Comment 312 for bug 1690085
Revision history for this message
|
|
#312 |
© 2004
Canonical Ltd.
•
Terms of use
•
Data privacy
•
Contact Launchpad Support
•
Blog
•
Careers
•
System status
•
e3ace39
(Get the code!)
(In reply to Panagiotis Malakoudis from comment #223)
> Blaming the PSU from AMD's part is wrong, since idle freezes have been
> documented with various PSUs, new and old. If you read my older posts, I did
> a complete replacement of parts between two systems: one excibiting the
> freeze on idle issue and one that didn't. The system not excibiting the
> freeze was using a really old (pre Hasswell) PSU. Putting this PSU on the
> other system didn't avoid idle freeze.
Two have one more documented check by Chris Hall can't be bad. But I fear I know the end of this test - even if I wish him it would work.
> The problem is related to power and actually I have "hide" the problem by
> overclocking my CPU.
This is a new interesting information.
> Hitting certain voltages/
> the issue (at least on my case with an ASUS X370 Prime). But even these
> voltages/
Me!
> here reported he "hides" the problem by using the automatic overclocking of
> motherboard ("optimized mode for daily computing" option) which actually
> overclocks to 3600 if I remember well.
You do! 3600 probably is enough for me, as I'm always running 4 VMs - which are mostly idle, too, but they add some additional minimal load, which could prevent the system from often entering very low current levels. Moderate overclocking additionally helps to circumvent the problem.
Next point is, that I'm seldom running this machine more than 15 hours - therefore, I can't say, if the problem disappeared completely. But before the overclocking, I could see freezes even after an hour or earlier.
> For me this didn't work, still had
> idle freezes. But idle freezes disappeared when I overclocked at 3900.
This is good to know! There are a lot of posts elsewhere about overclocking Ryzen (and RAM - that seems to be much more difficult) - mostly by Windows users. Overclocking doesn't cause difficulties but even could resolve problems! That's a cool new idea!
Could it be related to the relation between RAM and CPU clock?
[...]
> It is a hardware issue that AMD needs to fix and it
> seems it has been fixed in with this new option "Power Supply Option".
As I could read elsewhere [1], this option would disable C6 states among others. so, using zenstates.py to switch off c6 states would be the same "solution".
[1] https:/