CPU's supply. This is only a guess on my part. I'm starting to "clutch at straws". I don't know the 1st Player 700W PSU. Is it a high quality device?

Yes, this was another big problem and I think I solved it successfully with the helpful guys there but now in conjunction with this problem not sure. You know how it is: I have my folder where I keep about 4gb vids. If I enter that folder and set view size to - large or even medium, PC restarts immediately on every try only when I have both ram set, if I have only 1 ram set, then it doesn't restart. I removed this "new" RX 570 and put my old Nvidia card with both ram set and tried to enlarge video icons again and guess what - it does not restarted. So now logically this means that RX 570 makes it but I remember when building PC I also had random restarts but then I haven't yet bought RX 570 and was using my old Nvidia card. All this is very confusing now, I will continue to test with old Nvidia card and also can replace my new ssd with an old hdd to test it too. Btw temperatures on all parts are normal even in a graphics intensive game, I often monitor them.

I don't know the 1st Player 700W PSU. Is it a high quality device?

I bought that Psu new. I also don't know is 1st player high quality brand or not and how to check that.

Well some interesting posts I found in internet:

1. When I was searching for "unexpected procedure from cpu", I found this post: https://github.com/memtest86plus/memtest86plus/issues/180
One of the responses there is: "Ah, looking at the screenshot, your XHCI controller is mapped above the 4GB boundary. Please could you test the fix I've just pushed" and then the author of the post replies: "Yes that worked perfectly, and it's now doing the memtest, thanks!"... I don't know if this could be important and how can I also check this in my Bios.
2. I wanted to run Gpu stress test with Aida64 and when I pressed the checkbox it show this window:

I Searched for Video Driver timeout and found some articles where they say that this should be set to 8. This is one of them: https://www.nvidia.com/en-us/geforc...ards/5/508243/timeout-detection-and-recovery/
"Timeout Detection and Recovery set to low could mean that something has set the TDR or correlating values below their defaults. TDR should be at 8 Use Wagnard TDR Manipulator to make sure these are at Default."
I downloaded TDR Manipulator and it shows - 2 as Windows default. I'm not sure why that guy (Greybear) says it should be set at - 8. Well, maybe, when I flashed the bios of VC and if that bios was incorrect, then I also might have wrong driver for my card and maybe that's why it's not set to 8 and since I get restarts when entering video (only) folder and video is related to VC, then wrong TDR value might trigger restarts and this also explains why I don't have that with my old Nvidia card. But all these are just my guesses.

UPDATE: I tested with old card and hdd and still have restarts. So Video card shouldn't be related.
On the other hand, if it is bad cpu or motherboard, why would they work well on 1 ram module?!

 

[Harlow]

1). The number of layers in a PCB is an obsure detail, but sometimes it's mentioned by the manufacturer when trying to show their product is superior to other makes. Hence claims of 6 or 8-layers and 2oz (ounce) copper on more expensive boards attract buyers looking for quality and overclockability. You might find mention of your board's construction on the Asus web site or in reviews online. I wouldn't worry about it too much about it, until you find out what's causing the instability.

Given the relatively basic spec of the A320 chipset and the low number of power phases in the VRM (Voltage Regulator Module) without heatsinks, I doubt your board is more than a nominal 4-layers. The A320 chipset is not designed for serious overclocking with a high powered CPU, but your 2600 has a relatively low TDP of only 65W so you should get a mild overclock on both RAM frequency (XMP) and CPU frequency (multiplier), but only when all components are working properly.

When buying RAM for an old board, I scan eBay for cheap second-hand DIMMs. I only buy brand new (expensive) RAM for new builds. Unlike GPU scammers offloading duff mining cards on eBay, people selling RAM are usually honest and are selling old RAM after upgrading their systems with higher capacity DIMMs. I've bought dozens of second-hand DIMMs on eBay, but I run MemTest86 as soon as they arrive. I cannot remember receiving any duff RAM, but I don't recall all my purchases. It's up to you to decide on new or second-hand RAM. If you do buy new, check to see if the part number of the DIMM is on the QVL. As already stated, I normally ignore the QVL. I do however recommend looking for RAM that's specifically optimized for AMD chipsets instead of Intel. Most RAM works fine with AMD and Itel, but some systems are fussy.

