Are we sure it's a bad thing when RAM usage goes up?
 

I'm reading lots of people trying to figure out how to keep RAM usage down. What I'm wondering is if that's really productive. Seems to me that as long as the OS properly manages it, it should be OK if RAM usage get into the 90% range.

If the OS is using RAM to cache commonly used chunks of code, and if it has a decent routine for freeing up RAM as needed, what's the problem. It could actually slow things down to keep purging RAM.

I'll admit I don't know much about what's going on in the kernel or exactly why RAM usage grows over time. Is there anyone who has dug deep enough into the system to know for sure what's going on in there?

Re: Are we sure it's a bad thing when RAM usage goes up?
 

my RAM doesn't go down automatically. what do you mean by freeing up RAM? i have to use a seperate software to do that periodically (cleanRAM)

Re: Are we sure it's a bad thing when RAM usage goes up?
 

I think where you might start to notice problems is when you are at 90% usage and then you try to load something like opera. It will start spitting out memory errors kind of like you were on a mogul trying to run it. Another reason to keep memory low is if you play any games on the tp2. Games need all of the memory that they can get to run smoothly. That probably isn't the specific answer that you want but its the first few reasons that i can think of to keep the memory clean.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

Sure it does, you just don't let it get low enough.

WM6.x has RAM management built in: when it reaches a certain threshold, it shuts down programs based on a priority system--so things like Outlook are high priority and rarely will be shut down, over say a dialog screen or Settings.

This threshold value and memory level are changeable I believe via the registry.

To the OP: I agree. People put way, way, way too much emphasis on "free RAM".

Fact is, a system with 15mb of free RAM will perform the same as a device with 100mb of free RAM, makes no difference. I've run the Treo Pro with 38mb of free RAM on boot and WM6.5--fast as ever and you don't notice it. Sure, if you run Opera Mobile, Skype, TwitToday, etc. more RAM is better. But even then, the system itself does not perform any faster.

Confession: I don't manage my memory. I don't use "free ram" tools and never look at how much I have. This is non-sense and a waste of users time, imo. Especially with WM6.5 which has more VM and better memory management.

Now on custom ROMs, sometimes it is nice to know, only because you want to know what is running in the background--after all, background programs and processes take up RAM and CPU cycles. But if it is due to physical RAM or what the OS allots and not background apps, then the actual amount is trivial to the user experience.

That's why I jam up my glyphcache and system cache sizes--they eat RAM but make the OS zippier. I can care less about RAM.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

He means that the memory manager will properly expire/dirty unused pages and allow them to be swapped out when those pages are needed.

Proper caching (and even precaching) can be quite helpful in speeding up application loading/performance, but you do need a memory manager that will allow the appropriate amount of memory to be freed when needed, especially for applications with large footprints. If the memory manager doesn't allow this precached memory to be freed when not used or needed or doesn't free enough you'll see OOM errors.

This appears to be what HTC wanted to do with ResProxy - allow the precaching/caching or resources - but people reported that they were getting OOM errors when the memory usage climbed too high.



Well it will assuming you don't load something that needs > 15 MB of RAM, but the point is 100% correct. Assuming enough memory is free to load the application in question having N free or 10N free makes no difference from a speed perspective. Only when the memory has to become actively managed (pages dirtied and swapped to disk or swapped out and replaced) would you see a slow down.



100% agree here - I'd rather have a snappier performance for day to day use - I also don't run out of memory. I don't know what everyone is loading but it's clearly a lot more than I use - from the above post it appears "games" is the answer. Euchre isn't very taxing on the old memory it seems :)

Re: Are we sure it's a bad thing when RAM usage goes up?
 

I have yet to have my TP2 lock up in 6.5. I do reset it out of old habits, but most of the time its because of uninstalling or installing apps that require resets. I notice with 85% usage, the phone is actually snappier than it is originally at bootup

edit: one of the guys from the apple store wanted to see my phone (probably trying to spot an error so he can knock WM). ram usage was at 88% and the phone was moving along very smoothly with everything ready to go. after a few minutes of casual talk, and him playing with it, he handed it back to me with a mildly frustrated expression on his face :D

Re: Are we sure it's a bad thing when RAM usage goes up?
 

Great discussion, this is exactly what I was thinking. I'd rather be properly using 95% of the memory and have things loaded and ready to go than keep it sitting at 39% and have to constantly reload things every time I open something. Sometimes it seems like it's a competition to get the lowest RAM number.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

This is correct.

This is why I run RAM Sweeper periodically throughout the day. And also use Task Manager to close programs that are not being used or won't use for awhile. This helps improve battery life to have programs and tasks completely closed and out of memory...

It's an old habit since I'm a PC Gamer and always look for performance overall.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

I've never been convinced of this argument that programs sitting in memory, but not consuming CPU cycles, deplete battery life significantly or even at all.

Anyone care to give a detailed argument from a programming/engineering perspective to actually back this up? Because I think it's crap. Seems to be an old habit carried over from the WM2003 days with little evidence. I also hope for the sake of the OS it is not true because that is just awful to make the consumer manger their devices---the exact opposite reasoning give by Microsoft for having apps minimize and not close.
--Windows Mobile Team Blog

It's a sad state that WinMo users feel they have to waste time managing their memory throughout the day. This is the opposite of "smartphone" and I think it is misguided.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

Well, if the app is polling on any sort of interval for hardware status or other info then it will use resources (the amount of resources used will obviously depend on the hardware being queried). If it is truly sitting idle and only operating on its own data set or info then it shouldn't be using any additional resources by merely being open.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

Exactly and any programmer worth their salt would not have their software polling unless it actually needed to do so; off hand I can't think of many apps sans email, IM and automatically-updating data programs that would need to do this.

