Windows 8 RTC Bug analyzed and fixed!

Published by Christian Ney on 22.08.13
« 1 (2) 3 4 5 ... 7 »

Windows 8 testing on Intel

First we had to try again the tool to see if it behaves the same as before and if the new information it provides are right.

Without changing the bus frequency in windows everything is normal, all timers are displaying the same time.

Now after changing the bus frequency, in this case downclocking from 100 MHz to 95 MHz:

Here is something new, before we knew the RTC was affected but now we can see that the QPC Timer too is affected and displays the same time as the RTC Timer after 180s.

For the next test we booted at 95 MHz directly from the BIOS to show you that the issue only happens when changing the bus frequency in windows 8:

Here all timers display the same time like at 100 MHz which is normal but there is something that changed. The QPC Timer's frequency is of 3.25 MHz now and previously it was 3.42 MHz. This is a very interesting information since it shows that changing the bus frequency has an effect on the frequency of this timer. Furthermore we see the following: 100 MHz / 3.42 MHz = 95 MHz / 3.25 MHz = ~ 29.24. So it looks like the QPC is using the DMI frequency as timer source or a timer that makes use of the DMI frequency.

Before switching to the AMD platform here is a resume of what we learned from the Haswell testing:
  • RTC and QPC Timers use the same hardware source (timer or frequency).
  • This source is sensitive to bus frequency change. In other words Windows 8 seems to use an internal hardware timer or directly the DMI frequency instead of an external hardware timer such as HPET or ACPI.

Page 1 - Introduction
Page 2 - Windows 8 testing on Intel
Page 3 - Windows 8 testing on AMD
Page 4 - Windows 8 installed on AMD testing on Intel
Page 5 - Is AMD really problem-free
Page 6 - What about Windows 7
Page 7 - Conclusion

comments powered by Disqus

Windows 8 RTC Bug analyzed and fixed! - Articles - Reviews - ocaholic