X370 Taichi crashing on stock settings and C6
Printed From: ASRock.com
Category: Technical Support
Forum Name: AMD Motherboards
Forum Description: Question about ASRock AMD motherboards
URL: https://forum.asrock.com/forum_posts.asp?TID=8492
Printed Date: 27 Dec 2024 at 8:17am
Software Version: Web Wiz Forums 12.04 - http://www.webwizforums.com
Topic: X370 Taichi crashing on stock settings and C6
Posted By: ricesteam
Subject: X370 Taichi crashing on stock settings and C6
Date Posted: 24 Apr 2018 at 10:08pm
|
" rel="nofollow - Just picked up a 2700x and had the same issue with 1700. I'm using the latest 4.60 bios. For ram, I'm using G.SKILL F4-3200C16D-16GTZ. The problem I faced was with stock and auto bios settings, for both CPUs, was windows would crash within 2-5 minutes after loading. I would get a black screen, with fans going 100%, that I cannot normally reset or power off. I had to turn off my power supply. The fix I used was to manually set a fixed cpu voltage. Since I manually OC'd my 1700 anyways, stability returned and I just left it at that. Note: using offset instead of fixed would reproduce the problem. Now that I've switched to the 2700x, I want to use stock settings because XBR2 is better than manually overclocking, but the problem has returned. I don't want to manually set the voltages because it limits the boosts and it'll always run hot. It took me hours of trial and error, but when I disabled the global C-State, everything is working with stock/auto settings! Can someone explain to me what Global C-State is? Why do I have to disable it for stock settings? Also, what is C6 state? Where can I find that setting in the bios? Anyone else run into the same problem I had? Edit: I'm using a Sapphire 580 8G SE video card |
Replies:
Posted By: Onox
Date Posted: 25 Apr 2018 at 12:21am
|
Hi, I'm interested in this too. I have the same board as you and a 1700X, and I have experienced a few crashes since I built this computer last December. I have also read other posts regarding C6, and I haven't had any crash since I disabled it. However... I am not entirely sure that my issue is related to this, because unlike you and other posters, the crashes have been fairly rare for me. In total, it happened twice on Windows 10 and once on Fedora 27. The three of them occurred at idle or low load. The computer has been stable otherwise, except that I have had a number of LiveKernelEvent code 141 errors in the Windows logs. I believe this is a different issue, but again I don't know. It's very strange, because there was a time when such errors happened once or twice a day for several days, and now it hasn't happened for ten days, but I haven't changed anything to my system! When this code 141 error happened, my screen usually flashed white, but otherwise Windows continued to work (and I have never noticed such flashes on Fedora). My graphics card is an Nvidia GeForce GTX 1070. I would have preferred an AMD card, but couldn't get one due to the shortage and absurd prices because of crypto-currency mining. I left every settings in the BIOS on Auto and haven't overclocked. I would like to set my RAM to a higher frequency, though, but I haven't looked into it. I have these sticks : G.SKILL F4-3200C14D-32GTZKW, which are running at their default 2133 MHz and 15 timings currently Please could you tell me which settings you used with your Ryzen 1700 to fix the crashes without having to disable C6? |
Posted By: ricesteam
Date Posted: 25 Apr 2018 at 2:01am
|
I'm wondering if C6 and Global C-State are the same things. What ever documentation I can find is very bad. I'm worried it could be a hardware fault since another person with similar spec don't have any problems with C-state enabled. My next suspect would be my RAM. Off the top of my head, I manually set the clock speed of my 1700 to 3800, fixed voltage of 1.3v, fixed Soc of 1.1v. I had both LLC at Level 3. Keep in mind every chipset is different so the exact config may not work for you. The general idea is that I'm manually setting values and I assume when I manually OC all those nice little features, like C-States, are disabled. |
Posted By: datonyb
Date Posted: 25 Apr 2018 at 2:03am
|
@ onox you do have global c state in your bios (its just in a different place to the other c states and cool n quiet).....for reference i have always left this on since bios 1.6 and overclocked ,so dont worry as for your ram, theres a good chance those sticks will work on taichi and bios 3.2 with simply clicking xmp 3200 (to be fair such a nice pair of ram sticks are being very wasted at 2133) ------------- [url=https://valid.x86.fr/jpg250]  [/url]3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold |
Posted By: datonyb
Date Posted: 25 Apr 2018 at 2:05am
just a quickie as my dinner is served but 1700 at 3800 with 1.3v is on the low side for that , that may be an issue ------------- [url=https://valid.x86.fr/jpg250] [/url]3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold |
ricesteam wrote:Posted By: ricesteam
Date Posted: 25 Apr 2018 at 2:15am
|
Depends on your "Silicon Lottery". Besides, that's not the problem I'm facing. Just an aside of information. The main problem I faced was with stock (auto) settings that would crash my system unless I turn off C-States.