Have you scanned your disk drive(s) for errors. If any of the sectors holding your 4GB video files are bad, your computer may crash when opening a corrupted file. It's a good idea to run CHKDSK C: /F /R periodically from the Command Prompt (run as Administrator) to look for file errors and bad blocks. I use the Read-only scan test in Hard Disk Sentinel Pro to check the surface of any suspect drives.

1st Player isn't a PSU brand I know, unlike Seasonic, Corsair, EVGA, etc. You might be able to find it in a list or review online. For a decent 700W PSU, I'd expect to pay at least $100US for a good quality unit. If you paid less than $50 for a brand new PSU, I'd treat it with caution. I once bought a cheap 700W PSU, but it died after a year when I plugged the mains lead into the back. Nowadays I stick to mid-range Corsair PSUs between $80 and $160. Some of my old computers have very cheap PSUs which are more than 10 years old but are still working (on borrowed time). I've re-capped several PSUs and motherboards when I see the tops of electrolytics bulging with brown goo oozing out. I'd caution against opening up a switched mode PSU unless you know what you're doing. There's a good reason why the main bulk capacitor is rated at 420V DC. You have been WARNED!

I agree your problem does not appear to be video card related, but XHCI mapping errors are outside my experience. I have the benefit of a large collection of working computers, ranging from an ancient Intel Pentium 4 (socket 478) circa 2003, to a modern AM5 7950X, so I can swap components around until I locate the fault. If you can pick up a cheap socket AM4 mobo on eBay, it might isolate the fault. Unfortunaely, the only way to be sure is to spend some hard-earned cash.

 

[amrgul1]

1). The number of layers in a PCB is an obsure detail, but sometimes it's mentioned by the manufacturer when trying to show their product is superior to other makes. Hence claims of 6 or 8-layers and 2oz (ounce) copper on more expensive boards attract buyers looking for quality and overclockability. You might find mention of your board's construction on the Asus web site or in reviews online. I wouldn't worry about it too much about it, until you find out what's causing the instability.

Given the relatively basic spec of the A320 chipset and the low number of power phases in the VRM (Voltage Regulator Module) without heatsinks, I doubt your board is more than a nominal 4-layers. The A320 chipset is not designed for serious overclocking with a high powered CPU, but your 2600 has a relatively low TDP of only 65W so you should get a mild overclock on both RAM frequency (XMP) and CPU frequency (multiplier), but only when all components are working properly.

When buying RAM for an old board, I scan eBay for cheap second-hand DIMMs. I only buy brand new (expensive) RAM for new builds. Unlike GPU scammers offloading duff mining cards on eBay, people selling RAM are usually honest and are selling old RAM after upgrading their systems with higher capacity DIMMs. I've bought dozens of second-hand DIMMs on eBay, but I run MemTest86 as soon as they arrive. I cannot remember receiving any duff RAM, but I don't recall all my purchases. It's up to you to decide on new or second-hand RAM. If you do buy new, check to see if the part number of the DIMM is on the QVL. As already stated, I normally ignore the QVL. I do however recommend looking for RAM that's specifically optimized for AMD chipsets instead of Intel. Most RAM works fine with AMD and Itel, but some systems are fussy.

Have you scanned your disk drive(s) for errors. If any of the sectors holding your 4GB video files are bad, your computer may crash when opening a corrupted file. It's a good idea to run CHKDSK C: /F /R periodically from the Command Prompt (run as Administrator) to look for file errors and bad blocks. I use the Read-only scan test in Hard Disk Sentinel Pro to check the surface of any suspect drives.

