TWEAK: Enable HPET (in BIOS and OS) for better performance and FPS


Recommended Posts

By default Windows 7 uses different timers in the CPU to calculate stuff. HPET is the newest and best of these timers, but because of default combination of timers it takes longer time for CPU to keep up all the timers and sync between them. Forcing Windows to use HPET only improves performance and leads to greater FPS.

Steps to enable this tweak:

1. Enable HPET in BIOS. If you have HPET option in BIOS then your hardware can support HPET.

2. Enable HPET in Windows by giving this command in admin credential CMD:

bcdedit /set useplatformclock true

3. Reboot

  • Like 2

Enabling HPET in BIOS is just half way of enabling HPET, it needs to be enabled in OS too, and in a way that it's the only timer used.

By default windows uses combination of TSC+ACPI timers, not matter if HPET is enabled in BIOS.

TSC+LAPICs Low performance (slow timers + syncing)

LAPICs low performance (slow timer - no syncing)

TSC+HPET medium performance (slow and fast timer + syncing)

HPET high performance (fast timer - no syncing)

HPET + platformclock=true will give you best timer resolution, frame rate and lowest DPC latency.

You can test timer ratio and QueryPerformanceFrequency with WinTimerTester 1.1 http://www.mediafire...xzo9n84d8lze9nb

The higher the QueryPerformanceFrequency is the better is performance. You only get high frequency with HPET. The other timers will give you significantly less frequency. Also note that if your ratio is not 1.0000 you are off set (or you have wrongly OC'ed), enable HPET and you should be without sync problems.

If you ever want to go back to default timers admin cmd:

bcdedit /deletevalue useplatformclock

Varying depending on setup, one should get increase up to +30 FPS and from the between.Online games is a good example of boost from HPET.

TSC+LAPICs Low performance (slow timers + syncing) = 2.76MHz

LAPICs low performance (slow timer - no syncing) = 3.5Mhz

TSC+HPET medium performance (slow and fast timer + syncing) = 3.8Mhz

HPET high performance (fast timer - no syncing) = 14.3MHz

Run the WinTimerTester 1.1 to see your QueryPerformanceFrequency

Then try with HPET, you'll be amazed.

Enabling HPET in BIOS is just half way of enabling HPET, it needs to be enabled in OS too, and in a way that it's the only timer used.

By default windows uses combination of TSC+ACPI timers, not matter if HPET is enabled in BIOS.

TSC+LAPICs Low performance (slow timers + syncing)

LAPICs low performance (slow timer - no syncing)

TSC+HPET medium performance (slow and fast timer + syncing)

HPET high performance (fast timer - no syncing)

HPET + platformclock=true will give you best timer resolution, frame rate and lowest DPC latency.

You can test timer ratio and QueryPerformanceFrequency with WinTimerTester 1.1 http://www.mediafire...xzo9n84d8lze9nb

The higher the QueryPerformanceFrequency is the better is performance. You only get high frequency with HPET. The other timers will give you significantly less frequency. Also note that if your ratio is not 1.0000 you are off set (or you have wrongly OC'ed), enable HPET and you should be without sync problems.

If you ever want to go back to default timers admin cmd:

bcdedit /deletevalue useplatformclock

Varying depending on setup, one should get increase up to +30 FPS and from the between.Online games is a good example of boost from HPET.

A guy on another forum claimed that by disabling it, his DPC latency plummeted and his gaming performance improved; most notably the microstuttering.

So I tried it myself, and the difference to my astonishment, was very noticeable. First off, my DPC latency dropped big time. Not that it was high to begin with, but now it hovers below 10 for the most part.

Second, I noticed that general windows performance felt snappier. I know I'm not imagining things, because like many, I scrutinize my system so I'm sensitive to even minute changes in performance.

And last but not least, gaming performance is definitely smoother. Anyone with SLi will tell you that occassionally when playing a game, you experience a bit of lag or stuttering for no apparent reason whatsoever. Most people attribute these minor discrepancies to SLi, but what if it's something else? Because I can tell you, that my overall gaming experience has improved since turning HPET off.

