Adding 16GB RAM - Measuring performance increase?

[Mike Boreham]

Just received delivery of my 16gb ram upgrade for my 2012 MacBook Pro. Before I install it I'd like to take some readings of its current performance to compare with the new ram.

What do I look for? Is it page outs?

This is going back a bit now, but when I bought my 2010 MBP with 4Gb of RAM, I was very disappointed by the Photoshop performance measured by the Retouch Artists Speed Test. Amongst other things this test uses quite a lot of RAM. (That looks like an old thread but if you click on "last" you will see that it is still very much alive, must be the longest running active thread on MacRumours.)

I upgraded from 4 to 8Gb and the improvement was dramatic.

Lots of good advice in this thread. the only thing I would add is that if you want to use Parallels or VMware 16 Gb is well justified.

 

[Big Stevie]

Yeah dude, you're close to maxing out. Even though your system is using up some of that ram just to keep the memory from partially closed programs. You would likely benefit from having that 5th or 6th gig of ram during your heavy usage. If you've already bought it, and are trying to justify using it. I would say this is the perfect situation lol. I would do that rather than trying to "use" your computer conservatively just so that you don't worry about running out of ram.

Great, thanks. Im not so much trying to justify the purchase, more like wanting to monitor any improvements. I didnt know what to look for to check for improvements. I get a lot of spinning beach balls during what I would call 'normal use'.

I bought the 16GB as it wasnt much more ££ than the 8GB. I'll pop it in and see what happens

----------

This is going back a bit now, but when I bought my 2010 MBP with 4Gb of RAM, I was very disappointed by the Photoshop performance measured by the Retouch Artists Speed Test. Amongst other things this test uses quite a lot of RAM. (That looks like an old thread but if you click on "last" you will see that it is still very much alive, must be the longest running active thread on MacRumours.)

I upgraded from 4 to 8Gb and the improvement was dramatic.

Lots of good advice in this thread. the only thing I would add is that if you want to use Parallels or VMware 16 Gb is well justified.

Thanks, Ill give that thread a good read

 

[Big Stevie]

About 4 days since my restart and its now looking like this...

So, Page Outs are 4.24GB with 1.77GB swap. I will be fitting the new 16GB when I get back from hols at end of this month, I shall use these posts as a benchmark. Thanks everyone.

 

[scbond]

I would honestly say that 16GB would be of little use to you. If you really want to upgrade then get 8...you could probably be very comfortable with 6 even.

 

[NewishMacGuy]

About 4 days since my restart and its now looking like this...

Image

So, Page Outs are 4.24GB with 1.77GB swap. I will be fitting the new 16GB when I get back from hols at end of this month, I shall use these posts as a benchmark. Thanks everyone.

I would honestly say that 16GB would be of little use to you. If you really want to upgrade then get 8...you could probably be very comfortable with 6 even.

You'll probably be fine with 8GB unless you find yourself running a Vm with moderate usage on both OSX and the VM at the same time, but 4GB of page-outs in 4 days is a lot. You'll definitely notice an improvement from upgrading, particularly if you're still running an HDD.

>

 

[Big Stevie]

Thanks. Im about to go away on my hols, but when I get back I will be inserting the new RAM, so I will post how it goes.

My last check before the new RAM is fitted reveals nearly 5GB of Page Outs in 5 days!

 

[throAU]

Stop looking at activity monitor unless performance is bad. some page outs are normal depending on the amount of active vs inactive stuff you have in RAM.

going into details on how modern memory management operates is kind of beyond a forum post, but the short version is

• inactive = essentially free RAM currently used for cache as there's nothing better to do with it
• page out rate (NOT the total) under load = number to look at

You could put 500 GB of RAM in a machine, and eventually if you were to read/write enough stuff to/from disk, if you had inactive data in memory it would likely eventually get paged out.

 

[GermanyChris]

Stop looking at activity monitor unless performance is bad. some page outs are normal depending on the amount of active vs inactive stuff you have in RAM.

going into details on how modern memory management operates is kind of beyond a forum post, but the short version is

• inactive = essentially free RAM currently used for cache as there's nothing better to do with it
• page out rate (NOT the total) under load = number to look at

You could put 500 GB of RAM in a machine, and eventually if you were to read/write enough stuff to/from disk, if you had inactive data in memory it would likely eventually get paged out.

He's justifying a purchase, let him convince himself.

 

[Big Stevie]

No, im trying to overcome the freezing up & spinning beach balls. I was told that more RAM might help. 16GB was only a few more £ than 8GB so it seemed to make sense. And lots of folk report these issues get solved by increasing RAM.

The purpose of this thread is to find out how I can monitor if the extra RAM is performing. I figured using a stopwatch to monitor any increase in speed wouldnt be the best way.

 

[Bobby Corwen]

Barring the costs, are there any disadvantages to owning a Lamborghini to go grocery shopping down the street?

You'll have it, but it'll never be put to use. Let your wallet decide knowing that.

I highly disagree with this train of thought.

He should keep 16. The price is nominal.

OP there is no way of quantifying it. But it is a factor once in a while anecdotally speaking. And just the fact that you know you have no ceiling in that department is nice in itself and worth the money.

 

[pat park]

RAM is relatively cheap these days. Why is everyone making such a big fuss if he installs 16 GB? Sure he may not really need it, but it sounds like its already been purchased. Install it, use it and enjoy. Go out get a nice dinner and have a few drinks.

 

[throAU]

No, im trying to overcome the freezing up & spinning beach balls. I was told that more RAM might help. 16GB was only a few more £ than 8GB so it seemed to make sense. And lots of folk report these issues get solved by increasing RAM.

