Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Valkyre

macrumors 6502a
Original poster
Dec 8, 2012
525
410
So to the people who have installed Mavericks, what is your impression regarding RAM usage and such?

I know it is a BETA still and there is a lot of work to be done, but are the Mavericks more RAM hungry than ML or are they looking to require less ram, have less memory leaks etc?

They said that safari will not be as memory hog as it was in ML.

Please give us your impressions. ;)
 
My impressions: 10.9 seems to use the free RAM more aggressively for caches. Overall the system seems more responsive. Safari definitively uses less RAM. Funnily enough, kernel_task uses 0 (zero) RAM at my system, no idea if its a bug or a feature ^^

I also tried starting some very memory-intensive calculations just to see how the system responds under such circumstances. The RAM compression feature is just amazing. My script quickly grabbed over 17GB of RAM, but the swap area remained under 1GB and the system was still responsive (I could switch to Safari, open new tabs, navigate the Finder, even write some of this message) - of course slower, but still really responsible.

Overall conclusion: They seem to have made some massive changes to the RAM system.. and it appears to work quite well. I really like the new Activity Monitor btw!
 
As a side note, the recent security update for 10.7.5 seemed to have some changes around memory that improved performance. I expect what is in Mavericks is more of the same and a good thing.
 
Funnily enough, kernel_task uses 0 (zero) RAM at my system, no idea if its a bug or a feature.

If you show real memory it uses pretty much the exact same amount of ram as ML. 677.9MB for me right now.

Overall conclusion: They seem to have made some massive changes to the RAM system.. and it appears to work quite well. I really like the new Activity Monitor btw!

Hard for me to say with only a few hours use but I will agree that it definitely appears to have improved memory usage/handling over ML.
 
Here's what I have going on with a mid-2009 13" MBP. When I took this screenshot I was running TextWrangler, Mail, iTunes, Safari open with 6 tabs, and Quicktime X playing a 1080p movie.
 

Attachments

  • Untitled1.png
    Untitled1.png
    172.4 KB · Views: 3,326
If you show real memory it uses pretty much the exact same amount of ram as ML. 677.9MB for me right now.


Hm, what is the difference between 'real memory' and 'memory'? I assume that 'real memory' is the amount of memory allocated by the process and 'memory' is the amount physical RAM currently mapped to it?

Edit: no, it appears that 'real memory' is actually the amount of physical RAM in use. But what is 'memory', then? :confused:
 
Hm, what is the difference between 'real memory' and 'memory'? I assume that 'real memory' is the amount of memory allocated by the process and 'memory' is the amount physical RAM currently mapped to it?

Edit: no, it appears that 'real memory' is actually the amount of physical RAM in use. But what is 'memory', then? :confused:

I really have no idea what the memory column is supposed to indicate but here they are side by side:

LpqiJZF.png
 
I really have no idea what the memory column is supposed to indicate but here they are side by side:

Off-topic, but W0lf can you check to see if all the Apple processes are 64bit now? In 10.8 Safari typically loads something called QTKit Safari Web Content as a 32 bit process in Activity Monitor, has that changed?
 
Off-topic, but W0lf can you check to see if all the Apple processes are 64bit now? In 10.8 Safari typically loads something called QTKit Safari Web Content as a 32 bit process in Activity Monitor, has that changed?

QTKitServer-(1463) Safari Web Content is still 32bit. Appears as though everything else is 64bit though.

Here is all the 32bit process running (pretty much all Chrome):
 

Attachments

  • Region capture 2.png
    Region capture 2.png
    223 KB · Views: 947
With 16GB I don't feel like there is any difference at all.

If you weren't running out of memory before, you won't be running out now. The changes are really there to benefit people who are using all their ram. For example the memory compression feature is not really going to be noticed if you weren't using swap before, it's not going to make your existing ram faster.
 
With 16GB I don't feel like there is any difference at all.

If you weren't running out of memory before, you won't be running out now. The changes are really there to benefit people who are using all their ram. For example the memory compression feature is not really going to be noticed if you weren't using swap before, it's not going to make your existing ram faster.