If you have the HPET option in your BIOS, I highly recommend experimenting with it and see if it impacts your performance. There's a possibility that having it on could increase the amount of microstuttering in any given game; particularly if you're running SLi.

http://forums.nvidia...howtopic=183329

Without changing any settings in the BIOS or the OS, and with HPET enabled in the BIOS as normal I get this

Untitled.png

^Because if you enable HPET in BIOS you will mix it with other timers and CPU needs to sync between them. You need to force HPET only and reboot. Mixing HPET with other timers is not good, but using HPET only will give you best performance. People don't know about that Windows needs to be forced to use HPET only - that's why the bios setting alone can undergrade their performance.

bcdedit /set useplatformclock true (then reboot) enable HPET

bcdedit /deletevalue useplatformclock (then reboot) disable HPET

Without changing any settings in the BIOS or the OS, and with HPET enabled in the BIOS as normal I get this

Untitled.png

You are using TSC+HPET. Try with HPET only (bios and OS setting) you should get something like 14.xMhz for queryperformancefrequency.

^Because if you enable HPET in BIOS you will mix it with other timers and CPU needs to sync between them. You need to force HPET only and reboot. Mixing HPET with other timers is not good, but using HPET only will give you best performance. People don't know about that Windows needs to be forced to use HPET only - that's why the bios setting alone can undergrade their performance.

bcdedit /set useplatformclock true (then reboot) enable HPET

bcdedit /deletevalue useplatformclock (then reboot) disable HPET

Ok so I enabled it, rebooted

2.PNG

First things I notice are it took twice as long to finish loading the desktop and gadgets

Mouse pointer has a ghost trail now like I am using a cheap LCD with 20ms response rate

What exactly is "queryperformancefrequency" ? and how does it effect the OS ?

queryperformancefrequency is frequency of high resolution timer performance counter, timers are different as I have explained above, and better/higher the timer frequency the better responsive the system is. With HPET enabled in BIOS and OS you can achieve highest timer performance resluting in better responsive system.

This "tweak" is depending on each hardware setup. Some people say it slightly speeds things up, others start having micro-stutter in multimedia/3d gaming, especially SLI setups. the DPC latency goes through the roof and limits the FPS in the games. Best thing to do is try it for yourself, do some benchmarks and see what happens, if your system runs fine now. Leave it alone would be my advice.

Enabling HPET in BIOS mixes HPET with other timers - needing the CPU sync between the timers resulting in compromised performance. If WIndows is forced to use HPET only (in bios and os) it in modern system results in higher performance, higher FPS and better DPC.

Test and try what suits you.

Enabling HPET in BIOS mixes HPET with other timers - needing the CPU sync between the timers resulting in compromised performance. If WIndows is forced to use HPET only (in bios and os) it in modern system results in higher performance, higher FPS and better DPC.

Test and try what suits you.

I had mouse ghosting and other nice things when doing it "your" way. Disabling HPET is the best way for me on a Sandy Bridge-E CPU and an ASUS Sabertooth motherboard.

commands:

bcdedit /set useplatformclock true (then reboot) enable HPET

bcdedit /deletevalue useplatformclock (then reboot) disable HPET

TSC+LAPICs (useplatformclock false)*

LAPICs (useplatformclock true)**

TSC+HPET (useplatformclock false)***

HPET (useplatformclock true)****

* Windows default.

**platformclock=true and HPET disabled in BIOS will default to LAPICs, which is good compared to TSC, but doesn't not have not so high resolution and so low DPC latency as HPET.

*** Windows default with HPET enabled in BIOS.

**** HPET enabled in BIOS and in OS.

It seems to have had the opposite effect here too. I had HPET enabled in the BIOS already but hadn't set the useplatformclock boot parameter.

Without it, my clock ratio was mostly 1.0000, occasionally flickering to 1.0001. After setting it my clock ratio is mostly 0.9999, occasionally 1.0000.*

Here's the original post from the author of WinTimerTester 1.1. It explains the scenario better. Seems to be related to software overclocking utilities.

The problem is that for many computers, this function in Windows 7 is now based on the clock speed of the processor. If you overclock in the bios and then boot up, this timer is calibrated and will work 100% correctly. If you use SetFSB or a similar program and you change the bus speed from within Windows, you have now just screwed up this very important timing mechanism. The number of applications that can choke after you do this is surprising, especially games.

