PC Drivers & Software

[forgemaster]

System Setup:

Software:

Windows 10 x64 1709 Build 16299.15

Amd-chipset-drivers_18.10b

Hardware:

AsRock AB350M Pro4 - BIOS Revision 4.70

AMD Ryzen 5 1600

Corsair CX550W

Corsair Vengeance LED 16GB - Samsung Die

My system has been crashing a lot more often than before, especially when drivers are getting stressed. Problem is, the only way to get memory dumps is to just run it. I tried the integrated driver verifier, and it lets every driver to boot into Windows except AMD Chipset drivers, they deadlock the system as soon as Windows tries to boot with them using the checks, which leads me to believe there's something wrong with either the drivers or the newest BIOS revision for the motherboard.

A random crash may happen when using multimedia, internet browsing or game applications. It seems to crash more often when I stress the system with gaming, in particular Skyrim Special Edition which crashes with a different memory address every time with no pointers on which module or driver is causing the crash. The crashes get worse if the system is overclocked, adding more voltage doesn't alleviate the crashes.

I have found 0 remedies so far, even a fresh Windows install didn't help. Running the system completely stock with newest drivers installed still causes random crashes pointing to Windows kernel, and never a driver or a module.

Is there anything else I can try at this point? As these random BSODs have been plaguing me for 2 weeks now.

[sammyreynolds1977]

If you're getting a kernel error try going into your power options and turning off fast startup.

[forgemaster]

Wouldn't the Ryzen Power Plan already do this? And it should work perfectly fine for most people with it turned on, no? It's not just a problem with the kernel, anyway I realized when I've written "Running the system completely stock with newest drivers installed still causes random crashes pointing to Windows kernel" seemed rather vague.

Here is one of the minidumps created by windows just before it restarted: Instant Online Crash Analysis, brought to you by OSR Open Systems Resources, Inc - Pastebin.com

I get various crashes with different Windows crash codes, such as:

KMODE_EXCEPTION_NOT_HANDLED

SYSTEM_SERVICE_EXCEPTION

APPLICATION_FAULT_NULL_CLASS_PTR_READ

Unfortunately I did not keep any of the older minidumps, they were lost when I re-installed Windows after formatting. Creators update was recently shoved in my computer after I was happily running the older build which worked OK with older drivers. And was having no issues when AM4 was only a month old, my early adoption went smoother than most others. But now the system keeps crashing when it is in more heavy scenarios, on full load I'm barely going past half of what my power supply can output on a stress test, so it can't be the problem.

I will try to do as you said, I don't remember if I had fast startup disabled on my previous windows version. I vaguely remember having it off. My memory is a little fuzzy because I booted the system after 3 month hiatus to find that a bunch of updates installed, so OK, I download the newest drivers for the system to be compatible, then all of this started to happen.

[kingfish]

Windows 10 Fast Startup could be detrimental to your computer

"Faster -- like bigger -- is better, right? Not necessarily. Windows 10 includes a feature called Fast Startup which certainly sounds enticing. Microsoft's description in the Control Panel says, "this helps start your PC faster after shutdown". Who wouldn't want a slice of that pie?

Enabled by default on most clean installations of Windows 10, Fast Startup essentially hibernates the Windows kernel and loaded drivers to help slash startup times. It all sounds great, but Fast Startup may not be quite as good as you think. Not only can it seriously mess up dual-boot systems, it may also interfere with the installation of updates, and could cause problems with encryption software.

Over on How-To Geek, Walter Glenn reveals the pros and cons of this feature of Windows. Everyone's time is important, but you may well decide that the few seconds you can save booting up your computer are just not worth the price you have to pay. One of the more worrying side effects of Fast Startup is that some people have found encrypted drives are automatically mounted after a restart, potentially posing a security risk.

There is also the potential for Fast Startup to get in the way of Windows Updates. While updates will be downloaded for you if you have Automatic Updates enabled, not all of them can be installed without a restart. If you shutdown your computer with Fast Startup enabled, your machine will not perform a complete shutdown, so the installation of updates may not complete. Opt to perform a restart, and this will incorporate a proper shutown and allow updates to be installed -- something to keep in mind if you want to ensure that you keep your copy of Windows 10 fully up to date."

[elstaci]

What GPU card do you have installed?

When I had one of four RAM Module that was defective, I was getting mainly, plus some dealing with memory,  the first two errors you mentioned - Kmode and System Service errors.

Every time I tried to install something it would crash or when my computer was under load or when Updating to the newest version of Windows.