But browsers, file explorer, RSS readers that aren't updating, Twitter apps that aren't auto-updating, Outlook, Netflix, etc. don't do this as far as I know. In fact it should seem obvious which apps do truly "run" in the background and those that don't do anything.

Edit: Here's a good example direct from MS on coding of Widgets, battery life and what they strongly suggest to preserve power; this is why I think modern programs on a moder OS don't impact battery life when idle.

• Only consume network resources when your widget is active.
• Minimize animations when the widget is idle.
• Cache data locally as aggressively as possible, use the widget persistent storage as a repository since it will be preserved across runs (and even upgrades).
• Be aware of the current battery level and minimize network operations when the battery is running low.
• Always (and by that I really mean always!!) use asynchronous network requests.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

malatesta you are on point with the technical info. u back up your opinions with actual facts i love that!! thanks for the info man!

Re: Are we sure it's a bad thing when RAM usage goes up?
 

lol, it's my academic background ;-)

Heck, I even enjoy when folks prove me wrong or teach me something I don't know (which is often), so if people can enlighten me on these issues I'm all for it!

Re: Are we sure it's a bad thing when RAM usage goes up?
 

Your giving too much credence to MS and some random developers personally. MS operating systems are constantly running tasks even on an idle machine. Once these programs aren't in the que they won't have the ability to run. So I'm glad you think we are wasting our time and whatnot. I know what works in my own expierence since i'm the one doing it. Justify the means and go have a coke and a smile.....

Re: Are we sure it's a bad thing when RAM usage goes up?
 

What processes and tasks? I'm looking for specifics not generalities. And giving too much credence to the developers of the operating system as opposed to joe-schmo in a forum with zero credentials (no offense of course, even I can be lumped into that category)? Hmmm...

I just see lots of claims being made but little actual arguments to support them. I've seen these sort of urban myths passed on in forums and I just see them as inaccurate, hence why I'm looking for someone to explain it.

This reminds me of the "turn off beam" to save battery life meme.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

now thats a good attitude that we should all go by...... You da man Malatesta.....

Re: Are we sure it's a bad thing when RAM usage goes up?
 

all i know is with my verizon tp i couldn't multi task at all until i flashed an ssk rom that allowed lower ram usage. i couldn't open multiple tabs in opera and it was extrememly slow. i have no problems with my phone but don't see why i would want to let it get that high again.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

whenever I see more ram used it also means more processor used.. and I wish we had somethign showing how much processor is being used like on windows vista/7

Re: Are we sure it's a bad thing when RAM usage goes up?
 

Get an app like MemMaid to show CPU cycles.

There's no direct correlation between amount of RAM available and CPU usage. Just because RAM is occupied doesn't mean the CPU is doing any work, there's just no direct connection.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

It's definitely NOT a good thing.

WinMo leaks memory like a colander leaks water. I soft reset every day, and run a memory cleaning app all the time, otherwise I don't have enough memory to run memory hungry programs like iGO8.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

why not? For ram to be occupied it means the CPU has to be doing somethign ram just doesnt get used for no reason.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

Well, for the data set to get placed *into* ram the CPU must be used. However, it can remain in ram and if nothing is working with the data set then no CPU operations are required to keep it in RAM.

RAM gets used by any application in non sequential fashion. It is entirely possible that data in RAM is out dated or was placed there by a program that has since been closed or is sitting idle (which is why programs reading from an incorrect memory address exhibit sometimes unpredictable results).

However, having that data in ram doesn't require any further cpu operations.

What you may see is that extra work is required validating that the current state of the data in memory is no longer needed so that it is then dirtied/swapped out to allow for newer data to be placed in the same memory location. This occurs when there is not enough free memory (the limit may be less than 100%) to place the required data set into memory or garbage collection occurs. This threshold is going to be determined by the memory manager (and 6.1's is really not very good).

So, with a large amount of used memory you may see extra CPU cycles used when new data has to be written into memory locations that are already occupied by old/stale data, but the actual storage of the current data (assuming it's not being used and nothing is being written to those locations) does not in itself cause any extra CPU usage.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

To add to krohnjw's excellent explanation, there is a whole other area of cache and allocation of that--in short I devote a lot of my RAM to cache (file system, file system filter, GDI/font, etc),which is variable to a certain extent. In turn I have a "faster" device but consume more RAM; however just because certain areas of the OS are pre-cached doesn't mean it is being acted on by the CPU when not being called up for a process.

Remember, RAM is just volatile memory--it's a temporary storage area for data to be accessed quickly but not written to storage. The RAM chips themselves, AFAIK, are powered by the system regardless of what is being "held" in memory at the time and it is only when that information is called or acted upon does the CPU need to be used.

At least that, in my non-engineering background, is how I understand it.

Re: Are we sure it's a bad thing when RAM usage goes up?
 

Is there a specific way in which one goes about this?