Overclock.net banner

21 - 40 of 73 Posts

·
Registered
Joined
·
54 Posts
MoW, I'm at work now. I will check out after 4 p.m.

Deepcuts, can you check the new BIOS? You have the specification closest to my rig (CPU, MOB, RAM)
 

·
Registered
Joined
·
104 Posts
Discussion Starter #22
Trying now.
Don't hold your breath.
 

·
Registered
Joined
·
104 Posts
Discussion Starter #23
Updated to F31h
Clear CMOS via back IO button for good measure.
Entered UEFI and noticed FCLK is 104.4 Mhz. Weird. never seen it so high on auto. Set it to manual 100.
Rest on default.
Problem remains. Windows crashes, Ubuntu live does not even boot.

Set VCORE SOC to 1.000 and I could boot Windows, but crashes under Aida64 memory copy and most games.
Tried almost all combinations of VDDG, VDDP, VCORE SOC, IF, RAM Speed, LLC, etc.
The higher the VCORE SOC, the most likely Windows will not even boot.
No matter the combo, Aida64 will crash on copy benchmark as well as most games. 100% will crash at idle after some time or just internet browsing with WHEA uncorrectable error BSOD.

The only setting that gets this rig stable is setting "Core Performance Boost" to disabled.
With it disabled, I can set everything else on Auto and will work just fine.

On the upside, somehow it seems F31h is very stable when changing various voltages, like VDDP/G.
Up until now, a lot of times when changing anything voltage related, the board will beep twice and reboot 3 times. Somehow F31h fixes this for me.
 

·
Registered
Joined
·
43 Posts
WOW, reading these issues just makes me want to wait. And this is a ~$1200 motherboard in Australia. You'd expect the highest end boards to get the fastest release for fixes. But at least in Australia we are protected by Consumer Laws, basically I would have 0 problems returning the board and getting a full refund if it had these issues. Sadly in other countries you have to contact Gigabyte, here it's not my job to contact Gigabyte, the store you buy from has to deal with them.
I have the same board and live in queensland, am expecting my 5950x in a few days by post. I am relishing troubleshooting as just doing a fresh install with my 3900x has caused nothing but headaches.

What I suggest to everyone here is to check a few things, if youre running a pcie4 card make sure tbt and 10-bit tag support is ON. Also make sure gear down mode is enabled in every option on the BIOS and switch PCIE mode to auto. 10-bit tag support is used in pcie 4 iirc so it is likely if its disabled that would cause reboots.

While on BIOS settings make sure to go into the power config in the (middle tab of the) BIOS and try with HPET (high precision event timer) to OFF to see if it helps. I found turning mine on after changing dynamic ticks in windows helped a lot but if you havent touched that then maybe turning it off will help. I am not sure if its the new chipset drivers as well but I have found my 2080 TI, yes still waiting on 3090 stock D:, is getting a lot of high DPC latency spikes even with perfer maximum performance enabled in the control panel. Apparently this is due to windows 10 now requiring/preferring DCH version of drivers.

NVIDIA has those drivers but you will need to use their advanced driver search, NOT GFE or their usual web portal. You have to disconnect the net/disable NIC and reboot in safe mode with DDU to uninstall the standard drivers and then reboot again into safe mode and install the DCH version of the drivers and then REBOOT AGAIN into regular OS and get the NVIDIA control panel from the windows store.....ugh such a mouthful to explain driver updates in 20H2 now.....

I am about to do exactly that but apparently that has solved a lot of stuttering for people with using 20H2 and nvidia drivers. I also suggest to watch FR33THY on youtube, he has an awesome introduction and optimisation guide to 20H2 and has multiple google drive folders for all the presets and batch files he uses. His A-Z guides are something to behold but I swear the guy is like a chihuahua on coke, can't sit still for 1 second he is always tweaking hahaha. Really it's what I will be like once I get this new hardware up and running and I am on the modafinil playing FPS' til the cows come home.

OHHHHHH and another thing to try with on/off in the BIOS is maybe the spread spectrum option under the core multiplier. The changes above and this one all affect the windows timers and the new 20H2 timer is still relatively unknown in its actual function unless you're an admin or super-user.....which you would charge to spend the time divulging all the info.

Anyhoo hope that helps and look forward to joining you sweaty 5950x boiz in a few days so we can all cry thermal paste tears together as we do the re-seat dance.
 

