[INFO - DISCUSSION + Downloads] Windows 10 1809 FINAL b 17763.xxx (PC) [RS5]

As this went to GA many people will be running it fine already, thats a much bigger reason than with 1803 for MS to fix this some other way, otherwise they all get another full upgrade in the space of weeks which isnt much fun for business users especially those sporting celeron W7 era machines.

 

You need to login to view this posts content.

 

i remember doing some research on this a long time.. and basically what i found after reading lots of post is that hpet is deprecated it use a clock embeded on the motherboard hardware or something like that. and at some point windows 7 or 8 dont recall, windows start using its own internal OS timmer or something like that.

also just to clarify in ur first pic ur not using hpet i dont see the string

To see if it's on or off run "bcdedit /enum" and if you don't see a line that says "useplatformclock Yes" it's off.
To disable HPET in Windows run the command... bcdedit /deletevalue useplatformclock
To enable HPET as the only timer run the command... bcdedit /set useplatformclock true

Note: it is recomended to leave it off on newer systems/OS,,, even if u enable it on the BIOS windows 10 wont use it,
Unless you enabled it yourself at some point, HPET is off by default.

 

Software under W10 will still be able to use HPET when useplatformclock isn't set, as long as you don't disable its support on bios (and the driver is listed in device manager). It can still be useful for some applications when a more precise timer is required, but software must call it explicitly. In those cases, when it's not present, apps will fall back to other windows timers.

 

There might be Event Viewer events from "Kernel-Power" in System Event Log.

Example on VirtualBox, I get Kernel-Power Event ID 508:

Code:

The system has been constrained to a periodic tick

Reason: No HW support.

So, check out Event Log for Kernel-Power if you have timer-based issues.

 

To clarify a few things:

- the images show both 1803 and 1809, you can see the 3.6MHz timer on 1803 and the 10Mhz timer on 1809 (HPET Off/Default)
- forcing HPET is 24MHz for my PC and kills performance

The issue is NOT why and how HPET works, there's an article about how the Intel HPET "bug" and how performance is affected, this basically made Anandtech fail their 2700x review as they forced HPET and the 2700x came above the 8700K in games, a conclusion they had to revise.

The issue is why in 1809 the timer was changed to 10MHz. Intel Z370 on default should use Invariant TSC and that used to be 3.6MHz, now it's stuck at 10MHz (confirmed by a few others in Nvidia forums and now here) and nobody knows why.

PS:

- Kernel-Power only has shutdown transitions and sleep resumes, nothing related to this. Invariant TSC is shown in AIDA64. i mean, if it worked in 1803, it's not like it died in 1809 suddenly and replaced by something else that nobody knows what it is.

To be fair this is not a serious issue, because performance seems largely the same (or slightly improved even). It's just a curiosity. Why do it? Why set it to 10Mhz? And which timer is this exactly? Is it stil Invariant TSC? Can that run at different values?

Curious about this too. Can software call HPET if it's BIOS enabled, OS default (i.e. disabled system-wide)? How will that work? TimerBench for example requires a reboot to enable HPET. If it's possible to just use HPET on demand, why require reboots?

 

Yeah fresh install working just fine. I never ever do an update because it can cause all sorts of problems.

 

You need to login to view this posts content.

 

Same. And I usually wait a bit before I do the clean install. This time though I went through WUpdate. I was stressed once the tweets started to come out on the issues with the update. Fortunately I rebooted fine and had no problems. But it was a good reminder to stick to my routine.

 

1511 had Bitlocker bug, and was pulled, then returned (same media) with CU

 

I jumped on the fast ring and upgraded from 1803 spring update to 17755.1 RS5 and have used WU to upgrade every build released up to 17763.1 and got out of the fast ring with no issues at all. I never did a clean install or upgraded using an ISO so I don't understand how files are getting deleted on peoples computers. I wonder if it's hardware or some type of config related issue? They did mess around with storage sense in this update, could that have anything to do with it?

 

A good guess, they have increased storage sense in order to replace disk cleanup and it is probably too sensitive now.
This is one of the reasons, I hate everything automatic (thus I disable it), it all works great, until one tiny tiny bug occurs.

 

You need to login to view this posts content.

 

In Storage Sense and Disk Cleanup (cleanmgr.exe) By default the downloads folder is selected for deletion if they are stored for more than 1 day - 60 days

 

You need to login to view this posts content.

 

You need to login to view this posts content.

 

I really think Microsoft should wait when they feel that they hit RTM and keep that build in the slow or release ring for a few weeks to iron out any serious bugs before it's pushed out to the public. I mean it would give more time for insiders on that build to test on different machines and what not. I read somewhere that the deleting bug was reported in the feedback hub months ago but got lost in the shuffle.... That is something they need to fix also because they could of caught it sooner!

 

If MS will release new build to be RTM, I have to throw one media (disc).

 

@cuteee why would you still use dvds for windows images? Can't you make a bootable usb drive for install/upgrade purposes? They're faster and more practical than a dvd since you can always update the image or just format the drive and start anew. It seems to me like an unnecessary waste, especially if you consider than many new computers (be them desktop or laptops) don't even carry an optical drive anymore.

 

MSDN media has been pulled for 1809 and Server 2019 so Im going to say yes