1st Player isn't a PSU brand I know, unlike Seasonic, Corsair, EVGA, etc. You might be able to find it in a list or review online. For a decent 700W PSU, I'd expect to pay at least $100US for a good quality unit. If you paid less than $50 for a brand new PSU, I'd treat it with caution. I once bought a cheap 700W PSU, but it died after a year when I plugged the mains lead into the back. Nowadays I stick to mid-range Corsair PSUs between $80 and $160. Some of my old computers have very cheap PSUs which are more than 10 years old but are still working (on borrowed time). I've re-capped several PSUs and motherboards when I see the tops of electrolytics bulging with brown goo oozing out. I'd caution against opening up a switched mode PSU unless you know what you're doing. There's a good reason why the main bulk capacitor is rated at 420V DC. You have been WARNED!

I agree your problem does not appear to be video card related, but XHCI mapping errors are outside my experience. I have the benefit of a large collection of working computers, ranging from an ancient Intel Pentium 4 (socket 478) circa 2003, to a modern AM5 7950X, so I can swap components around until I locate the fault. If you can pick up a cheap socket AM4 mobo on eBay, it might isolate the fault. Unfortunaely, the only way to be sure is to spend some hard-earned cash.

Thank you for so much educative guide which is based on your experience! I really learned many things. Today I went to IT service center for diagnostic and the guy put 2 Gskill 2133mhz rams instead of mine. As I already wrote here, I can easy try to test those restarts when entering a 4 gb folder with videos and trying to set large size - my rams restart on every try, but with his Gskills it didn't restarted! and I did try many times to be sure. Just it was weird at first try with my rams (I wanted to show him the problem) it also didn't restart, but on second try - it did. After that as already said we tried with Gskills many times and never got restarts. So, I think that this should mean that 2 of this particular made of modules are incompatible with my mobo, I just can't see another logical explanation. So, now, what I have to do is to buy new ram and I will do buy just one stick and try with on of the mine. I asked him which brand he would advise to buy and he said - Corsairs are working well. So I think I will search either for Corsair either for Gskill (only not with default 2133mhz).

XHCI mapping errors are outside my experience.

As for my rams, I think it would be almost impossible to diagnose why they are incompatible with my motherboard, maybe there are some setting in the bios which can correct this problem for them, that's why I get interested in XHCI controller mapping thing, but I'm not sure what it is exactly and how to find it or similar in my mobo's bios (I will try though).

Thanks again to all for the help!

 

[Harlow]

Excellent news. Well done!

When choosing RAM, if you plan eventually to install two DIMMs, it's better to buy both of them at the same time in a "matched pair". If you buy a single DIMM to check it's OK, then buy a second supposedly "identical" DIMM later on, it probably won't come from the same batch and it might have subtly different primary. secondary and tertiary timings. I have a bunch of seemingly identical Corsair DDR3 DIMMs pulled from old computers and they have three different sets of XMP timings.

Usually, mis-matched timings don't affect performance, but when the difference between DIMM timings is large and the BIOS selects tighter timings based on the faster DIMM, instability can result because the slower DIMM is being overclocked. It then becomes necessary to disable automatic BIOS timings and apply manual timings which can be a pain for the uninitiated. Play safe and buy a matched pair. You can always RMA them if they don't work.

As for speed, remember that most DDR4 DIMMs are programmed with the JEDEC standard speed of DDR4-2133 or 2400. These low speeds are guaranteed to POST (boot up) with virtually any CPU/motherboard combination. If you pay extra for faster DDR4 RAM, you get additional XMP settings guaranteed to work with Intel CPUs, e.g. 2666, 2933, 3000, 3200, 3400, etc. The higher the maximum XMP speed, the more you have to pay. Normally it's OK to use Intel XMP speed settings in an AMD motherboard, but at very high speeds, it's better to buy RAM that's qualified for use on AMD systems. You don't need to worry at sub DDR4-3000 speeds but above DDR4-4000 on a compatible mobo, things can get tricky.

