OCed Threadripper 1950x Time Drift after sleep
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=6399
Printed Date: 27 Dec 2024 at 11:24am
Software Version: Web Wiz Forums 12.04 - http://www.webwizforums.com
Topic: OCed Threadripper 1950x Time Drift after sleep
Posted By: llitz
Subject: OCed Threadripper 1950x Time Drift after sleep
Date Posted: 24 Oct 2017 at 2:14am
|
Wanted to check if anyone else experience this and, if not, what CPU and Motherboard they are using. Would also be nice if someone on AMD could confirm this. I running the 1950x on an ASRock Fatal1ty and when I resume my computer after sleeping, time starts to pass faster. I use the website https://time.is" rel="nofollow - https://time.is and reload it sometimes. Running a 10% OC makes time pass 10% faster. If I use Ryzen master to put my CPU back to 3400mhz before sleeping, I don't experience time drift. While overclocked, the only way for the computer to work normally was by enabling HPET in bios and on Windows, which made everything slow, but would allow me to sleep and resume while OCed.
|
Replies:
Posted By: MisterJ
Date Posted: 24 Oct 2017 at 4:34am
" rel="nofollow - llitz, this is a user forum. I suggest you open a ticket with AMD and ASRock (or you MB vendor) to see what is really going on with your system. Please see my system specifications in my signature. I Sleep my system regularly but do not OC and have seen no problem. I assume you are talking about the TOD clock. When you set HPET in the BIOS (what version?) did you also set it using BCDedit? You can check it by opening an Administrator command prompt and typing bcdedit. Look for:
------------- Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD |
Posted By: llitz
Date Posted: 24 Oct 2017 at 4:50am
|
Hello John, Thanks for taking the time to talk to me (=. Right now, I am trying to identify if someone else is seeing the same behavior as I am, hence posting on the user forum. I have tickets opened with ASRock and AMD. ASRock still has to reply to me, but AMD want me to RMA the cpu. Considering I still have some days to return it to Microcenter, I am more inclined to just drive down instead of not having my workstation for a week (just did 2 RMAs on a Ryzen 1700x). About the useplatformclock, I am away of it. Played with that a lot. With that enabled, the clock doesn't drift anymore because Windows will use a different mechanism to sync the ticks. Unfortunately, when that is enabled performance tanked for me (not only in gpu related activities, but anything that had lots of instructions, even reading from the ssds was slower). If that parameter is on and HPET is off performance is bad, with HPET on performance is terrible S= but no time drift. For now, I am just reducing the OC before sleeping and returning it when I need more power. If someone else see the same behavior as I do, I may just not return the CPU (2 days left on return).
|
Posted By: MisterJ
Date Posted: 24 Oct 2017 at 5:23am
|
" rel="nofollow - May be too late, llits, but did you ask AMD about an Advanced RMA? Some vendors will send you the replacement before they receive yours. Probably will require a credit card. Please keep us informed. Enjoy, John. ------------- Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD |
Posted By: llitz
Date Posted: 24 Oct 2017 at 5:28am
| Did that with my first Ryzen Chip, then they wouldn't do with the second but said "as soon as you ship and send the tracking label we will send you another"... Well... the guy who checks support emails took 4 days to see it, by then my CPU had been at their warehouse for 2 days already... |
Posted By: parsec
Date Posted: 24 Oct 2017 at 10:12am
|
" rel="nofollow - Seems like a strange issue, I would suspect either Windows or a problem with the UEFI/BIOS. Wake from Sleep issues with a CPU can be caused by registers in the CPU not being restored to their correct values when the system is waked. A common simple example of that are the core multipliers being restored incorrectly. I'm not referring to Ryzen or TR specifically, but any system can have that problem. That is blamed on the UEFI/BIOS as I recall, at least in some cases, but exactly why I don't know. I'm curious which version of Windows you are using. Which "clock" in the CPU would be the master clock for the whole system? I'm surprised AMD issued an RMA for this, unless they are aware of this happening with TR. Given the similarities between Ryzen and TR, Ryzen processors could have this bug, IF it really is the processor at fault. I just have an X370/1700X system, but will test disabling HPET in the UEFI (enabled by default), if only to test the performance issues you mentioned. I use Sleep with all my PCs, and have not noticed a problem with Windows time. You did not describe specifically how performance is affected, so I'm curious about that too. ------------- http://valid.x86.fr/48rujh" rel="nofollow"> 
|
Posted By: llitz
Date Posted: 24 Oct 2017 at 11:59am
|
Ryzen had a problem like this around March, but it was fixed. The HPET isn't enabled in Windows unless you have enabled it with "bcdedit /set useplatformclock yes" or you run Ryzen Master "beta" version that would automatically enable it for you. It doesn't matter if it is enabled in the bios or not (= Running Windows 10 right now. I am not surprised AMD would issue the RMA. I have had different problems with different CPUs from them (2x RMA on Ryzen 1700x with segfaults) and I am very specific in writing on how to reproduce it. I do believe no one tried to reproduce it though. I can still return/exchange at Microcenter for more 2 days, I might go down there and see if they can test if there's no one (on the different forums I am asking) with the same problem or no one available to test. |