Question Ryzen 7 5700G APU - Freezing and/or BSOD

Page 2 - Love gaming? Join the PC Gamer community to share that passion with gamers all around the world!
Jan 5, 2022
20
7
15
Visit site
Hi guys, new here.

I come in search of desperate help. Just built my PC a month ago, and can't seem to get it right. There's definitively something wrong and I don't know what else to do.

It will randomly freeze, sound keeps going, and after some seconds screen goes black, and complete lockup. Will do it every couple of hours at least.
Sometimes it will go BSOD too, or tabs in Firefox crash randomly. Very, very unstable.
The lockups happen especially when gaming or doing 3D rendering stuff.
I have tried with W11 and W10 with no noticeable difference.

Specs:
APU: Ryzen 7 5700G
MB: Asus ROG Strix B550F Gaming WiFi
RAM: (2x8)16Gb Corsair Vengeance RGB PRO 3600Mhz
PSU: Corsair RM650W 80+Gold
SSD: Samsung 980 Nvme 500Gb

Corsair H100i Capellix AIO 240mm
Corsair 4000x Chassis

So far I have tried the following with no success:

- Reinstalling SO (W11 or W10)
- BIOS Update from 2403 to 2423
- Enable/Disable DOCP to 3600
- Manually entering 3600 Mhz / 1800 FCLK
- Enabling / Disabling Precision Boost Overdrive (PBO)
- Installing drivers from AMD and from Asus, different versions. Still having lockups.
- With and without Armoury Crate
- With and without iCue
- Reseating PSU cables (x3)
- Changing RAM modules (installed in A2/B2 as per MB instructions)
- Using a single stick of RAM. (This gave me hope as I could play for about 1 hour)
- Tightening/ Loosening AIO pump mount screws
- Tightening / loosening MB screws
- Cleared CMOS
- Set SOC Voltage to 1.05
- Disabling C-States
- (Everyone says to change Power Supply Idle Control to Typical, but this is a setting I can't seem to find in my BIOS)
-Reinstalling RAM, APU, checking MB connections.
-Can't really remember what else I have done so far.

Benchmarks passed without any issues, and with max 75C for the CPU package during the most demanding tests.
- Heaven (with and without OC)
- Cinebench R23
- OCCT
- MemTest86 (x4 times. Twice with both sticks, and once for every RAM stick)

I have read most of (if not all) of the forums and topics covering similar issues.
I am afraid there's something wrong with my hardware, but don't know where to start to get a clear diagnose.

Guys, I really need your help.
I can get similar sticks of RAM but with different timings, I would change from C18 to C16 which should be better. But can't really tell if this is the issue.
Do you think it's the RAM? APU? MB?

Your help is greatly appreciated,
And if you just read all of this, thanks!

[Mod edit: changed title from "Desperate Help! Ryzen 7 5700G APU - Freezing and/or BSOD".]
 
Last edited by a moderator:
  • Like
Reactions: Brian Boru

Another guy who passed memtest but RAM turned out to be the problem recently.

I've seen it on several occasions over the years across different platforms both Intel and AMD, that's why it always sets off alarms to me when someone rules out RAM issues because of Memtest.

HCI is what I used to tes the overclock the RAM on my 7700K a few years ago, I dont have that kind of time for stress testing anymore so I'm out of the loop on overclocking newer platforms, but I do know a few tricks here and there :)
 
  • Like
Reactions: Brian Boru
Another guy who passed memtest but RAM turned out to be the problem recently.

yeah, memtest isn't perfect but its pretty good. I have helped people where ram tested fine but replacing it still ended up being fix.

It is why running Prime can help as sometimes you get rounding errors and they can be fixed by giving more voltage to ram. But I won't get ahead of the results.
 
  • Like
Reactions: Brian Boru
yeah, memtest isn't perfect but its pretty good. I have helped people where ram tested fine but replacing it still ended up being fix.

It is why running Prime can help as sometimes you get rounding errors and they can be fixed by giving more voltage to ram. But I won't get ahead of the results.

Prime 95 works! But I've heard mixed messages as to whether Blend or Large FFT's is better for RAM. I guess Blend catches some of everything so its a better place to start. If you know which FFT's are causing the issue you can just cycle straight to those to get faster results.

Sounds like OP already bumped SOC voltage a little, so DRAM voltage up to 1.4 might be worth trying too. Pending HCI results, 1000% of 90% RAM over all threads should be pretty comprehensive. But freezes, hangs and browser tab problems sure sounds like RAM still to me.
 
