Another look at HPET High Precision Event Timer

From Microsoft, still a nice read IMHO even if from 2002
https://web.archive.org/web/2008041...ft.com/whdc/system/sysinternals/mm-timer.mspx
Quote:
The High Precision Event Timer
The High Precision Event Timer (HPET) was developed jointly by Intel and Microsoft to meet the timing requirements of multimedia and other time-sensitive applications. Originally, the HPET was called the Multimedia Timer (MM Timer), but the name was later changed to avoid confusion with a Microsoft DirectX timer, and to better describe the timer.

Benefits of the HPET
Experiments on prototype HPET hardware by Microsoft Test Engineers have shown that, in addition to extending the capabilities and precision of a system, the HPET also improved system performance.

I still have Vista on an old laptop and that's how it worked for me. But lets quote Microsoft again.
https://docs.microsoft.com/en-us/windows/win32/sysinfo/acquiring-high-resolution-time-stamps
Quote:
QPC support in Windows versions
QPC was introduced in Windows 2000 and Windows XP and has evolved to take advantage of improvements in the hardware platform and processors. Here we describe the characteristics of QPC on different Windows versions to help you maintain software that runs on those Windows versions.
<snip>
All computers that shipped with Windows Vista and Windows Server 2008 used a platform counter (High Precision Event Timer (HPET)) or the ACPI Power Management Timer (PM timer) as the basis for QPC. Such platform timers have higher access latency than the TSC and are shared between multiple processors. This limits scalability of QPC if it is called concurrently from multiple processors.


So why do you say no?

 

The PM timer was the QPC timer unless platformclocktrue was set on Vista.

 

Won't the 10mhz timer on the newer versions of Windows 10 cause latency issues? Or is that just fake news from FR33THY and crew?

 

Have you seen him elaborating on this subject?

 

The best way to control the mitigations is through the windows registry switches (I mean, I recommend defaults but if you wish otherwise...). You can and should keep both Windows and the CPU microcode updated (microcode updates are not a new thing, they have been used to patch CPU bugs for decades, spectre/meltdown is a security bug but there are also stability/performance/datacorruption bugs in CPUs...).

 

Where is the connection to that specific Windows version when it was fixed in the Nvidia driver?

Is there any benchmark that shows any performance penalty with that Windows MC update when mitigations are turned off via registry/group policies?

 

InSpectre does that, a free small tool.

 

No, I can't say I have. That's sort of the reason why I'm asking here, you guys seem more on top of it than he does. So, do you have the answer? Like I said man, I'm an open book here and I'm willing to learn.

 

In fact I never really see him elaborating on anything, but it seems like he's getting most of this info from people like you guys which actually know what they're talking about. So he's probably at least second hand right about most of the stuff he says, even though he doesn't have an understanding of it.

 

I get what you are saying, but so far the microcode 22h for CPU's like mine, such as Haswell, seems to work the best on older systems.

https://www.overclock.net/forum/5-i...pectre-performance-stability-differences.html

https://www.overclock.net/forum/5-i...de-update-cpu-microcode-through-software.html

Also any microcode past that has security mitigations. Imo spectre&meltdown "security" is just another bloatware, I don't need that hindering my gaming experience and I have not seen a single "attack" to any home user that doesn't have this "protection". I doubt very much a newer microcode would make a CPU perform a lot better "magically" or get more performance score on benchmarks. In fact, past microcodes 22h seems to ruin Haswell CPU's, which make sense, any microcode past 22h has spectre&meltdown mitigations, so I doubt it would provide any "bug fix".

 

As I mentioned before, his videos have bits of good info, but a great majority of the time he doesn't elaborate on anything, literally he just tells people to do it because.

 

I found some users worried about this:

https://answers.microsoft.com/en-us...-windows/44946807-5355-4b36-ba3e-43aa86ce30c0

https://social.msdn.microsoft.com/F...ncyqpc-in-windows-10-1809-onward?forum=windbg

https://www.reddit.com/r/windows/comments/bm4w8l/in_windows_10_1903_queryperformancefrequency_is/

https://www.windowsphoneinfo.com/th...ows-10-build-1809-causing-performance.140399/

https://www.overclock.net/forum/132-windows/1711242-my-windows-timer-seems-off-after-new-update.html

Decide for yourself, so far it seems to be "bad", no word on why MS decided to do this, I am waiting for Windows 2004 release to see if MS changed this or gives any info on the subject.

