{Don't} Turn Off 10.9 App Nap

Moderators: Site Moderators, PandeGroup

{Don't} Turn Off 10.9 App Nap

Postby TomJohnson » Sat Feb 15, 2014 11:50 pm

Mac Mavericks' App Nap from zdnet 1/6/14 --
http://www.zdnet.com/mac-mavericks-app- ... 000024792/
"New in OS X 10.9 Mavericks is App Nap, a power-saving technology that looks at the applications that are currently being used and slows down the apps that it decides aren't being used. Apple says that if a program window is hidden from the user and isn't "actively working," then its process will be slowed down, perhaps as much as 23 percent..... "
Turn Off FAHControl's App Nap setting to see if your PPD improves.

Mod Edit: Edited Thread Title - PantherX
iMac 15,1 27" 5K Retina Intel i7 4.0 GHz Dual Quad 24 GB RAM 1 TB Fusion OS 10.10
Mac Pro 5,1 3.06 GHz 12 Core 52 Gig RAM OS 10.9.5 -- Plus 20 more
Image
TomJohnson
 
Posts: 271
Joined: Mon Dec 03, 2007 1:29 am
Location: Oak Harbor, WA

Re: Turn Off 10.9 App Nap

Postby calxalot » Sun Feb 16, 2014 12:40 am

Turning off app nap for FAHControl will decrease your PPD.
It's best to quit FAHControl if you're not using it anyway.

App nap does not apply to the client or cores.
calxalot
 
Posts: 439
Joined: Sat Dec 08, 2007 1:33 am
Location: San Francisco

Re: {Don't} Turn Off 10.9 App Nap

Postby TomJohnson » Sun Feb 16, 2014 8:16 pm

I am just running 10.9 on one computer, a Mac Pro 5,1 512GB SSD, via the default FAH 7.3.6 installation 24/7. FAH is never "open" and showing on the Desktop and therefore meets the App Nap criteria for being utilized; Norton Internet Security, Hardware Monitor, Little Snitch and smp Fan Control are the only other programs that are running in the background.
Note the significant increase in my PPD over the last 4 days since 2/11/14 when I turned FAHControl/App Nap Off. All I am recommending is that you try turning Off FAHControl/App Nap to see whether your PPD increases.
http://folding.extremeoverclocking.com/ ... s=&u=33541
TomJohnson
 
Posts: 271
Joined: Mon Dec 03, 2007 1:29 am
Location: Oak Harbor, WA

Re: {Don't} Turn Off 10.9 App Nap

Postby calxalot » Mon Feb 17, 2014 12:06 am

App Nap is for GUI applications, not daemons and agents.

If you run Activity Monitor, you can see what is being app napped. It is never the client or cores. For FAHControl to nap, you have to use the Hide menu item and wait about a minute.
calxalot
 
Posts: 439
Joined: Sat Dec 08, 2007 1:33 am
Location: San Francisco

Re: {Don't} Turn Off 10.9 App Nap

Postby TomJohnson » Tue Apr 01, 2014 5:43 am

App Nap test results for March 2014 with App Nap selected to "Prevent App Nap" --

For month of March, 31 days, I received 6,426,236 points for 2,740 WUs
For month of January, 31 days, I received 5,615,785 points for 2,302 WUs

WU points increased 810,451
WU production increased by 338 WUs

No changes were made to any of the 23 computers used in this test except for installing the F@H V7.4.4 update in mid March.
Seven computers were running OS 10.9.2 that had App Nap availability.

For month of April I will set App Nap ON to see if there is a noticeable difference in results ie not selecting “Prevent App Nap”

Data results can be viewed here:
http://folding.extremeoverclocking.com/ ... s=&u=33541
TomJohnson
 
Posts: 271
Joined: Mon Dec 03, 2007 1:29 am
Location: Oak Harbor, WA

Re: {Don't} Turn Off 10.9 App Nap

Postby P5-133XL » Tue Apr 01, 2014 6:23 am

Absolute points; WU's; or even points/WU are poor metrics. The problem is that the sizes of the WU's and their points vary far too much between different projects and the projects that are available changes over time. Average PPD is the correct metric because that is normalized to be relatively consistent for a given HW set.
Image
P5-133XL
 
Posts: 4034
Joined: Sun Dec 02, 2007 4:36 am
Location: Salem. OR USA

Re: {Don't} Turn Off 10.9 App Nap

Postby davidcoton » Tue Apr 01, 2014 5:06 pm

The ppd (or total points over 31 days, it amounts to the same thing) varied by about 14% of the earlier reading between the two periods quoted. There is no information as to whether that is within the normal range of variation, or January was abnormally low, or March was abnormally high. If the difference (either way) in April (relative to March) is about the same or less, this test will not be conclusive. Even if the difference is greater, particularly downwards, it will be difficult to draw conclusions without significantly more data.

In many cases FAH metrics are not useful just because of the random (?) variation in the baseline. It's not a flaw in your experiment, just a barrier to drawing conclusions from too few data points -- and more data points of themselves will not solve the problem. Unfortunately my ability with stats is only enough to spot the problem, not to solve it :(

David
Image
davidcoton
 
Posts: 940
Joined: Wed Nov 05, 2008 3:19 pm
Location: Cambridge, UK


Return to V7.3.6 Public Release Windows/Linux/MacOS X

Who is online

Users browsing this forum: No registered users and 1 guest

cron