DRanck
Hmm. This idea seems to be in direct contradiction to this:
http://forum.cakewalk.com/Possible-Key-to-SONAR-Performance-Breakthrough-Check-this-Out-m3103263.aspx
Unless I am missing something...

That thread is a completely different topic. Nothing to do with cpu clock. Good advice over there as well.

kevinwal
I gave this a whirl today. HPET was enabled in bios but not in windows. I enabled it in Windows  and my latency increased with my aging evga x58 Mb and i7/970 hex core cpu. On the other hand I experienced a significant boost in performance by disabling it at the bios and in windows. I also found that Intel speedstep was enabled and I turned that off, but the boost was noticeable even with speedstep off.
 
I've read that certain chipsets show this behavior and it looks like I won that lottery. Or something. Thanks for the idea!

Try running window update, updating your motherboard BIOS, and get latest chipset drivers from the intel driver update site.

Yes, all up to date, thanks. There is some literature on the web about specifics chipsets that do not benefit from HPET being enabled but it seems counterintuitive to me that it should be so. I'm still researching this.

That sounds interesting please post a link if it's any good...
 

mauryw
I found the HPET enable/disable option in my bios.  But, when I ran the command line in windows and tried to disable HPET with you command, I got the following error:
 
Microsoft Windows [Version 6.1.7601]
Copyright (c) 2009 Microsoft Corporation.  All rights reserved.

C:\Users\Larry>bcdedit /deletevalue useplatformclock
An error occurred while attempting to delete the specified data element.
Element not found.


What is wrong with this syntax?  How do I get the command to work?

Hi mauryw.
 
The error message indicates that you have not forced
Windows to use the platform clock exclusively to any other options.
bcdedit cannot delete an entry you have not set.
_____________________________________
That said, here's me eating crow...
 
for the last 8 months I have be noticing and silently blaming
Sonar for an ever so slight glitchyness in midi files I've loved for years...
nothing I could quite put my finger on, just not as tight as I remembered.
Ran my whole system on present specs(sig) only the cpu was O.C.'ed @4400.
 
1.
Research convinced me that my specs didn't require that, so cpu speed is then set @ 3800 via xmp
memory setting (recommended  by ASUS) but left  C-States, Speedstep  and TurboBoost off.
seemed better but something still niggling. stopped worrying about for a while.
THEN I SAW Doktor Avalanche's post...
 
2.
Did some more research just make sure I wouldn't hurt my system,
decided I HAVE to try this.
 
First thing I did was undo all my personal settings on the motherboard, set EVERYTHING
to DEFAULT.
Made sure HPET set to ENABLED in the bios. 
I wanted no variables to color my opinion so I did the same with the cpu, everything DEFAULT.
(Of course, every fan alert went off on reboot and had to be reset.)
 
So now it's time for the entry to be set...
followed the Dok's  procedure and I have to say...
WHEN YOU'RE RIGHT YOU'RE RIGHT.
 
DPC Checker came back with about a 30% INCREASE in latency...
but I noticed a PRONOUNCED INPROVEMENT in the tightness of midi AND audio; puzzling.
 
Still more research, however, indicated that this is due to the increased resolution
of the HPET, and it's enhanced ability to catch and report spikes
(already there, DPC Checker didn't catch them).
 
CONCLUSIONS:
 
My desktop's up in 8 seconds, down from 28.
ALL my music related apps are snappier, even look-ahead plugins have
ceased to glitch Sonar's engine, and best of all;
The FEEL has returned to my tunes.
 
Thank you, Doktor Avalanche.
 
Mel
 

Doesn't look like this works for me ...
 
I had pretty good performance and stability @ low buffer settings and 96 kHz up to Everett, but in the past 2 weeks I suffer from repeated random like crashes (only system changes were Everett & 4 NI updates of which I only have Kontakt in the project) ...
 
So I went through the whole exercise ... BIOS update, chipset drivers, etc. win 7 64 bit updates, then disabled speed stepping and the moderate overclocking I had in place ... no real change in stability, I do still get crashes when using the low buffer settings I used to use (128 @ 96 kHz) ... it even crashes at 192 samples, strangely this happens long before it gets even close to drop-outs from heavy system load ...
 