 

useplatformclock wasn't an option in BCD until W7, just as useplatformtick wasn't an option until W8. QPC is a counter, not a timer and both the PM Timer and HPET are platform clocks of which under Windows HPET has precedence.

From your link

The 10MHz QPC is also from TSC.

 

And how do you know its also TSC? Also I never claimed anything, I just posted that because he seemed worried about the 10 MHz QPC and apparently, it creates a bit more of system lag. When I tried both 1809 and 1909, the system felt like if I was playing a game with v-sync on, on 1803 and earlier of course, the system feels snappy and agile again.

 

...came across this site

http://www.neowin.net/forum/topic/1...n-bios-and-os-for-better-performance-and-fps/

i used to have hpet disabled in bios and it was also disabled in my windows.

now i enabled it in the bios and i enabled it in windows.

i have yet to test any games and i dont really notice any difference

my max latency in dpc latenacy check is 132 us

in the topic there is wildy different results

but with my computer in non gaming related usage i dont see any differnce at all.


That's the first post on this thread from 2012.

 

Uh oh, this thread is about to loop again.

 

Well, for starters, it should be move from Geforce Drivers to some other section of the forum (like Operating Systems or Game Tweaks may be).

 

That thread is ancient and the user who posted that was explaining things from his point of view, it doesn't mean it will be the same for all users or that he is right. If HPET active on Windows was so amazing and epic and whatever, then the O.S would have it active as default; yet it doesn't. Windows uses TSC; HPET when its needed, no need to force it on or off, I believe if an user fully understand its functions, then they could force it on for a particular program or application, the rest of users don't need it on because Windows manages it.

Stock HPET+active on BIOS is the best setting:

I'd rather believe a real video than some telling me to force a timer on my O.S that doesn't need to be force, because its actually dynamic.

He also said this (from the thread you linked): "Fortunately I have to say that useplatformclock defaults windows to the best that the clock has (cpu), but is it so with mb's?"

He doesn't even know what he is doing or changing, it just seemed ok for him, that's it

As for latency, well, this my system: https://imgur.com/a/XqOkjuS

And if your whole post is sarcasm, well then, I fully fell for it, because taking advice from a random post from a random user that has no idea on anything its something I refuse anyone from here would just do.

 

We have no means to check that. I mean we can't switch that frequency on the fly. And if you use older version of Win10 then the test is not relevant because whole OS is different.

 

I tested Windows 10 1909 that has the 10 MHz QPC, latency measurement with Latencymon was as low, if not lower than 1803 with some tweaks I do. Very good in the regard, but the system felt "muddy", like most of the stuff I did had an small delay to respond, fast, sure, but not as snappy as 1709 and 1803 versions are. Even with Spectre&Meltdown disabled, the 10 MHz QPC will remain, there is no way to change it. Speaking in general terms, it won't affect the normal user because not everyone can notice these changes as some of us do, gaming felt the same, no difference that I can remember.

I decided to go back to 1803 and everything feels ok, its a bit snappier. When Windows 2004 comes out, I will try it, it promises a lot, but we'll have to see.

Just stay on any version you have, don't rollback, try to tweak it as much as possible, fr33thy guides are ok, even so he doesn't explain certain things, some of that stuff just works. Keep drivers up to date, use DDU in case you think your GPU driver its giving you issues and install the newest one. ISLC is not needed anymore imo, but if you use Windows Defender, disable all ASLR security layers if you feel your games have some problems. You won't get hacked or anything by disabling this, don't worry.

Remember MSI modes (guide from mbk1969) here: https://forums.guru3d.com/threads/w...ge-signaled-based-interrupts-msi-tool.378044/

Set affinities:

Try using Park Control, great tool: https://bitsum.com/product-update/parkcontrol-v1-0-3-0-released/

Disable Windows spying "features": https://www.oo-software.com/en/shutup10

Optimize the visual effects to minimize system lag due to the graphics interface.

Google Windows 10 unneeded services and disabling them slowly, check what they are for and if you don't need it, disable, then try your system and so forth. Generally, this is to lower latency a bit and RAM usage at idle.

DO SYSTEM RESTORE POINTS BEFORE DISABLING WINDOWS SERVICES because if your Windows stops booting, you will be able to recover. If not, you will have to reinstall.