* After a few minutes it seems to have settled at 1.0000. I don't have any mouse ghosting.

I had mouse ghosting and other nice things when doing it "your" way. Disabling HPET is the best way for me on a Sandy Bridge-E CPU and an ASUS Sabertooth motherboard.

Funny as I don't have any problems like that. my CPU is i5 3570K in Z77 mobo and Asus HD 7850. I do use CRT monitor.

Perhaps it's something else you need to adjust if you want to have best timer.

Well RAGE was smooth when it ran, but every few seconds was locking up for a few seconds

BF3 let me play it with CFX for the first time so it seems to have helped there

I`ll test a few more games but it seems to have negatives and positives for different things so far

I've read many people with the Intel X58 chipset have issues with HPET, checked my BIOS and it was enabled (in 32-bit mode yet I run a 64-bit system). Disabled it because lately my computer has hard locked twice while browsing the internet (hasn't done it in awhile though). So maybe my X58 chipset + HPET in 32-bit was part of the issue. Either way my system seemed more responsive on boot than usual, won't be able to tell if it helped until I get a chance to use it more or see if it freezes for no reason again.

I've read many people with the Intel X58 chipset have issues with HPET, checked my BIOS and it was enabled (in 32-bit mode yet I run a 64-bit system). Disabled it because lately my computer has hard locked twice while browsing the internet (hasn't done it in awhile though). So maybe my X58 chipset + HPET in 32-bit was part of the issue. Either way my system seemed more responsive on boot than usual, won't be able to tell if it helped until I get a chance to use it more or see if it freezes for no reason again.

I have X58 chipset boards and never saw an issue with 64bit HPET

Funny as I don't have any problems like that. my CPU is i5 3570K in Z77 mobo and Asus HD 7850. I do use CRT monitor.

I have 5?s latency with HPET disabled and 60?s with it enabled using your method, plus no mouse ghosting or stuttering which I get with it enabled. There is nothing else concerning this that I can 'adjust'. I guess my X79 board just feels better with it disabled or something.

Is this on TSC+LAPICs?

wintimertester.png

Plus I'm unsure if the Asus P7P55D Pro has a HPET toggle, will have to check later.

Hard to tell, it's either TSC+LAPICs or LAPICs. Your ratio is excellent so it might indicate that there is no sync - meaning you might have LAPICs. However hardware differs and even if your ratio is that exact in 1.0000 it still might sync TSC+LAPICs well - depending on hardware. Windows obviously uses LAPICs mainly for hypervisor/virtual machine and TSC as general, but it's a bit mystery what Windows 7 uses, a combination is the best assumption. Vista was heavily pushed into using HPET. I would say that you have at least TSC. If you want TSC + HPET enable HPET in bios. And if you want to test HPET only use the bcdedit /set useplatformclock true.

I think that your mobo has HPET in BIOS as HPET isn't actually anything new.

-----------------------------

Good to hear that people are getting good results with HPET (bios & os). However I have provided details about timer settings in the first page and how to revert back from HPET.

Hard to tell, it's either TSC+LAPICs or LAPICs. Your ratio is excellent so it might indicate that there is no sync - meaning you might have LAPICs. However hardware differs and even if your ratio is that exact in 1.0000 it still might sync TSC+LAPICs well - depending on hardware. I think that your mobo has HPET in BIOS as HPET isn't actually anything new.

Couldn't find anything in the BIOS, think it's hidden. Though I ran bcdedit /set useplatformclock true and I'm getting 14.31818 MHz now. But the ratio took longer to stabilise at 1.000 (took 180 seconds), it was alternating between 0.9997 and 0.9999 a lot.

Not spotted anything bad yet, like ghosting mouse pointer or such. My DPC latency remains unchanged, averages at 60-80us

Im running it on a i7 [email protected] (clocked in BIOS), and Asus P8Z68-V Pro mobo/NV 570GTX and enabling HPET & setting windows to solely use HPET via command prompt has fixed my micro-stuttering id see in BF3.