I would run MEMTEST86 just to eliminate bad Memory module. After two passes it should show Zero errors. one or more errors indicates a bad memory module.

[forgemaster]

I've run memtest and it would pass, but on a bad overclock there would definitely be errors popping left and right.

[elstaci]

I not sure if MEMTEST86 works on Overclocked RAM or not. But either way you shouldn't get any errors at all.

Remove all your RAM except one and run MEMTEST86 with the same condition as when it had many errors and see if it passes with Zero errors. If it does pass with Zero errors install the others and do the same until you find which RAM module is acting up during OC.

If MEMTEST86 is showing errors left and right it means that what is going into RAM is not what is going out of RAM thus causing your crashes.

[waltc]

So what happens when you switch the chipset drivers from 18.10b (I am using 18.10b on my x370 mboard, Win10x64, version 1803, build 17133.1, R5 1600 without a problem) to the standard Windows drivers--eg, simply switch from the Ryzen "balanced" power plan to one of the standard Windows power profiles--and reboot--making sure that fastboot is not activated?

If that doesn't work it would appear you *might* have more than one problem, probably, and it's hardware--either your PSU is under-amped on the 12v rail (that would correlate with the system crashing under stress)--and the problem with the chipset drivers might be related to a faulty DIMM--possibly even both.

It would appear from what you have written that you leave your machine running for months at a time without attention?   That must be the case as had you powered it off during that time it could not have updated anything in your system, of course.   Very poor thing to do--leaving a system running 24/7 unattended simply cuts the time between hardware failures, fan motor failures (another thing to check), etc.  If you are overclocking anything clock back to stock, and try downloading your mboard's chipset drivers directly from the  manufacturer's web site and using those. One last suggestion--if you use sleep mode at all--disable it entirely, and then reboot and see what happens.

I think that somewhere in these suggestions you will find the culprit...;)

Edit: one last comment, you should be getting GSODs instead of BSODs, IIRC...;)

[forgemaster]

I'm not getting GSOD's I wasn't even aware of such thing until I looked it up. It's only for Windows 10 Insider Builds.

The system wasn't running unattended, it was simply taken apart completely until I was ready to build it again after I had a faulty motherboard (front panel sound problems). Windows 10 updated after I was trying to boot into my old WIndows. But that's the past now, because I did a clean install, and that didn't help.

The power supply has been more than enough, in fact if I really wanted I could run the fastest GPU on it without a hitch though I would be running into its limitations.

I don't use sleep mode, ever. Because it has been proven to reduce performance when the system "wakes up" until the next reboot. Hibernation is a no-no as well.

I've disabled overclocks because as I've said the system failures become more frequent. I'm concerned about the BIOS being finicky, because the last stable revision was version 3.10, as recommended I updated to 3.40 first and then applied the last update (4.70). I updated in hopes for better stability, but it seems to be proving to run much worse than the old version or it might just be placebo because I have done so many things at once and I'm not sure what is the root of the problem. I've been unable to diagnose it with standard procedures.

Running 3.7GHz w/ SMT disabled + 3200MHz XMP on RAM was going well until a combination of factors got involved:

1) I changed out for an identical motherboard, which seems to be running worse?

2) I updated the motherboard's BIOS.

3) Windows 10 updated to Creators Update.

4) I installed the newest AMD Chipset drivers, and I don't remember which ones I used before. (not that a roll-back would help because of the last Windows 10 build, mixing old drivers with new 10 builds is not a good thing)

Now I can't even run the system stock.

One or more of these could be a problem, and now I have no idea which one, because it never occurred to me that I would have stability issues so I never tested before-hand. Usually doing these one-by-one and testing I would figure out what went wrong. I'm really frustrated at this point.

[sammyreynolds1977]

time to take it to a tech

[forgemaster]

Yesterday night I was testing a few things, running with Hibernation and Fast Startup forced off seemed to help, then ran through some Windows updates. And some programs had compatibility mode applied, so it doesn't use as many Windows 10 useless trash"features".

I was able to run the system OK, the games still crashed but it took them much longer. I'm still getting memory violation exceptions and I have no idea why programs can't read and/or write to memory at certain points (running in Admin mode doesn't matter). It's like something else keeps reserving that same memory space, which shouldn't happen. Maybe it's just bad game coding/optimization as I suspected.

I didn't try reverting my changes and force a crash due to time constraints. I'll investigate further tonight.

[sammyreynolds1977]

Sorry like bad memory

[mrperforations]

hello,

       I was wondering could this be to do with installing AMD™ ramdisk™ software please?