I have 16GB on my 2011 iMac and running 10.7.5. I constantly run into page-ins due to OS X and other apps cacheing things and not releasing it. From a fresh boot and a couple hours of normal usage, I'll see more than half of my RAM eaten up and a lot of it stuck as "inactive" RAM and after a day or two, it will get to the point of paging-in. Better memory management in Mavericks will be a more than welcome addition to the inner workings of OS X.
 
Not being off topic, but I see the lack of color for the activity monitor (the bottom quadrant) is a bit surprising.
 
upgrade or not upgrade?

so.. i want to buy a MBA 13" with i5, and i would like to run lightroom and photoshop (casual use, nothing harcore), is it ok if i stay with the 4gb of ram right now with Mountain Lion?... i mean.. once that Maverick goes out it should be just fine those 4 gigs of ram on my MBA, right?

thanks
 
DP1 wasn't much better than 10.8.4 but DP2 made a significant stride in reducing the possibility of page outs (swap used in Activity Monitor).
 
My memory situation seems so much different... how can it be?
I'm on a retina MBP 15...
 

Attachments

  • Screen Shot 2013-06-28 at 11.05.53.png
    Screen Shot 2013-06-28 at 11.05.53.png
    484.3 KB · Views: 1,247
I have it running on a 2008 iMac C2D 2.8 with 4GB Ram and a 256GB SSD and it's running fine. No lag whatsoever. I'm sure the SSD plays a big part in that but I'm amazed how well it works. I will be upgrading this machine to 10.9 when it's out no question.
 
I missed the pie charts on all the Tabs on Activity Monitor...

Can't see Wired memory
Can't see Active memory
Can't see Inactive memory
Can't see Free memory

The only pie chart I can find is the one on the Dock when Activity Monitor is active, but it looks dead. It stays at the same percentile whether or not I use RAM-hungry apps...

Apple had just increased my level of noob-ness on OS X... :mad:
 
How does your Safari take so less memory? I only have LastPass as my extension enabled and I have a very high memory usage.


Here's what I have going on with a mid-2009 13" MBP. When I took this screenshot I was running TextWrangler, Mail, iTunes, Safari open with 6 tabs, and Quicktime X playing a 1080p movie.
 

Attachments

  • Screen Shot 2013-06-28 at 10.38.15.png
    Screen Shot 2013-06-28 at 10.38.15.png
    264.5 KB · Views: 672
How does your Safari take so less memory? I only have LastPass as my extension enabled and I have a very high memory usage.

No idea. I don't really do anything "special". The only extension I use is Adblock, and I don't have flash installed. If I need to view flash content(more rare than you would think) I jump over to Google Chrome.
 
That's only because you aren't being nerd enough to use "top" in Terminal.

top no longer reports out statistics on the various page-lists (Active, Inactive, Wired) in 10.9, just Memory Used and Memory Free.

To find this information, you either use vm_stat or the new memory_pressure tool.

To be honest, the way memory accounting and demand paging has been reworked in 10.9 means that keeping close table on the inactive list really isn't as meaningful as it once was. As long as you aren't using any swap, you're fine even if it looks like you have no "free" memory. 10.9 seems to have a higher affinity towards stealing pages that are backed by the v_node pager (read: memory mapped files) now; you can see this by simulating high memory pressure on the system with the memory_pressure tool and watching the size of the "File Cache" drop down to zero before App Memory is impacted and the swap file grows.
 
To be honest, the way memory accounting and demand paging has been reworked in 10.9 means that keeping close table on the inactive list really isn't as meaningful as it once was. As long as you aren't using any swap, you're fine even if it looks like you have no "free" memory.

IMHO, that's always been true, at least since IBM invented virtual memory ~50 years ago -- swapping is what's bad. "Free" memory isn't particularly good either because that's memory that's going to waste. And while "inactive" is better than free, it still can be considered memory thats going to waste. Seeing how poorly people have been judging their needs (or lack of them) for adding memory, as seen on many threads here in the forums, I see merit in throwing out these old misinterpreted metrics for memory "pressure".
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.