  • Like
Reactions: Brian Boru
2 of the BSOD may be caused by Internet drivers. I would run https://www.intel.com.au/content/www/au/en/support/intel-driver-support-assistant.html and see if it has newer LAN/WIFI drivers

all 3 look like driver errors.
one occurred when all you had running were intel and AMD drivers so we can always update chipset drivers as well - https://www.amd.com/en/support/chipsets/amd-socket-am4/b550
try updating GPU drivers as well.

crash conversion - https://jsfiddle.net/gd2hcubo/show (mainly to show source)
 
Jan 5, 2022
20
7
15
Visit site
Thank you, updating now.
Checking the crash conversion all of them show "Probably caused by : memory_corruption."
Should I take this as a clear sign of defective RAM?

EDIT: Updated drivers. Still the same, crashing with Doom / Dirt 2.0. Freezing when in normal use.
Should I just return RAM, APU and MB for warranty? :LOL::(
 
Last edited:
memory corruption: I see this on a lot of BSOD, I don't feel it always means ram (I don't expect people to read conversions, its just there more as proof)

as the guy who wrote conversion said today
The "probably caused by: memory_corruption" message is probably in about 99% of the dumps. On rare occasions it will say hardware_disk (or something similar to that).

Memory can be multiple things. It can be the memory where the drivers are stored. It can be the RAM. It can be the virtual memory on the hard drive.
I wouldn't get too caught up in that.

You could RMA the ram and get new sticks from same company and see if it fixes it.
 
Last edited:
  • Like
Reactions: Brian Boru
memory corruption: I see this on a lot of BSOD, I don't feel it always means ram

You could RMA the ram and get new sticks from same company and see if it fixes it.

If it were me I'd try raising DRAM voltage to 1.4 in BIOS and see if it helped first. AFAIK some IC's get less stable at higher voltage so it might not work, but it wont blow anything up to try and if its stable it will run fine at that voltage for longer than the life of the RAM. Ran mine at 1.4 v for 4 years with no problems.

Also QVL list specifically mentions the part number and version 4.31 for that RAM. Can it be that they changed IC's on that part number at some point in the run and the sub timings are slightly different on other versions?
 
Jan 5, 2022
20
7
15
Visit site
Well guys, I managed to RMA the APU and the RAM with Amazon.
New 5700G should arrive next week or maybe the weekend if I'm lucky.

I'll change the RAM too, just to be sure. But, still can't decide between these two.

Amazon.com: Corsair Vengeance LPX 16 GB (2 x 8 GB) DDR4 3600 (PC4-28800) C141.45V AMD Optimized Memory -Negro memoria : Electrónica

Amazon.com: G.Skill Trident Z NEO Series 16 GB (2 x 8 GB) 288 pines SDRAM (PC4-28800) DDR4 3600 CL14-15-15-35 1.45 V Dual Channel Desktop Memory Model F4-3600C14D-16GTZNB : Electrónica

Should I just throw some money at it and get the GSkills? :devilish:
Both are QVL approved and CL14. It seems like both of them are Samsung B-Die.

Which would you choose?
Thanks again!
 
Just checking, as designated spokesperson for the ignorati:

I can't find info on that site which is not at least a decade old. Not that I looked everywhere, of course—but still… are you comfy the Prime Directive is still suitable for this mille cent decade?

I think the Blend test without AVX enabled is still a good test of RAM stability.


It used to be 24 hour Prime 95 stable was considered to be as close to 100% rock solid CPU overclock as it gets. These days I think its considered better to run for a shorter time as part of stability testing with a other tests, and the main test without AVX enabled is good for soak testing for maximum 'normal' CPU temperatures to test cooling.
 
  • Like
Reactions: Brian Boru
Jan 5, 2022
20
7
15
Visit site
Hi guys! Changed CPU for the new one...

Still having same behavior, random BSOD and I also got error message with a crash after 5 mins of playing Doom.

Pretty much convinced it's the RAM at this point. Hope it arrives this week!
 
  • Like
Reactions: Brian Boru
Have you tried setting RAM to default 2133 and everything else ram related to auto? Doing this will rule out faster ram timings and frequencies being the issue. If it's still unstable after that and you try different ram sticks then there's a motherboard yield issue more than likely.
Am I that dumb? I mean it can easily be.

Had it in my head that if the sticks werent on the QVL they might not train anyway even at JEDEC speeds. But somehow now you mention it feels like I might not always have thought that. :oops:

If I've been wasting @Juanchoyos time completely, then I'm sorry.
 
  • Like
Reactions: Brian Boru
Jan 5, 2022
20
7
15
Visit site
LOL

I was just talkin with a friend on the phone and told me exactly the same. As fas as I remember it still crashed without DOCP enabled. But I'm not 100% sure. I'll disable it and try at stock speeds tonight, we'll see.

New G-Skills are on the way. QVL approved in both the AMD and ASUS lists.
 
  • Like
Reactions: Kaamos_Llama

TRENDING THREADS