• i think seem to remember it leaving a setting file behind or something?

Best regards

    Mr Perforations

[mrperforations]

note.

I'm using the Adata™ 60gb ssd write now.

and the three little drives two as one stools every now and then.

[elstaci]

Good  advice about the Hard drive. A bad Hard drive can cause data corruption. But he mentioned that when the RAM was Overclocked he got many errors which indicates a bad memory module. But I am not sure if MEMTEST86 is reliable when the RAM Memory module is overclocked. I would believe that it shouldn't have any errors, whether it is overclocked or not.

I would suggest installing only one RAM module. Testing it with MEMTEST86 to be sure it passes without errors for at least two passes. Than try to use the computer and see if it crashes. If it doesn't, then install another RAM Module. Run MEMTEST86 again. If it passes, than continue to use the computer. If no crashes than install the other sticks. If it starts crashing, then he will know it is one of the last two sticks that he installed last that is bad.

By the way, I see you are using a Ryzen 5 - 1600 CPU with the B350 Motherboard. Have you checked to see if that Memory is on the Motherboard's Support list for Compatibility for using a Ryzen CPU?

In another thread, an OP is having problems with his Motherboard because it seems like he was using RAM Memory that wasn't totally compatible with the Ryzen CPU and Motherboard. He wasn't having any crashes thou. Just wasn't showing the full memory in BIOS.

just an observation.

[forgemaster]

EDIT: TL;DR, I managed to get everything up and running with basic testing. I haven't tried any real-world gaming yet to see if the application crashes are as often as they were before.

Things that helped were:

1) God-tier level of patience

2) Letting Windows do it's thing with updates

3) Disable Hybrid Sleep/Fast Startup and Hibernation

I'm running 4 brand new drives, 2 of them are SSD's. They pass with flying colors in any test. I'm using MEMTEST x64, which is more reliable, it makes the system throw more errors because on a typical scenario you will be running Windows and stressing the system out. It makes the CPU go 100% and you can allocate as much memory as you want, and it's not linear, you can have as many loops of testing running. EDIT: All loops now run successfully even with my old BIOS settings.

I've been using my RAM modules almost ever since launch, they are not Ryzen "compatible". And the QVL lists are just peace of mind for newbies who aren't sure what they are buying. My memory sticks are Intel optimized, but ever since AGESA 1.0.0.6 I've been able to run them at their rated XMP profile since day 1 of the BIOS release without a hitch. It's just that now something is breaking down. I don't have any other RAM of similar speed to test, and it's far too expensive to buy a new DDR4 kit just for that.

What might be right is to try and overclock sticks but leave them on memtest one by one instead. But that only leaves the CPU IMC with less work, usually more voltage helps get rid of stability problems, in my case it did not. It's true that Dual-Rank kits weren't even bootable at 2933+ at some point, but that's the past.

My point is something changed software wise or the motherboard I received wasn't doing as much as a good job as I expected, but it shouldn't influence memory overclocks, the CPU silicon quality does, almost every B350 guarantees 3200MHz working. And other people are having no problems running the same DDR4 kit on their AM4 systems.

I put a picture just in case. Don't worry about it having set at 6 threads. I'm running SMT off for higher single-thread performance for un-optimized games. Apparently now the system is running fine again after some time passed. I swear not only are the crash prone games are mocking me by not crashing when I am debugging them, but the system decides to run OK again. I give up. Part of me feels guilty for choosing Ryzen because I essentially lost 10% single-thread performance, which is important for high FPS/Refresh rates. My favorite game series are STILL releasing with engines based on old API's that are limited to quad-cores because most people are still using competition CPUs. The Ryzen 5 1600 is a great CPU, I don't deny that. But I am getting more issues than it is worth, and I was banking on it's longevity for gaming (which it won't fail at, the FX-8350 is a testament to that when playing on high resolutions where the GPU is the bottleneck). The games I play simply prefer clock speed over more cores and that hasn't changed since launch, that's a problem. And upgrading to the next-gen processor is a waste of money, because the big leap was already made for me. I'm in love/hate status with the way things currently are.

Anyway, the rant is over. Seems like resetting/tweaking some Windows settings and getting updates again seemed to bring everything back to normal, sort of. I haven't tested the games but I will this weekend to see how well everything runs.

I will post my final thoughts on the ordeal later. Thanks for reading. These issues aren't widespread and I know I'm the drop of water in a sea, so hopefully I didn't waste too much time, and someone reading this might be able to deal with similar problems in the future.