Result :) Cheers OP I owe ya a beer :)

This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
  • Posts

    • Uninstalr 3.1 by Razvan Serea Introducing Uninstalr: Easy to use and very accurate software uninstaller for Windows. It can uninstall multiple apps at the same time and we think it’s pretty cool. Developed with expertise by Macecraft Software - the minds behind jv16 PowerTools. Key Features Batch uninstall many apps at the same time. Supports unattended uninstallation of apps. Supports monitoring of new software installations. Also detects portable apps and previously uninstalled software leftovers. Shows all the data added to your system by installed software on a file by file basis. Shows all the data it will remove before starting the uninstallation. Filter and search the list of installed software. According to our benchmark, Uninstalr is the most accurate software uninstaller by leaving the least amount of leftovers when uninstalling apps. Supports detection and uninstallation of Microsoft Store, Steam, Big Fish Game System, Chocolatey, NuGet and Ninite installed software. Supports Windows Dark Mode. Supports Windows 11, 10, 8 and 7. Comes with these translations builtin: Chinese Simplified, Chinese Traditional, Czech, Danish, English, Filipino, Finnish, French, German, Greek, Hindi, Hungarian, Indonesian, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Romanian, Slovak, Spanish, Swedish, Thai, Turkish, Ukrainian and Vietnamese. Has a single executable file portable version and a normal setup version. Uninstalr is freeware, lightweight and easy to use. No bells and whistles, no nonsense. Uninstalr’s custom uninstallation engine has a dedicated support for the detection and uninstallation of 15 types of apps: Normal Windows apps Microsoft Store apps Portable apps Chocolatey apps Ninite apps PortableApps.com apps Steam games EA App games Epic Games Store games Riot platform games GOG Galaxy games WarGaming.net games Battle.net games itch.io games Big Fish platform games Uninstalr 3.1 changelog: Key Changes Uninstalr now starts and shows the list of installed apps faster after the initial scan has been completed, and with much smaller memory usage. Uninstalr now detects and highlights apps that automatically start with Windows. Greatly improved the detection of portable apps. Improvements New feature: Uninstalr now detects and highlights apps that automatically start with Windows. New feature: Uninstalr now highlights possible leftovers and apps from Russia and China. This can be disabled from the Settings. New feature: A new filter that allows you to show only software that is installed to other than the system drive. New feature: Users can now select to always do the deepest and the most accurate scan for installed apps, at the cost of the analysis taking a longer time. Greatly improved the detection of portable apps, such as added dedicated support for MiTeC, EZ Tools and SysInternals tools. Improved support for portable apps installed via Windows System Control Center (WSCC). NirSoft portable apps are now listed with "NirSoft" prefix for easier identification. Improved the speed of uninstalling apps. The main installed software listing search will now find "Xbox GameBar" if you search for "Game bar" and vice versa. The tooltip now displays more detailed information of the installed apps, such as its registry key and uninstaller path. The links in the About section now look more like clickable links. The main menu is now more clearly indicated in the main user interface. Microsoft Teams Meeting Add-in for Microsoft Office ships with some Windows 11 installations and is now considered a builtin Windows app and only listed if builtin Windows apps filter is enabled. Added a Help button to the main user interface that opens the help section of the website. Added an option not to close Uninstalr after uninstallation. If you open the Uninstalr website from the app, the website now receives the version number of your current Uninstalr version and warns you if you are using anything but the latest version. Improved the accuracy of the New Software Monitor. Improved confirmation messages for Steam and other platform related uninstalls. Improved the uninstallation performance of Steam games. Fixes: Known bug fixed: Some installed app names are capitalized incorrectly, such as "CCleaner Portable" is listed as "ccleaner portable". Known bug fixed: Some apps can be listed twice, for example, Smart Defrag can be listed once as Smart Defrag and then Smart Defrag Home. Known bug fixed: On the pre-uninstallation screen, the Scripts checkbox can be checked by default on Dark Mode but not on the normal mode. Known bug fixed: Perform Deep Analysis can be started only by clicking the button, not via the Right Click menu, main menu or F4 keyboard shortcut. Muse Hub could be incorrectly listed as Adobe Muse. SyncTrayzor was incorrectly detected as two unrelated software, SyncTrayzor and Syncthing. Smart Defrag was incorrectly listed twice as Smart Defrag 11 and Smart Defrag Home. It was possible to enter non-printable characters to the search input boxes of the main screen, and the path listing screen, which caused the UI to look