Just updated my system from a Ryzen 1700 to a 3700X. Brand new MSI X570 Gaming Edge Wi-Fi motherboard, 16 gig Gskill 3200 memory, Ryzen 3700x with Coolermaster 240 AIO. Fresh install Windows 10 on Samsung EVO Pro NVME ssd. I'm using my MSI RTX 2080 and Corsair HX850 Pro power supply from the 1700 system. The new system keeps randomly rebooting whether gaming or watching YouTube. Can go for days then all of a sudden it just restarts. No BSOD, no info in event logs except a Kernal 41 (63) error. Its not a temperture problem as the the cpu never goes above 75 under max load all cores running prime 95.. Idles around 45-50. Have stress tested for hours running Heaven, Prime95, Heavy Load etc without a single problem then I could be surfing browsing the web or watching a YouTube video and it will suddenly crash restart. Reading other posts with the same issue, people have suggested its a power supply problem. So I swapped out the Corsair for a Brand new Thermaltake 750 Gold and still the same problem. My Corsair never gave me any problems in my 1700 system so I doubt its a power supply fault. Board has latest BIOS running default settings apart from memory xmp. No overclocking and all drivers up to date Windows version 1909. I ran memtest overnight, no errors. This is driving me nuts, cant seem to find a cause.
almightynasty, please do a Clear CMOS. This should turn off XMP. Please run that way and and let's see what happens. Also look at the Details (lower pane of your Kernel 41 error.) and tell us what the BugCheck code is. My system gives a 26 (Decimal) code. Look at several, please. Also post a screenshot of Ryzen Master (RM) - simply drag-n-drop the image into your reply. Thanks and enjoy, John.
first thing I recommend is a BIOS update, my X570 has had several updates since I bought it back when they first surfaced
that Corsair HX850 is a respectable PSU, there should be no problem with it
Another confused/angry 3700x user.
I'm having stability issues as well. I assembled the damn thing, updated bios to F11 and installed fresh W10 Pro last week. I left everything on default bar XMP profile. Everything went well, I did some 3-4 hours gaming sessions.. cool, great. Went to sleep and next morning the damn thing restarted. OK... and than during the day again, some bsods, some restarts. I reseated RAM, GPU and NVMe drive. Same thing. Cleared CMOS and left everything on default. Same problems, especially during idling during the night.
Then I ran OCCT for 6 hours with bsod error - system thread exception not handled/page fault in nonpaged area.
I also tried:
- fresh windows with just amd chipset drivers and default nVidia W10 driver - crashfest (almost always kernels)
- xmp on or off - it crashes even on stock setting
- xmp off with manual voltage and timings as recommended by ryzen calculator - crashfest
- different bios setting - pbo on/off, core boost on/ff, csm on/off, different psu settings... - crashfest
- ran memtest86 for 10 passes with xmp profile on - no errors
- swapped RAM with G.Skill F4-3200C16D-16GVKB... same issues.
- swapped GPU (gigabyte 2070S with KFA2 2070S)
Today I tried with high and medium LLC and it wouldn't even boot to windows. BSOD. Then I set defaults in bios again and switched CPU VCORE from Auto to Normal (didn't touch offsets). It ran 24 hours without a problem. Then I enabled XMP with profile 1 - issues. Switched to profile 2 and set DRAM voltage to 1.40. Testing OCCT for about 5 hours now without issues. The only issue I had was in Prime95 (26.6 and latest version) - rounding was 0.5 expected 0.4 on workers 15 and 16.
Now I really don't know what to do anymore... I'm thinking of reverting to BIOS F4 with agesa 1.0.0.3 ABB.. Or should I not touch the BIOS again? Is this common, should I RMA CPU and MoBo...?
The funny thing is, I also assembled another ryzen. (3600, MSI Gaming Max (tried with both mentioned RAM, default latest bios + xmp.. and it runs great..)
Samsung 970 NVMe on first m.2 port, another Samsung 860 1TB on SATA and WD black 2TB FZEX on sata.