The purpose of this thread is to find out how I can monitor if the extra RAM is performing. I figured using a stopwatch to monitor any increase in speed wouldnt be the best way.

Just put the RAM in the box, you'll notice it.

The difference between 4 GB and 8 GB is massive. The difference in my experience between 8 GB and 16 GB in day to day subjective "feel" is about 25% of the difference between 8 and 16 GB - not even doing anything special - because the OS has so much more RAM available for caching. If you're running VMs, etc then the 16 GB advantage grows further, as you have multiple virtual machines getting their disk access cached, and much more memory being consumed by them.

Given the price difference between 8 GB and 16 GB, if your system can handle it, it's a total no brainer to go for 16 now.

 

[talmy]

To put a quantitative value on it, 1GB of paging takes about 10 seconds, so if you are accumulating 1GB of paging a day (as the OP is) that means 10 seconds of lost time per day, or about an hour a year.

 

[bniu]

Holding the cost of the chips aside, is there any disadvantage to having more RAM than you might need?

slightly higher cost, not by much though. Difference is probably in the neighborhood of $50, not something I'm too concerned about. I've got 16GB as well, maybe it's a bit overkill for me, but I'd rather just max out my machine and then not have to think about it again. If 16GB SODIMMs ever come out, are reasonably affordable, and are confirmed to work with my MBP, I'll put two of those in to take my MBP to 32GB of RAM and then proceed to use my computers for several more years and not be constantly trying to figure out if I need to upgrade or not.

 

[Shrink]

slightly higher cost, not by much though. Difference is probably in the neighborhood of $50, not something I'm too concerned about. I've got 16GB as well, maybe it's a bit overkill for me, but I'd rather just max out my machine and then not have to think about it again. If 16GB SODIMMs ever come out, are reasonably affordable, and are confirmed to work with my MBP, I'll put two of those in to take my MBP to 32GB of RAM and then proceed to use my computers for several more years and not be constantly trying to figure out if I need to upgrade or not.

You're right on the mark with cost. Two 4GB chips were $78, two 8GB chips were $125. And i agree that not having to think about Ram, page outs, etc. was worth the $47 to me.

 

[jdsipod]

No, im trying to overcome the freezing up & spinning beach balls. I was told that more RAM might help. 16GB was only a few more £ than 8GB so it seemed to make sense. And lots of folk report these issues get solved by increasing RAM.

The purpose of this thread is to find out how I can monitor if the extra RAM is performing. I figured using a stopwatch to monitor any increase in speed wouldnt be the best way.

My friend and I just upgraded my MBP from 4GB to 16GB, and to see the performance difference he had me shut down my computer with the old RAM in it, then we timed it from when I hit the power button, to the time everything was loaded and logged in ( with automatic login). Then we did it again with the 16GB after we installed it, and the time dropped from 2 minutes to 30 seconds. That is just a easy way to test your performance.

 

[alexrmc92]

My friend and I just upgraded my MBP from 4GB to 16GB, and to see the performance difference he had me shut down my computer with the old RAM in it, then we timed it from when I hit the power button, to the time everything was loaded and logged in ( with automatic login). Then we did it again with the 16GB after we installed it, and the time dropped from 2 minutes to 30 seconds. That is just a easy way to test your performance.

This makes little sense. OS X doesn't use more than 1GB for itself, 4GB could easily handle OS startup and shutdown and 16GB would make no difference.

The only way it would help is if you have a ton of startup apps and apps open on shutdown. Adding more RAM only helps when your machine runs out of it, which will never happen on startup or shutdown, unless your doing what i previously stated.

 

[talmy]

This makes little sense. OS X doesn't use more than 1GB for itself, 4GB could easily handle OS startup and shutdown and 16GB would make no difference.

And my experience has been bootup time actually slows with increased RAM because of the POST of RAM taking longer. Typically disk accessing speeds is the major component of boot times which is why SSDs make such a large difference.

 

[alexrmc92]

And my experience has been bootup time actually slows with increased RAM because of the POST of RAM taking longer. Typically disk accessing speeds is the major component of boot times which is why SSDs make such a large difference.

Very true, i didn't even think about POST times.

 

[cube]

Activity Monitor is not the proper tool for measuring paging activity.

Use 'top' in the Terminal.

 

[talmy]

Activity Monitor is not the proper tool for measuring paging activity.

Use 'top' in the Terminal.

Why would you say that? Activity monitor shows the page ins and outs in units of bytes. On the other hand, top (ironically, at the bottom of the screenshot) gives the same values but in units of 4KB pages. Same information but not consumer friendly.

 

[cube]

Why would you say that? Activity monitor shows the page ins and outs in units of bytes. On the other hand, top (ironically, at the bottom of the screenshot) gives the same values but in units of 4KB pages. Same information but not consumer friendly.

It's not the same information. Activity monitor gives only cumulative values. You can watch instant values in 'top'.

 

[Mike Boreham]

It's not the same information. Activity monitor gives only cumulative values. You can watch instant values in 'top'.

Isn't the bit in brackets on the page-outs line the instant rate (0 bytes/sec on the last screen shot)?

 

[cube]

Isn't the bit in brackets on the page-outs line the instant rate (0 bytes/sec on the last screen shot)?

top shows the number of pages in the sampling period, not MB/s.

 

[Mike Boreham]

top shows the number of pages in the sampling period, not MB/s.

But it is the instant rate, and more accessible for most people.

I agree that the instant rate is theoretically more important, but who has the time to watch it all day and know when/what is causing page outs? Mostly when I look at instant rate it is zero.

If you restart the computer (reset page outs)...use it....and next day have a few Gb of page outs, then you need more RAM, though you may not know what app or usage was causing it.