Windows 7, Turbo enabled, HPET disabled test project running for 12 hours at 32 buffers.  Maximum DPC latency spike overnight 438us.

Same test project, same test settings HPET enabled.  Would not run at 32 buffers, minor crackling at 48 so run at 64 buffers.  Maximum DPC spike 2,600 + us after 45 minutes.  Retested 2308 us reported again after a few minutes, final try also ended in me not bothering further.

Which would you choose?

32 buffers on the Quad comes out at 3.52 ms @44,100k RTL measured by Centrance over USB 2, 5.1 ms reported in Sonar.

Who said the Quad wasn't low latency?...

yo jonbouy

did you read the thread i posted earlier?

jm24


yo jonbouy

did you read the thread i posted earlier?

Indeed I did, I've also come across it before.

I can understand Scott's frustration with this thread but at least we have two things going on here.  One is that Eric has posted from a position of being more than just a technical enthusiast and here we are (or should be) exclusively concentrating on DAW performance.

WRT to the Core Parking issues I find it hilarious that much of the research and suggestion that it should be turned of initially was done by Scott and pertaining only to Sonar and older chipsets (he provided a link to a thread where he proposed these ideas that many have since stuck with regardless of the fact things have moved on).  Now advocates of that now outdated idea are sticking to much of what Scott originally proposed, he's been saying for some while now since 1155 boards don't need to mess with turning all that stuff off turning HPET off in itself mitigates much of the problems originally caused, and now those same people are not listening.

There's no doubt in my mind because I base my findings in what I'm doing merely related to the Audio applications, OS and drivers I'm using currently, rather than basing it on some theoretical ideal that I've imagined.

That is to leave Core Parking, Hyperthreading, Speed Step, Turbo and C-States all enabled and merely turn off HPET base a Power Plan in Window on the High Performance on and turn off all the hibernation crap and you'll be getting the most from your CPU running higher performance at lower temperatures and get the best low latency performance from your particular interface.

"Jonbouy"
Sorry, I must dash I gotta go and impress my GF with my DPC latency figures...oh my.

"Jonbouy"
Windows 7, Turbo enabled, HPET disabled test project running for 12 hours at 32 buffers. Maximum DPC latency spike overnight 438us. 

Same test project, same test settings HPET enabled. Would not run at 32 buffers, minor crackling at 48 so run at 64 buffers. Maximum DPC spike 2,600 + us after 45 minutes. Retested 2308 us reported again after a few minutes, final try also ended in me not bothering further.

"Jonbouy"
Because it would be against the TOS for him to refer to the person in question that way.

Yet the other guy seems to be able to be consistently annoying for the sake of it with impunity. And for that matter how are you entitled to shoot your mouth off to another member with a personal insult?

Reference :: http://forum.cakewalk.com/fb.ashx?m=2390503

"Jonbouy"
I've never seen the forum in the state it's currently in where pretty much every thread is reduced to rubble where at one time there used to be reasoned debate. Sides used to be taken on point not personalities.

Reference :: http://forum.cakewalk.com/fb.ashx?m=2390520

"Jonbouy"
The art that is lacking here lately is to engage in the actual point not what you've judged the person making to it be like. That's just called prejudice.

Reference :: http://forum.cakewalk.com/fb.ashx?m=2390556

"Jonbouy"
Some people obviously have felt threatened by my doing that though judging by some of the real nastiness that's been directed at me the last few days, which I find strange and a bit disconcerting.

Reference :: http://forum.cakewalk.com/fb.ashx?m=2390609

And my favorite one...

"Jonbouy"
I don't recall ever having a desire to villify anybody on this board, I'd like to have the same courtesy extended toward me.

Is that a big ask?

Reference :: http://forum.cakewalk.com/fb.ashx?m=2391083

What was that about mirrors and whines again, Jon?

Reiterated for clarity's sake... (from my post #32 in this thread)

"Alegria"
The fact that I'm demanding of the tools I chose to work with has nothing to do with religion, shrines and rhymes. The fact that you also use your machine for non-audio related tasks, makes it much more difficult/unwise/impossible for you to optimize/streamline/tweak your machine as I would mine. The fact that you overclock your machine, albeit reasonably, makes it unwise for you to turn off core parking. It's all about choice, I completely agree. And these facts don't make you or me bigger/longer/holier than the other, matey. We obviously have different needs and budget to meet those needs.

"Jonbouy"
That probably wouldn't concern you though as usually you'd need freinds or colleagues in order to partake in a collaboration...

That's something you know nothing about outside of these forums and my wife and my kids are my friends, that's all I need. Anything else is..., a bonus if it's meant to be. And if I knew someone that had no friends, I would be the first one in line to offer my friendship, such as I did with "jimmy". But only to find out that he was someone else's imaginary construct. But just like you, I didn't donate.

From this thread: http://www.neowin.net/forum/topic/1075781-tweak-enable-hpet-in-bios-and-os-for-better-performance-and-fps/

""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.



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.
""

It seems clear to me the interaction of these timers will degrade performance.

How does this not make sense?

j

It makes perfect sense but it isn't what is wanted here.  HPET is working very well and performing more interrupts to the CPU for power management functions.
 
When we have the HPET time enabled these devices are constantly interrupting the CPU likely for PPM policy requests.
 
Instead of keep posting the same link which contains no information on it, re-read this bit that does.
 
http://forum.cakewalk.com/fb.ashx?m=2647881

JB: thank you for your patience and guidance.

Yo all of yous others: great exploration.

My summary:

Disabling HPET helps to obviate the impact of mal-coded drivers.

Setting windows power settings to high performance (with some tweaks) will reduce the impact of C state changes and core parking,...

If ALL drivers were properly coded, and we could disable all power reducing aspects of these drivers, using only HPET for timing would be prolly be the best choice.

j

Yo all of yous others: great exploration.

 
Yes it was a great exploration, and some pretty useful stuff from the guys at ADK.  It was a pity it turned into playground as there is some really good insights among the mud-slinging.
I think it is a real shame when we get input from some highly respected DAW builders that are dealing with this kind of thing daily on various systems for many users and they share their knowledge with us, only to have some point scoring smart alec's trying to big themselves up by being disrespectful, ignorant and contrary for the sake of it.
Thing is Windows isn't a dedicated DAW platform so we have to make the best use of it and give the dedicated Audio parts of it (like ASIO) the best chance possible against the underlying themes that Windows imposes such as networking and power management.
 
It's always about the best compromise.  And allowing turbo to run for performance sake and HPET disabled to mitigate the problems caused by other hardware drivers interrupts is easily the best way to go for the most power and effective low latency.  The testing is simple, as the differences are pretty obvious.