GIGABYTE X570 AORUS ELITE, BIOS Ver: F11
GIGABYTE GeForce RTX 2070 SUPER WINDFORCE 8G - swapped with KFA 2070 SUPER.
Ryzen 3700x
600 Watt be quiet! Pure Power 11
G.Skill F4-3600C16D-16GTZ - swapped with G.Skill F4-3200C16D-16GVKB
Win 10 Pro
Thanks AMD for great, fun and productive hollidays.
I am suspecting your PSU might be able to handle the workload
I use Corsair which are very respectable. My HX1000i has so much capacity I can ignore it and do what I need with my rig.
I don't get it sorry. My PSU is okay or not? I was under impression that even 550W PSU (lower quality than PP11) are fine. I don't or intend to OC.
usually for a single elite gaming card a 650W PSU should be the minimum to have ample reserve for fans, extra RAM and hard disks etc
a PSU is more efficient at 50% load
OK, I understand that.. but I really think 600w PP11 shouldn't be to blame for idle BSODs.. ?
So yesterday I downgraded BIOS to F10, set everything to default (except XMP, PciE set to gen3 (it supposed to be better for Samsung 970..) and silent mode for CPU stock cooler)... Then I ran OCCT Linpack for 6,5 hours, Prime95 26.6 for an hour and Cinebech R20 for an hour and a half. No issues, except with latest Prime95 (29.8 b6) which gave me BSOD after 10 minutes. I set 1.35v manually for RAM and then it ran for 15 minutes but workers #15 and #16 had issues again (FATAL ERROR: Rounding was 0.4760124369, expected less than 0.4 Hardware failure detected, consult stress.txt file).
Let it idle through the night when BSOD happened again hour or so when I left.
I've rebooted, enable CPU VCORE to normal again, disabled PBO... aaand.. mouse froze in Windows. Keyboard was running fine. Rebooted, ran Prime95 29.8 again and complete system freeze after 8 minutes.
If this is normal behaviour than I'm a Pope. I've overclocked my previous 6600k @4.7 Ghz and never had an ounce of problems during OCing as I have with this stock 3700x.
This ain't normal. I'll RMA CPU and mobo.
drivers are a constant problem in my shop
I use a 1000W PSU as I often have dual cards installed etc
I agree regarding drivers. But I haven't installed almost nothing. Just Win + AMD chipset driver and GPU driver. Funny thing, I just ran Prime95 in linux. Same s.... - cores #15 and #16 - FATAL ERROR: Rounding was 0.4760124369, expected less than 0.4 Hardware failure detected, consult stress.txt file
mviper wrote:
I agree regarding drivers. But I haven't installed almost nothing. Just Win + AMD chipset driver and GPU driver. Funny thing, I just ran Prime95 in linux. Same s.... - cores #15 and #16 - FATAL ERROR: Rounding was 0.4760124369, expected less than 0.4 Hardware failure detected, consult stress.txt file
Depending on the video card older drivers may be more successful
I have also experimented with RAM timings which radeon drivers are unusually sensitive to
Some have used old 17.2 drivers in a bid to get around BSOD problems
Well... I also let system run at nvidia drivers, that windows found 432.00 I think.. I also tried 431.68.. But this is reeeeaaaally weird to have completely unstable system and BSOD due to driver version. What are we beta testers in 1992? I would understand errors, stutterings, weird fps, game crashes but not BSODs
mviper wrote:
Well... I also let system run at nvidia drivers, that windows found 432.00 I think.. I also tried 431.68.. But this is reeeeaaaally weird to have completely unstable system and BSOD due to driver version. What are we beta testers in 1992? I would understand errors, stutterings, weird fps, game crashes but not BSODs
441.66 is the current driver for Nvidia cards, 19.12.3 is the latest for AMD cards
I know.. You're not reading what I wrote. I tried 432, 431.68, 441.66.
mviper, I would like you to know that this is a user forum and we seldom see an AMD employee here. Please share your anger with them: Online Support. Thanks and enjoy, John.
I apologize if I offended anyone, I'm just so frustrated. I apologize again.