When you invoke XMP, the BIOS raises the Voltage on the DDR4 memory from the JEDEC standard of 1.20V to 1.35V (or even higher) and increases the memory speed. N.B. You need more Volts for reliable operation at higher speeds. The extra Volts mean the DIMMs get hotter, but you're unlikely to detect much difference in RAM temperature in a well ventilated case. I have one system with 4 tightly-packed DDR3 DIMMs in a server where I've added a dual fan RAM cooler because the RAM was hot to the touch.

Good luck.

 

[The King]

Are both your dimms the same, meaning a kit?

Please post Zentimings. Sometimes all it takes a slight increase in VDIMM or VSOC to get stability on some RAM kits when running dual channel

 

[amrgul1]

When choosing RAM, if you plan eventually to install two DIMMs, it's better to buy both of them at the same time in a "matched pair". If you buy a single DIMM to check it's OK, then buy a second supposedly "identical" DIMM later on, it probably won't come from the same batch and it might have subtly different primary. secondary and tertiary timings. I have a bunch of seemingly identical Corsair DDR3 DIMMs pulled from old computers and they have three different sets of XMP timings.

Usually, mis-matched timings don't affect performance, but when the difference between DIMM timings is large and the BIOS selects tighter timings based on the faster DIMM, instability can result because the slower DIMM is being overclocked. It then becomes necessary to disable automatic BIOS timings and apply manual timings which can be a pain for the uninitiated. Play safe and buy a matched pair. You can always RMA them if they don't work.

As for speed, remember that most DDR4 DIMMs are programmed with the JEDEC standard speed of DDR4-2133 or 2400. These low speeds are guaranteed to POST (boot up) with virtually any CPU/motherboard combination. If you pay extra for faster DDR4 RAM, you get additional XMP settings guaranteed to work with Intel CPUs, e.g. 2666, 2933, 3000, 3200, 3400, etc. The higher the maximum XMP speed, the more you have to pay. Normally it's OK to use Intel XMP speed settings in an AMD motherboard, but at very high speeds, it's better to buy RAM that's qualified for use on AMD systems. You don't need to worry at sub DDR4-3000 speeds but above DDR4-4000 on a compatible mobo, things can get tricky.

When you invoke XMP, the BIOS raises the Voltage on the DDR4 memory from the JEDEC standard of 1.20V to 1.35V (or even higher) and increases the memory speed. N.B. You need more Volts for reliable operation at higher speeds. The extra Volts mean the DIMMs get hotter, but you're unlikely to detect much difference in RAM temperature in a well ventilated case.

I see, thanks.

I have one system with 4 tightly-packed DDR3 DIMMs in a server where I've added a dual fan RAM cooler because the RAM was hot to the touch.

I also plan to add heatsinks to them when I resolve this problem.

Are both your dimms the same, meaning a kit?

Please post Zentimings. Sometimes all it takes a slight increase in VDIMM or VSOC to get stability on some RAM kits when running dual channel

Both identical, Patriot Signature Line 2*8gb PSD48G266681

 

[VuurVOS]

Is everything memory releated on auto in the bios or did you set this by hand?

Can you also make a photo of the ram modules?

 

[The King]

What is your stock VDIMM voltage 1.2V? Try increasing by 0.05 . So 1.2V will be 1.25V.

I would also try to increase you VSOC to at least 1V from 0.8688V it can help with stability issues.

 

[amrgul1]

Is everything memory releated on auto in the bios or did you set this by hand?

Can you also make a photo of the ram modules?

I didn't change anything, bios is on its default

Can you also make a photo of the ram modules?

Sorry, I can't do that right now. Why do you need that? I can write by hand any of the codes on it if yo want, if you suspect they might be used and botched, I can tell that they were new in their boxes and no harm when I looked at them visually.

