Discussion in 'Videocards - NVIDIA GeForce Drivers Section' started by Bukkake, Sep 18, 2012.
Not latest Windows 10, I guess?
Hi thank you it's 1709, never updated.
Is it normal or is it a "buggy" implementation, or it's the same on all 7y30 ?
I appreciate the help.
I see no help needed. I mean it can be that CPU (or ACPI) in your laptop implements TSC with that frequency, or Windows configures the QPC/QPF implementations due to laptop platform.
You should not have issues from 1.570 MHz QPC timer.
Do you witness bad behaviour in timer benching apps?
Yeah no issues, beside the acpi.sys spikes in latencymon, but they disappear when I disable the "ACPI battery control" in the Device Manager.
It's just that it's the FIRST TIME ON THE INTERNET that I have seen this timer, and the fact that is roughly half "standard" TSC frequency, made me think of a bad implementation, of badly configured bios (update don't fix the timer).
TSC should be in the CPU, then it should be the same for all 7y30 chip at 1.750MHz , right ?
It could not be a bad bios config from the manufacturer ?
I read BCLK affect the timer, but mine is at 100MHz in CPUZ.
Could it be affected by the RAM speed or bus ? Because I saw my RAM speed vary from 1066 to 1866 dynamically, until I disabled "SA geyserville" in the bios.
So can it be related to the ram or the System Agent configuration ? Possible or not ?
QueryPerformaceCounter and QueryPerformanceFrequency are Win API functions. They do not return raw hardware timer/counter output.
You can check the BCD configuration by executing "bcdedit" in elevated command prompt.
But I suggest you can relax. Issues should alarm you, not some numbers.
is it bad having a 1.570mhz qpc ? you dont even know what it is..how it works..there is no "standart" QPC before build 1809..it is related to your hardware..so it doesnt have to be around 3.2mhz..so the problem is your misunderstanding about qpc timer..
Hi guys and sorry for my questions.
But have you ever seen a 1,570 Invariant TSC before my post ?
I read most of all the thread on forums and it was the first time I saw this number, everyone is around 3.XX MHz (the only exception I saw was around 2,7X MHz).
That is why I posted here because you are more knowledgeable than me and I was thinking if it is a problem you might know about it, it's true that I don't know anything about timer I just compared my results with yours.
I know it might be irritating to you guys but I did not have a choice but to ask.
SatsuiNoHado you said it is related to my hardware, and since TSC is in the CPU, SO everyone with the same CPU Intel 7Y30 should have the same timer on Windows 10 before 1809 ? Right SatsuiNoHado ?
EDIT : I forgot to thank you mkb1969 for your answer, so thanks.
In my Bcdedit even with :
bcdedit /set useplatformclock No (I know I can just deletevalue this)
bcdedit /set disabledynamictick Yes
I have 1.750MHz Invarient TSC.
If I add useplateformticks same value of 1.570MHz
If I useplateformclock yes, I have HPET a 24MHz.
yes i ve seen different tsc values more than you can imagine..and i can make mine down to around 750mhz..
more than that i can make it what i want it to be..!!! stop complaining for nothing and enjoy your pc the way you like..
and one more thing to add for bcdedit settings so i can answer to this whole topic too..
after a fresh windows installation for tweaking purpose about bcdedit options just change as below nothing more or less;
bcdedit /set disabledynamictick Yes
bcdedit /set useplatformclock No
bcdedit /set useplatformtick yes
and LOCK this thread..!!!
Ok thank you guys, have a good day.
And again sorry if it was bothersome to you.
Ignore SatsuiNoHado. I would understand if he was old member and spent hours and hours battling in this thread. Also he started to advise unnecessary BCD tweaks. May be "disabledynamictick" and "useplatformtick" are worth to try and test (not believing in universal advises on modern PC platforms), but setting "useplatformclock" is not needed.
We've been over this. Thread is 80 pages long. This is how you optimise your timers:
1) Install the latest Windows 10
That's it. Windows no longer needs to be tweaked for timers and will automatically use the best combination for your system for optimal performance.
If you've been experimenting with the BCD commands, here's how to get back to 'default', the same as a clean Windows 10 install:
bcdedit /deletevalue disabledynamictick
bcdedit /deletevalue useplatformclock
bcdedit /deletevalue useplatformtick
There is no harm in changing these values, change and test away. It's easy to change back, as @Martigen mentioned, and there is no lasting effects.
But, the test part is probably the most important. For something to be "recommended" you need real proof.
Without that, the best you can hope for is "i feel like these changes help my computer, but recommended is default".
Thank you for the kind response and help mbk1969, I read your post from long time ago (years) and you were always helpful, and knowledgeable.
I am not a pro at timers, but I am an audio engineer, and I tested my laptop with all combinations of settings.
It does not make a "real" difference under LatencyMon, or when I record in real time (64samples).
Even disable "disabledynamicticks yes" does not make a difference, but I keep it because I disable every energy saving features on my laptop.
I also experienced with the 0.488ms vs 0.5ms (useplateformticks yes) and latencymon is the same,on Windows10 1709. I can't feel the change under Cubase or Live.
I am not a gamer yes, but I work with real-time audio processing (it should be close enough), and there is no difference with those settings.
The only difference is between using HPET vs using Invariant TSC. Use TSC.
If you want low DPC and ISR disable everything running "in the backside" of windows, or in the hardware, the invisible programs causing interrupts. The less stuff your PC has to schedule/compute/manage, the faster the response time.
Thank you for your help.
I am curious whether you toyed with MSI mode - https://forums.guru3d.com/threads/w...ge-signaled-based-interrupts-msi-tool.378044/
It was the next step on my radar, after I tackled the timer issues and settings.
Couple of years ago, on my desktop, I remember setting everything to MSI mode with your tool, but I can't remember... I wasn't as anal, and my computers where not "as optimized" as today.
Do you need me to test and post my results ?
I will do it and post my results when I finish battling with an EDID override in Windows.
Personal big thanks to you mbk1969, and for the work you put in all these years.
I would love to hear your comparison "legacy mode vs MSI mode" from your professional engineer point of view.
OK I will post them in the other thread when I am finished.
Also try to toy with interrupt priority (for audio card?), please. Since nobody knows what exactly that setting affects (and whether affects at all)...
Yes I will see.
I don't know what it is for, but I will change the settings and see If it does change anything when I record or use the audio card in a realtime playback scenario (with multiple realtime effects, like a compressor and equalizer).
But don't expect too much from me I am not that well versed in computer science.
But I will definitely see if it make a different as in realtime audio DPC or ISR spike cause "crackles" in the audio. I will also check in LatencyMon.
More detailed way comparing to LatencyMon