|
Posted By: Onox
Date Posted: 25 Apr 2018 at 3:07am
|
" rel="nofollow - ah, I forgot to mention that I tested my RAM with Memtest for about 1.5 hour, and no errors were found. It might not be enough to conclude anything, but that's reassuring nonetheless. Also, when I experienced my crashes, they weren't the same as you ricestream : instead of getting a black screen, the first time, Windows froze completely (the HDD LED was still blinking from time to time, but I had to press the reset button), the second time (Windows), it made a blue screen and rebooted, and the third time (Fedora), it was a random reboot as well, with an "hardware error" message in the logs. I thought about something : could it be possible that due to the "silicon lottery" you mentioned, the default/auto settings of the X370 Taichi are not enough for some CPUs that need a bit more voltage to function properly on idle (not sure whether this would be considered a hardware defect...)? @datonyb : I have been afraid to change the RAM settings because I'm not sure if I need to change other voltages at the same time, such as that of the SoC. And I use this machine for my work too, so I'd rather have it slower than unstable :) Because I've heard that 2933 MHz doesn't bring much over 2666 MHz, wouldn't it be safer to try that? The problem if I want to try such a frequency is that then, I have to set the voltage myself, don't I (because IIRC, there isn't an XMP profile that I can choose for lower speeds)? I would be interested in swapping parts to see if it changed something, but for me it is a difficult thing to do, as building PCs makes me very nervous, and I have a physical disability which makes it even more difficult (couldn't build the computer without help, and the people who helped me don't have much time anyhow). I could probably test another GPU on the Taichi, but not the GeForce 1070 in another PC, because I don't have another machine where it would fit. I also don't have another PSU available. My current PSU is a Corsair RM650x. As to the storage, I have a simple Sata mechanical HDD, which limits the problems on that side I guess (since I've seen people whose crashes were actually caused by SSDs) |
Posted By: datonyb
Date Posted: 25 Apr 2018 at 3:14am
|
@ onox setting xmp will set 1.35v on your ram and with bios 3.2 you shouldnt need to mess with soc voltage its not that scarey it will either work or it will give you the beep beep beep of shame and revert back to stock safe bios if it works and boots then just test it for windows stability and see if its all good ssd problems ? i dont know i have none and havnt used a spinning drive for a long time samsung once again are the name to look for either a nice nvme or the normal 2.5 inch drives now they really make a difference to a spinning disc and for the love of god CAN NO-ONE ACTUALLY WRITE MY USERNAME CORRECTLY  ------------- [url=https://valid.x86.fr/jpg250] [/url]3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold |
Posted By: Onox
Date Posted: 25 Apr 2018 at 3:20am
Sorry about that, I generally pay attention to this kind of thing. Edited my message  |
Posted By: ricesteam
Date Posted: 27 Apr 2018 at 8:11am
|
I like to add that Wake up on Sleep no longer works. When I wake up the computer, it will turn on, then nothing happens for a minute, and it shuts down. I wonder if this is because I have C-States disabled.
|
Posted By: ricesteam
Date Posted: 28 Apr 2018 at 4:50am
| " rel="nofollow - Edit: false alarm...sigh. Thought new Samsung ram solved the issue, but nope still instability. So motherboard hardware issue or bios issue? |
Posted By: ricesteam
Date Posted: 28 Apr 2018 at 6:41am
|
FYI, the different C-States and what they mean: https://gist.github.com/wmealing/2dd2b543c4d3cff6cab7" rel="nofollow - https://gist.github.com/wmealing/2dd2b543c4d3cff6cab7 tl;dr - basically, power saving features on the CPU.
|
Posted By: ricesteam
Date Posted: 01 May 2018 at 9:00am
|
" rel="nofollow - Updated to the 4.64 beta bios. Same problem as before if I have C-states on Auto (enabled). However I started to play with the Power Supply Control and switched it from Auto > Typical Current Idle. So far, my PC is stable. I guess my problem has been solved? Only time will tell... Edit: To be clear, C-states are on 'Auto' and Power Supply Control is on 'Typical Current Idle'. This resolved my issue.
|
Posted By: Onox
Date Posted: 03 May 2018 at 2:05am
Hi! So, you still have that wake up from sleep issue with this BIOS? I have never had the problem you describe, but I'm still on BIOS 3.2 (although Fedora sometimes throws a non-fatal error when waking up, which doesn't seem to have any serious consequence, as far as I can tell. It might not even be related) What I wonder is how much power savings I lose by keeping C6 disabled? My idle temps don't seem higher than before |
Posted By: ricesteam
Date Posted: 03 May 2018 at 2:17am
|
Sorry, my previous post wasn't clear. The issue is resolved by setting Power Supply Control to 'Typical Idle Current'. This setting allowed me to enable global C-states without any problems. Thus, when my computer sleeps I can now wake it up.
|
Posted By: shmerl
Date Posted: 18 May 2018 at 11:53am
|
" rel="nofollow - Sounds very similar to this bug: https://bugzilla.kernel.org/show_bug.cgi?id=196683 I thought it's fixed in Ryzen 2700X. My Ryzen 1700X was plagued by freezes on idle (on Linux) until I enabled "common current idle". I recently bought 2700X and it didn't have that bug anymore. Common current idle disables package C6 state (but keeps core C6 enabled).
|
Posted By: Onox
Date Posted: 14 Jun 2018 at 9:30pm
Interesting, thank you! I just had a freeze + reboot today, still with BIOS 3.2 and C6
disabled. I hadn't had a crash for weeks. I don't really have the time to investigate now, but it could be that same bug
Found this in the logs : mce: [Hardware Error]: PROCESSOR 2:800f11 TIME 1528987490 SOCKET 0 APIC e microcode 8001129 mce: [Hardware Error]: TSC 0 IPID 300b000000000 mce: [Hardware Error]: CPU 14: Machine Check: 0 Bank 3: b2a00020003f0000 Strangely, the
indicated time for those three entries is also wrong (16:44 but it was 14:45,
as the next entries also tell). Those lines were displayed when the
computer rebooted too I'll try to upgrade the BIOS soon |
Posted By: Onox
Date Posted: 15 Jun 2018 at 9:23pm
|
" rel="nofollow - Froze twice today on Fedora... if this is the same issue as before (I'm not sure, since I didn't use to have the same kind of error message (as shown in my previous post), and since I haven't had Windows crash again yet), that's a huge increase in the frequency of crashes =( The latest kernel updates might have changed something for the worse Edit : I have found out about the Ryzen "segfault bug", I tried the kill-ryzen test, and it seems that my processor is affected... RMA time is approaching I guess
|