"bcdedit /set useplatformclock true" causes crashes to happen more frequently, sometimes even when hitting play after just booting and loading the project ... so I disabled again
 
BTW, LatencyMon reports super low latency in either case ...
 

I, personally, would stay away from this type of "tweak".
 
Unless you understand the Windows OS inside out, and are very "tech-savvy(?)", then leave well alone.
 
Many "Power Users" on here seem to get on well without the need for this type of "optimisation",  beyond disabling wifi and "Power" settings.
 
I have seen more people run into serious problems with "tweaks" like this, than I have seen people solve issues.
 
Unless you understand EXACTLY what you are doing, then leave well alone.
 
Be aware..
 
JMHO

jb101
I, personally, would stay away from this type of "tweak".
 
Unless you understand the Windows OS inside out, and are very "tech-savvy(?)", then leave well alone.
 
Many "Power Users" on here seem to get on well without the need for this type of "optimisation",  beyond disabling wifi and "Power" settings.
 
I have seen more people run into serious problems with "tweaks" like this, than I have seen people solve issues.
 
Unless you understand EXACTLY what you are doing, then leave well alone.
 
Be aware..
 
JMHO

I think this is pretty good advice because Windows is pretty good at choosing the right setting to use. This is one of those deep, murky Windows internals subjects dear to the hearts of uber geeks and I've been trying to determine what the best approach to this is for me. This is my interpretation of what's going on with this setting. I'm sure if I'm too inaccurate a more knowledgeable geek will step in and correct me.
 
To grossly simplify things, when Windows is installed it inspects the hardware for the optimum time keeping mechanism to use for its various and sundry scheduling needs. If the system supports the invariant Time Stamp Counter (TSC) which most newer Intel chips do, Windows will disable the use of HPET, preferring the low overhead and precision of the TSC. This assumes of course that HPET was enabled in BIOS in the first place; if it isn't and the TSC is supported it will use the TSC. If the TSC isn't supported and HPET is, it will use that, otherwise it will use ACPI PM timer. Many BIOS's don't even allow the enabling of HPET these days, a further indication of HPET's growing legacy status.
 
So, what is the guidance on this? Here's my take, YMMV, etc.
 
1. If HPET was enabled in your BIOS but Windows has it disabled, your system is optimized for the best performance in timekeeping.
2. If your BIOS has no setting for HPET, your machine is probably running the preferred TSC timekeeping unless it is a very old or non-Intel box, and there's no need to change anything (indeed there is nothing to change.)
3. If HPET is disabled in BIOS, you can try this tweak to see if the performance improves, keeping in mind that if your processor is an i5 or i7 of a newer variety, you'll probably not see an improvement. But hey, you might!
 
One last thing; there shouldn't be any danger trying this tweak if you're careful with the typing.

jb101
I, personally, would stay away from this type of "tweak".
 
Unless you understand the Windows OS inside out, and are very "tech-savvy(?)", then leave well alone.
 
Many "Power Users" on here seem to get on well without the need for this type of "optimisation",  beyond disabling wifi and "Power" settings.
 
I have seen more people run into serious problems with "tweaks" like this, than I have seen people solve issues.
 
Unless you understand EXACTLY what you are doing, then leave well alone.
 
Be aware..
 
JMHO

I will test this and get some bench marks...
 
I am pretty ok with the internals of windows...I think.
 
Built the system I'm on... got several bkup's if I shaft somethin'...
 
I did extensive testing with my I7 2600k and the Asus board in my sig. a couple yrs. ago and the conclusion
I came to is that whether a given "tweak" (E.G. O.C'ing) works or not is
often system setup and motherboard specific.
 
Given that I am confident that I will not screw up my system(beyond repair*)
my general tendency is to try it(the tweak) on a non-critical system,
make an assessment regarding whether or not it improves my musical/production
experience and remove it if I judge it wanting.
 
* I back up my entire sys. every two weeks.
#NO FEAR
 
Mel

That's the spirit! I found that my best performance is with HPET disabled in the BIOS, but as with  the Doctor's results, many find an improvement.