funny. Changing the translation from Settings, especially many times in a row, caused the UI to distort. If you had multiple instances of portable apps on your system, such as the 64b and 32b versions of the same portable app, typically only one of them was detected, not both. In some very rare cases, Uninstalr UI could start with random characters in its search input boxes, which could make the UI look rather confusing. This was a rare issue, only reported by two users. The pre-uninstallation screen could display non-existing paths for example as the software's installation directory or main exe file. This was a cosmetic issue. New Software Monitor cannot detect the installation of Claude. Selecting all the found software made the UI look funny with the top panel covering everything else (because the names of all the selected software were listed there). Sometimes a Steam game could be listed a normal app instead of a Steam game. If the system restart after an uninstallation is delayed, e.g. because of Windows Updates being installed, this additional delay is incorrectly added to the time how long the uninstallation process took. This cosmetic bug could cause the program incorrectly report an uninstallation time longer than the actual uninstallation time. Uninstalling Minecraft could simply fail. The Only scan the system drive for installed apps setting does not fully work. If some apps are installed to a non system drive and this setting is enabled, the app could still be detected and listed on the main user interface. Changing any settings could also incorrectly alter the Only Scan The System Drive For Installed Apps setting. Microsoft OneDrive and Copilot are not always detected. If you enter something to the search filter field, then select the text and press the Delete key, this triggers the Uninstall button click even if your intent was to delete the text input. If you press the F5 key to refresh the screen during the uninstallation loading screen, the program will crash. If you enabled some setting, such as "Do not analyze installed app installation sizes", it could automatically be unchecked later. Uninstalr doesn't warn you if you try to remove Fortec antivirus. There should be a warning if user attempts to remove any antivirus or antimalware type program. Such programs should not be uninstalled using a third party uninstaller, as they are typically protected against automated uninstallation, for security reasons. With "Do not analyze installed app installation sizes" option checked from the Settings, Uninstalr could still display some installation size related elements in the UI which was confusing. The "Only scan the system drive" option moved under Improve Scan Speed from the General settings. If two software have the exact same name and version number, selecting both of them for uninstallation fails because only one is actually selected. Sorting the installed apps by size sometimes fails and the order is incorrect. The "Don't show which paths are currently analyzed" did not work correctly - some parts of the UI still show the currently analyzed path with this setting checked. The "Don't list software less than 10 MB" filter did not work correctly - some apps smaller than 10 MB could still be listed. Uninstalr could start very quickly and display an empty list of detected apps. Restarting the app usually fixed the issue and the list of installed apps was properly displayed. If you placed portable Uninstalr to a same folder with other portable apps, those were not detected because Uninstalr automatically added its installation folder to the ignore list. When trying to uninstall some specific software, Uninstalr could get stuck on the Searching for more data relating to the app phase. Uninstalr could sometimes do a silent uninstallation even if user had unchecked the Perform a silent uninstallation option. Known issues: Uninstalr can fail to run with an Out Of Memory error in systems that have a lot of installed apps. Using the New Software Monitor tool multiple times during one session can cause the program to get stuck on the Scanning stage. The "uninstallation completed" message box sometimes closes when the user moves the mouse cursor over the button before user clicks it. There is no feedback for the user after Fix Information feature has been used. The Right Click menu's Select by publisher option can display the number of apps per each publisher without correct vertical alignment. The default user interface might not display all of the found installed apps if you have over 600 installed apps. If you do, using the Screen Reader Compatible Interface solves the issue. Leftover apptype filter checkbox is shown in red font only in Dark Mode. Clicking the app's icon from the Windows Taskbar doesn't minimize/restore the app like other apps. The warning about an app that user wishes to uninstall being related to some other app user did not select can sometimes be inaccurate. If app's language is changed without restarting Uninstalr, the list of installed software might not automatically refresh. When software is being uninstalled, the UI can say it is processing paths unrelating to the uninstalled app. This is purely cosmetic and does not mean these paths are removed. Uninstalr might not properly detect and/or uninstall Steam games if they are installed to a drive different than Steam's default location in C:\. You might see "This action is only valid for products that are currently installed" error message from Windows Installer during uninstallation. This is a cosmetic issue. Download: Uninstalr 3.1 | 7.1 MB (Free, paid version available) Download: Uninstalr Setup 3.1 View: Uninstalr Website | Screenshot Get alerted to all of our Software updates on Twitter at @NeowinSoftware