Hi folks, I solved my issue after installing a an EVGA 1000w power supply. . I had tried everything i could think of, swapped out everything bar MB and CPU. Was gonna` RMA both. Decided to give the power supply theory another go, swapped in the EVGA from another system and haven't had any more crashes or reboots, system now rock stable. Couldn't believe it, thing is both previous power supplies now work flawlessly in other Intel I7 systems.......weird
Wait wait wait... You've already had 850w PSU and you had problems. Now you've tried with 1kw PSU and it works? Was 850w faulty or what?
almightynasty, I too have seen the 0 code and do not know what it means. Did things get any better with XMP off? My only suspect is memory. W10 has a memory diagnostic, so please run it. Please let us hear. Thanks and enjoy, John.
mviper, your power supply is too low power. I agree with hardcoregames™. 1000 Watts may be a little over, but it is dependent on video, memory (& XMP), etc. I would suggest 850 W for your system. I have seen the Prime 95 round off problem and I suspect memory. Please run “mdsched.exe” and see what you get. But at the bottom, you need more power. I am a firm believer in over kill. I am running a 2990WX using a 1250 Watt supply. Please let us hear. Thanks and enjoy, John.
Memory is fine, I ran 10 passes of memtest86 on both of them on XMP and without XMP profile.. 0 errors
Now you got me thinking if PSU is underpowered. But wouldn't system crash during high loads (3-4 hours of gaming (warthunder, battlefield V) went without an issue. Also cinebench and occt runnings. Underpowered PSU would crash out....? I'm running 2070S on 2k, 144 display..
Funny these psu calcs... one tell me 390W, next one 450, an another one 530
bequiet recommends me:
DARK POWER PRO 11 650W..
Hm.. hm.. hmmm....
Thanks, mviper. In my opinion, it is. These PSU calculators are all over the place. Save a few bucks on a PSU and have an unstable system. Your Prime95 is a loaded failure (probably memory due to a sagging 12 Volts), other than that I cannot explain. Ryzen is a VERY responsive processor - much more than predecessors. So perhaps when a request (from OS) hits, Ryzen pops to it and the voltages sinks. Basically the system runs off 12 Volts - need lots of Amperes. That is why I buy a good PSU, a good MB (lots of VRM resources) and a good CPU cooler. I hope you could at least beg, borrow or steal an 850 W or 1000 W PSU to test. Please check the maximum Watt draw for your Video cards - as hardcoregames™, says that is where the devil is. Good luck and enjoy, John.
Thanks for your explanation.
I'm always for good and reliable stuff and hates to cheap out on stuff that are the core basics. I was googlin for PSU power before I bought all this stuff and there was some king of universal agreement that 550/600/650w was enough but not budget models ofc. So I went with 600w PP11.
You guys changed my mind so I ordered another RM850x PSU. Keen to see the difference.
I ran R5 3600 (2070S) on 620W Seasonic SII PSU and had none problems in two days of toying around. I never ran any benchmarks but never had any BSODs during the night or anywhere. You think it's adequate PSU or may I expect issues in the near future?
Thanks, mviper. How many 2070 video cards do you have? If one card really requires 226 Watts, then make allowance for at least 300 Watts. If the PSU estimator says 650 Watts then make allowance for 800 Watts. You are up to 1100 Watts for only one video card. This is how I do it and 100% my opinion. Just 226+650 Watts is 876 Watts. Perhaps the estimators are already including all components (CPU, Video, Memory, etc), so my opinion is some high. I would much rather be high than low. Thanks and enjoy, John.
Sorry for misunderstanding. I have 2 machines. 1st is X570 elite, 3700x, 2070s, Samsung 970 nvme, Samsung 860 sata, wd black 7.2k sata, 600w pure power 11.
2nd B450 gaming max, 3600, 2070s, adata 8200 nvme, kingston a400 sata, 620 seasonic
I have issues only with the 1st one. Well I have not benchmarked or tortured the second but it ran 3 days 24/7 without a hitch.
I had seasonic 620 laying around and I thought to myself that will suffice for r5 3600 and 2070s....
mviper wrote:
Thanks for your explanation.