·
Registered
Joined
·
43 Posts
Updated to F31h
Clear CMOS via back IO button for good measure.
Entered UEFI and noticed FCLK is 104.4 Mhz. Weird. never seen it so high on auto. Set it to manual 100.
Rest on default.
Problem remains. Windows crashes, Ubuntu live does not even boot.

Set VCORE SOC to 1.000 and I could boot Windows, but crashes under Aida64 memory copy and most games.
Tried almost all combinations of VDDG, VDDP, VCORE SOC, IF, RAM Speed, LLC, etc.
The higher the VCORE SOC, the most likely Windows will not even boot.
No matter the combo, Aida64 will crash on copy benchmark as well as most games. 100% will crash at idle after some time or just internet browsing with WHEA uncorrectable error BSOD.

The only setting that gets this rig stable is setting "Core Performance Boost" to disabled.
With it disabled, I can set everything else on Auto and will work just fine.

On the upside, somehow it seems F31h is very stable when changing various voltages, like VDDP/G.
Up until now, a lot of times when changing anything voltage related, the board will beep twice and reboot 3 times. Somehow F31h fixes this for me.
Try my tip for spread spectrum, that affects the base clock mate, should hopefully see some stability.

If it has the options still also try the CPPC and CPPC preferred cores to enabled and disabled C states with cool n quiet ENABLED with pstate 0. Also i believe in the advance core settings there is also a low idle voltage option, try that on and off as well, reboot each time to test (its annoying but youll thoroughly rule things out).

Those settings are also how i got 20H2 stable with my 3900x and still retaining 4.65ghz light threaded and 4.4ghz all core in gaming. A good way to check if core boosts are being assigned properly and wont get "stuck" is to check event viewer when you load in and open the system section, look for kernel entries that have preferred core in the details and make sure they all have slightly different numbers like 130, 133, 142 etc... thats the chipset confirming with the OS so that boosts and cores are assigned properly.
 

·
Registered
Joined
·
104 Posts
Discussion Starter #26
@Tweedilderp
Thank you for the information. Will try some of your suggestions.
Please note that setting manual FCLK to 100 disables Spread Spectrum, which I did and also that a Ubuntu live USB also crashes before desktop. So slim chances that nvidia has anything to do with this.
 

·
Registered
Joined
·
54 Posts
This makes me more and more convinced that we have broken processors...: GIGABYTE Latest Beta BIOS - TweakTown Forums

It is impossible that it works for one and not for another. The same CPU, the same motherboard, and the new BIOS works with him, but not with us?

Maybe check this CPU on another motherboard?
 

·
Registered
Joined
·
1 Posts
I upgraded from a 3300X to a 5950X on the Gigabyte B550I AORUS PRO AX motherboard and get WHEA_UNCORRECTABLE_ERROR BSOD in games. What worked for me was unplugging every single USB device (including keyboard and mouse). If you have things connected to the internal motherboard USB headers you might need to unplug those too.

Usually I would BSOD within 2-10 minutes in Shadow of the Tomb Raider. After unplugging every USB device I went 2 hours without a crash and then plugged them back in and was fine. Then I quit my game, restarted Windows (keeping all USB devices plugged in) and crashed within 2 minutes of the same game. After the BSOD I noticed in Device Manager I had a "Unknown USB Device (Set Address Failed)" showing. I checked Event Viewer and saw every time I had the BSOD, Windows Error Reporting was taking a kernel dump and a USBHUB3 dump. I checked what the Unknown USB Device was (it goes back to normal after a restart) and it turns out to be the Gigabyte RGB Fusion 2.0 controller.

I have G.Skill TridentZ Neo RAM too. Is this some incompatibility with Ryzen 5000 CPUs, Gigabyte motherboards, this RAM and some USB devices?
 

·
Registered
Joined
·
104 Posts
Discussion Starter #29
Tested with only a Bluetooth mouse. No other USB connected.
BIOS reset to factory settings and optimized defaults loaded.
Same crash with Aida64 copy benchmark and games.
It seems to be a bit more stable nevertheless, or just placebo in my case.
 

·
Registered
Joined
·
104 Posts
Discussion Starter #30
To rule out RAM, I have tried with one stick of G-Skill F4-2400C15S-8GNS and one KIT of 2 sticks Corsair CMK8GX4M1A2400C16.
Same problem, BSOD with or without XMP loaded.
So up until now I have replaced everything but the mainboard and CPU.
 

