Windows 10 Insider Preview Dev Build 21343 (RS_PRERELEASE) - March 24 Insider

f14tomcat said:

Hmmmmmmmmmm

Wonder if they'll be back in 5 minutes yelling April fool!

Nah..........

I just hope they get all their ducks in a row before their next release so that fewer people have fewer problems with installations and startup issues.

magilla said:

I just hope they get all their ducks in a row before their next release so that fewer people have fewer problems with installations and startup issues.

Like my problem with Notification Area that I had to roll back to previous.

FWIW I can confirm that this build boots quite a bit slower. The question I have for myself is whether it worth my while to investigate where the slowdown is occurring. Perhaps those that actually using the Feedback Hub app can tell me whether the discussion there has proceeded to the point where other people are using tools to explore this. For example, I know the Russian programmer who founded Greatis (author of Regrun and another program that analyses boot-up times) but it would mean spending a lot of time looking into something that Microsoft programmers know quite about already., but choose not to share with "insiders".

The April fool was on us. We will see if they acknowledge and fix the boot up time problem. My boot up time went from from ~30 to ~55 secs - almost a 100% slow down.

Hi there
As I suspect there's stuff in the kernel that adds debugging and overhead - a lot of it could be at boot and "hardware detection" - but surely even if it needs to do sophisticated hardware detection on first boot any sensible OS would have some sort of specific hardware database created so it just can skip that on further boots on the same machine unless new hardware added.

On VM's of course in general the hardware is usually pretty standardized paravirtulised hardware whatever VM software you use and built in to the virtual Bios so the amount of hardware detection required is pretty small. So we get the unusual situation with some of these builds is that they actually seem to perform better running as a VM than on a Native machine.

Servers of course won't have all that extra stuff in them -- servers are generally designed from the outset to be lean and mean - hence the mega performance disparity between these builds and a commercial / retail version of Server W2K19 or W2K16 !!

Before RTM Ms will have to get rid of a load of the cludge that appears to be gumming up these new W10 builds.

Here's a W10 running as a VM on a dual XEON system just running Outlook 2019 -- not the biggest process hungry job on the planet to READ a few emails never mind sending them

just look at the CPU consumption -- nothing is running on the HOST !!!!! apart from the Hypervisor which is negligible (like HYPER-V) . No other VM's running either.



The CPU usage should barely be registrable on a dual processor XEON system -- OK these processors aren't the most modern but they are still blindingly fast !!!.

Reading same email on W2K19 server - Full Desktop enabled so not just the "base edition: CPU significantly lower -- same resources allocated as to W10 VM. Here again just 1 VM running.



BTW Happy Easter everyone.

Cheers
jimbo

OHHH, I'm sorry. I was so involved reading JImbo that I forgot I was on the Insider build thread.

Please ignore my comment. it's more 20H2 related.

Hi, I'm using the new Z590 motherboard. I did not witness boot slowdowns but I noticed that the new 4080 USB sound chip is a cause of boot problems on the new motherboard. Driver loads take almost a full 10 seconds more on Windows loads.

I opened a ticket with Asus and even sent a boot trace.

But I do agree. Is MS sloppy when it comes to general use of Windows like Pro and Ent vs Server ?
I have no idea but it's interesting to see how much difference between the two there is.

Hey guys, it’s been a while, Happy Easter, a little late but downloading :)

Fedora, and most modern Linux OS's, give testers the ability to download code without elaborate hardware detection (essentially non-debug versions of the kernel). VMware gives testers the option of by-passing their debugging code for performance reasons - although they are not happy about it. Microsoft should do the same if, as jimbo45 suggests, this is the reason for the slowdown. I mean, they pretty much have complete control over my physical (and VM) instances - they could allow me to experience and test with the RTM version will be like. Actually, speaking of that does anybody know (Brink maybe?) whether we are working on 21R1 or have they already ditched that nomenclature. Wish there would be more communication from the Insider team but I'm not expecting it. Guess we'll get new emojis next week or redesigns of the Trash bin icon.