    • I and many others did not vote to get out of the E.u because of Putin or Farage, we did so for our own reasons. You don't have to tel me what my own did or did not do when it comes to the E.U. The EEC is or was the European Economic Community, a different beast to what the E.U is now.The EEC was a mainly about trading, the E.U have gone far beyond that and as I have said before, is now more of a United States of Europe. The U.K did not vote to join a United States of Europe. Anyway, they did not want us in there in the first place, Charles de Gaulle stopped us joining as he claimed we didn’t agree with the core ideas of integration. He was not wrong and that is why we voted out of the E.U when the time came. I was not old enough to vote the first time. My only regret is that we did not have the referendum years ago and got out years ago. If we rejoined, we would have to agree to join the Euro and no doubt Schengen, agree with freedom of movement, we have enough problem with people coming over here as it is. i have no problem with people coming over here if they work and don't try to push their way of life onto us. The E.U has a currency, freedom of movement, an anthem a flag, a parliament, well they are there, not sure if they do anything. Don't sound like something that is just for trading. Oh yeah, also wanted a euro Army. How many stupid rules have the E.U made that we had to follow? I doubt I will see the Uk rejoin the E.U, which suits me. Oh yeah, my partner is Polish, she came over here before Poland joined the E.U and she got fed up of people just coming over here with ease, while she had to struggle. She is now a British citizen and have been for a fair few years
    • Hello, Paul. Thanks for the editorial. It was interesting. I'm going research more into the app and its concept. Of course, if you know me at all, you know that I'd say your articles needs some editing! I always do, don't I? For instance, the article occasionally mentions relays before defining it.
    • Screamer is 50% off on Steam, making it £24.99 here in the UK: https://store.steampowered.com/app/2814990/Screamer/ You might remember the series from the mid 90s / early 2000s, this new game is also by Milestone who created the older games.
    • U.S. partially reverses Anthropic AI ban for Mythos but keeps Fable 5 off the market by Karthik Mudaliar Anthropic says that the U.S. government has finally allowed it to restore Claude Mythos 5. But of course, there's a catch. The rollout is again for a limited set of U.S. organizations that operate and defend critical infrastructure. The company announced this in a post on X (formerly Twitter). This does not mean that Anthropic's latest frontier models are back to normal availability. Fable 5, which was a tuned version of Mythos 5 for public release, remains unavailable. Anthropic said that it is still working with the government to expand Mythos 5 access and make Fable 5 available again, but there's no timeline. Reports from Bloomberg and Reuters say that this decision actually came through a letter from the U.S. Commerce Department. According to Reuters, this would allow more than 100 companies and institutions access to Mythos 5. Reuters also reported that Commerce Secretary Howard Lutnick’s letter removes the need for export licenses for approved companies’ non-US citizen employees, as well as Anthropic’s own non-US citizen employees, while restrictions remain for organizations outside the approved list. Anthropic isn't alone with this kind of controlled rollout. OpenAI's newest model family, GPT 5.6, was announced just yesterday, but isn't available for everyone yet. In its announcement, OpenAI also said that access to these models is initially limited to a select group of trusted partners and organizations, with broader access planned later this year. Both of these cases show that frontier AI launches are no longer just ordinary product releases and more like slow and vetted deployments shaped heavily by the U.S. government.
  • Recent Achievements

    • Week One Done
      flexorcist earned a badge
      Week One Done
    • One Month Later
      Woland13 earned a badge
      One Month Later
    • Week One Done
      Woland13 earned a badge
      Week One Done
    • One Year In
      bernmeister earned a badge
      One Year In
    • Week One Done
      Scoobystu earned a badge
      Week One Done
  • Popular Contributors

    1. 1
      +primortal
      495
    2. 2
      +Edouard
      226
    3. 3
      PsYcHoKiLLa
      154
    4. 4
      Steven P.
      75
    5. 5
      FloatingFatMan
      71
  • Tell a friend

    Love Neowin? Tell a friend!