·
Registered
Joined
·
54 Posts
Even if the motherboard was replaced. Maybe someone of your friends has, or a neighbor in the area, a colleague from work?

For me, the computer service says that it will check in 2 weeks because they have a lot of work now. Absurdity.
 

·
Registered
Joined
·
104 Posts
Discussion Starter #33
Have a couple of X570 and B550 boards I can test the CPU on, in remote offices. Will try to test there.
In my 20 years working in IT I never once bought or had to service a defective brand new CPU. I had lots of defective boards though.
All over the internet people are starting to post about issues with 5950X, 5900X and WHEA crashes and reboots. I guess if more units were available at launch, even more people would have posted.
I am 99% sure this is a firmware issue, with AMD AGESA or Gigabyte (or both)
I just hope the fix comes sooner rather than later.
Pretty disappointing to pay for a Ferrari only to find you have to keep it under 90 Km/h.
Nevertheless, will try to test the CPU on an Asus and MSI X570 and B550 and get back with details.
 

·
Registered
Joined
·
12 Posts
Got a reply from gigabyte (what ever that is worth) as per below:

As we are looking into this issue, have you done a clean install to your system before after upgrading hardware?
To our understanding, this issue is not identical through all systems. Majority of people with this setup has the system running stably.
What we are concerns is that your old operating system is not quite compatible with the 5950X processor.
We can only confirm if you test Window 10 latest version to a spare drive and booting into the system with the 5950X.
As it is fair to assume the motherboard, and your components are all functional as it is working in previous setup, it is most likely a issue with the software (OS) or motherboard (BIOS).
If you can have the system run stably without rebooting/crash when in BIOS menu, it is less likely an BIOS issue.
BSOD only occurs when Windows system is loaded and certain detail has failed to work.

Regards,
GIGABYTE
 

·
Registered
Joined
·
54 Posts
I listened to the geniuses from GIGABYTE - I put a new system - did't help. It still reboots itself.

On Friday, the CPU is flying to the RMA. I'm fed up with this unequal fight.
 

·
OG AMD
Joined
·
8,884 Posts
the amd experience still strong.

wonder will that experience double with 6800xt..
:rolleyes: Of course Intel doesn't have that problem, they don't change tech and force you to buy a new motherboard for each new CPU.
 

·
OG AMD
Joined
·
8,884 Posts
Have a couple of X570 and B550 boards I can test the CPU on, in remote offices. Will try to test there.
In my 20 years working in IT I never once bought or had to service a defective brand new CPU. I had lots of defective boards though.
All over the internet people are starting to post about issues with 5950X, 5900X and WHEA crashes and reboots. I guess if more units were available at launch, even more people would have posted.
I am 99% sure this is a firmware issue, with AMD AGESA or Gigabyte (or both)
I just hope the fix comes sooner rather than later.
Pretty disappointing to pay for a Ferrari only to find you have to keep it under 90 Km/h.
Nevertheless, will try to test the CPU on an Asus and MSI X570 and B550 and get back with details.
Early adopter's tax. No hardware is immune. I always wait for the kinks to get ironed out.
Don't forget this could also be caused by Windows. Try Ubuntu.
 

·
Registered
Joined
·
104 Posts
Discussion Starter #38
F31i same problem.
 

·
Registered
Joined
·
54 Posts
Gigabyte on my website (POLISH) removed even previous versions. Now the newest one is the F30.

My package is on its way to the service - I bought it at Proshop.

Deepcuts, you did too many things and checked it was the BIOS fault.
The faster you send your CPU to the RMA, the sooner you will get a new one.
From what I saw on the Proshop website, new processors (5950X) will only be available on December 28, 2020. But the service is not included in the store's range. After all, the service also has to get you a new processor from somewhere.
 

·
Registered
Joined
·
12 Posts
Returned my 5950x to the retailer, they tested it on the asus dark hero. Still the same issue, Bsod with Whea errors.
Kinda sucks isn't it. Don't tell me we have all with us faulty chips ?
All the 5950x sold out in my area. So can't get a new replacement on the dot.
It's either wait for new stock to arrive (that will take ages) or get a refund
 
21 - 40 of 73 Posts
Top