What is your stock VDIMM voltage 1.2V? Try increasing by 0.05 . So 1.2V will be 1.25V.

I would also try to increase you VSOC to at least 1V from 0.8688V it can help with stability issues.

Yes, stock voltage is 1.2 in the bios. Ok I will increase them and try.

 

[VuurVOS]

Sorry, I can't do that right now. Why do you need that? I can write by hand any of the codes on it if yo want, if you suspect they might be used and botched, I can tell that they were new in their boxes and no harm when I looked at them visually.

I just want to know which chips are on both modules. The numbers listed on the memory chips will tell us more what the capabilities are.
This is a better approch instead blindly increasing voltage and memory clock speeds.

Yes, stock voltage is 1.2 in the bios. Ok I will increase them and try.

Please keep voltage at 1.2v until we know more about the memory chips.

 

[amrgul1]

I just want to know which chips are on both modules. The numbers listed on the memory chips will tell us more what the capabilities are.
This is a better approch instead blindly increasing voltage and memory clock speeds.

I can't do photo because those writings on the chips are very small and my webcam can't focus at all. Maybe if I buy magnifying glass, then I could do it. Sorry about this, if you give me a day, tomorrow I try to buy one.

 

[VuurVOS]

If a photo is not possible, just write it down and post it here.

 

[amrgul1]

If a photo is not possible, just write it down and post it here.

I also can't read

 

[The King]

Yes, stock voltage is 1.2 in the bios. Ok I will increase them and try.

I gave you those voltages based on a review of a similar RAM kit here running on a 3700X.

Patriot Signature Line Premium 32GB DDR4-2666 Memory Kit Review - Overclockers

A closer look at the Patriot Signature Line Premium 32GB DDR4-2666 dual-channel memory kit on the latest AMD Ryzen platform.

www.overclockers.com

You can find out what IC you have using CPUZ or Thaiphoon Burner which was used in the above review.

You could also try lower increments like 1.21V or 1.22V etc...
Sweet.

 

[VuurVOS]

I gave you those voltages based on a review of a similar RAM kit here running on a 3700X.
View attachment 285380

Patriot Signature Line Premium 32GB DDR4-2666 Memory Kit Review - Overclockers

A closer look at the Patriot Signature Line Premium 32GB DDR4-2666 dual-channel memory kit on the latest AMD Ryzen platform.

www.overclockers.com

You can find out what IC you have using CPUZ or Thaiphoon Burner which was used in the above review.

You could also try lower increments like 1.21V or 1.22V etc...
Sweet.

Using Thaiphoon Burner or CPU-Z is not a guarantee it will be correct.

I have PNY modules which dont have that information even listed in the SPD (its undefined).
The reviews are telling me that it should be SK Hynix CJR but the modules I have are actually Micron Rev E.

Visual inspection is the best thing you can do when there are no heat spreader on the memory. If the IC's aren't rebranded then you have 100% guarantee to know which IC are used.

 

[amrgul1]

I managed to read the numbers and also made screenshots by my camera to look visually, because I'm already not sure that it was really unused.

These are the numbers on the 4 chips on the right side (on the left side sticker covers other chips) and on all of them they are identical including second stick:
PATRIOT
PM(maybe-W)1G8D4BU-266
2213EA0303

Also on the back side of the board there are some numbers:
BP 4M-1 E186014
94v-0

Also on the top side of the board:
B84URCA2 HF 2.10

And finally on the sticker:
PSD48G266681
9DS00208
DDR4 8GB 2666MMHz CL19 1.2v
RoHS Compliant
Under the barcode: 9DS208-80794

What is your stock VDIMM voltage 1.2V? Try increasing by 0.05 . So 1.2V will be 1.25V.

I would also try to increase you VSOC to at least 1V from 0.8688V it can help with stability issues.

I can see VDIMM voltage in the bios but can't find which setting is for VSOC. I found - "SOC Overclock Vid" and its value is set to - 0. Is this VSOC?