I'm always for good and reliable stuff and hates to cheap out on stuff that are the core basics. I was googlin for PSU power before I bought all this stuff and there was some king of universal agreement that 550/600/650w was enough but not budget models ofc. So I went with 600w PP11.
You guys changed my mind so I ordered another RM850x PSU. Keen to see the difference.
I ran R5 3600 (2070S) on 620W Seasonic SII PSU and had none problems in two days of toying around. I never ran any benchmarks but never had any BSODs during the night or anywhere. You think it's adequate PSU or may I expect issues in the near future?
The RM850x should be able to handle the hardware
The Seasonic 620 is not bad but its not the level of capacity I prefer.
I know Seasonic ain't wow but I got it laying there in the drawer, it's around 2-3 years old and figured it'll do the r5 3600 + 2070s fine..
mviper wrote:
I know Seasonic ain't wow but I got it laying there in the drawer, it's around 2-3 years old and figured it'll do the r5 3600 + 2070s fine..
As long as not too much additional hardware is added it should handle it fine.
I often use dual video cards which can be brutal for PSU loads which is the motivation for the big fat HX1000i which has more power than I can afford hardware.
Nah, nothing serious.. single 2070S, additional usb3 and network pcie cards... oh and dvdrom..
Yeah, for dualcards I believe you need small powerplant. How's it in gaming? Insane fps?
mviper wrote:
Nah, nothing serious.. single 2070S, additional usb3 and network pcie cards... oh and dvdrom..
Yeah, for dualcards I believe you need small powerplant. How's it in gaming? Insane fps?
Many old games run at insane frame rates. This the reason for vsync.
This is just for your information only:
The RTX 2070 Super Maximum power consumption around 226 Watts according to Tom's Hardware tech site:
The Ryzen 7 3700X Maximum power draw around 150 watts under heavy Stress test (AIDA65):
That's around 380watts. Why so powerful psus than? Because of more powerful 12v and 5v rails?
You need to realize the above results was from stressing the hardware. So most likely it would never really reach 380 watts between those two in regular computer use such as gaming.
Even though your PSU is adequate for your computer system. It all depends on the 12 Vdc amperage draw from the rail.
But if you stress tested the PSU using OCCT PSU Test and it passed without any symptoms, it does seem to indicate your rebooting is caused by some other reason then low power.
On this current thread here at AMD Forums concerning something similar but not exact: My system reboots after installing AMD driver for Ryzen 2200G
I post this reply today:
From a Windows Tech site about what could cause a computer to reboot constantly:
Hardware failure or system instability can cause the computer to reboot automatically. The problem could be the RAM, Hard Drive, Power Supply, Graphic Card or External devices: – or it could be an overheating or BIOS issue.
Yea, that's what puzzles me. Linpack ran 6 hours twice.. No problems. Let machine run on idle during the night - bsod... BSODs happen on underpowered PSU too?
Found this tech site that gives Troubleshooting tips on how to troubleshoot a computer that randomly restarts: Check these 13 solutions to quickly Fix 'Windows 10 restarts randomly'
This are the 13 steps they suggest:
By no means download any recommended software. Most of the time it is causes more problems than they fix from the above website.
what are the BSOD errors you are getting when it crashes during idle?
Man, I've tried all this stuff already. I ain't no noob either (except for ryzens). It's not the drivers or corrupted windows. Tried that 3 times already. I get bsods on stock setting with pbo off. Even xmp is off.
Usually some kernel issues and random errors... irq not equal, system thread exception, page fault in non paged area, kernel auto boost lock acquisition... name it, got it
memtests found nothing after 10 hours..
It's got to be PSU ir CPU.
Open a AMD Warranty Request from here: https://www.amd.com/en/support/kb/warranty-information/rma-form
See if they suggest to RMA the CPU IF it isn't still under the Retailer's Warranty. Otherwise you need to go through the Retailer's Warranty first.
Thanks man.
I opened alreasy. Waiting for Hollidays to pass and than full action.
Happy and healthy New Year everybody.