Personal computing discussed

Moderators: renee, mac_h8r1, Nemesis

 
Mr. Camel
Gerbil
Topic Author
Posts: 31
Joined: Wed Nov 16, 2011 12:08 pm

Slow System Performance

Thu Dec 22, 2011 11:24 am

Hi,

I am having performance issues with my Dell Latitude E6520 laptop. I believe that I have identified the performance bottleneck to be the Seagate ST320LT014-9YK142 HDD. However I have been unable to resolve the problem.

Here are the additional specifications of the laptop:

CPU: Core i7 2620M (2C/4T)
Memory: 4GB DDR 1333
GPU: Integrated GPU + nVIDIA NVS 4200M
OS: Windows 7 Enterprise 32-bit

I believe that the HDD is the issue because whenever I am running multiple processes which access the disk simultaneously, the paging activity goes to 100% (I found this by monitoring the LogicalDisk:% Disk Read Time event using Perfmon) and when this happens, the whole system slows down to a snail's pace. I find this behavior somewhat puzzling because when I look at the memory utilization in Task Manager, it only shows 1.7 GB as being used and I still have almost 1 GB of free memory. In other words, why should there be so much paging when I have so much free memory?

I ensured that the HDD is accessed using UDMA 6 and not PIO.

I have Windows 7 set to automatically manage the paging file size for all drives - it has set the page file size to be 2984 MB.

Can anyone help?

Thanks.
 
DancinJack
Maximum Gerbil
Posts: 4494
Joined: Sat Nov 25, 2006 3:21 pm
Location: Kansas

Re: Slow System Performance

Thu Dec 22, 2011 1:11 pm

Seems odd to me especially considering the HDD is 7200rpm. Maybe a x64 OS might have some effect?
i7 6700K - Z170 - 16GiB DDR4 - GTX 1080 - 512GB SSD - 256GB SSD - 500GB SSD - 3TB HDD- 27" IPS G-sync - Win10 Pro x64 - Ubuntu/Mint x64 :: 2015 13" rMBP Sierra :: Canon EOS 80D/Sony RX100
 
mmp121
Gerbil
Posts: 31
Joined: Tue Aug 15, 2006 12:09 am

Re: Slow System Performance

Thu Dec 22, 2011 2:31 pm

I presume you have checked the SMART codes of the HDD to make sure all is in order with it. I would start there, if you truly suspect the HDD.
 
JdL
Graphmaster Gerbil
Posts: 1025
Joined: Thu Aug 29, 2002 11:45 am
Location: United States of America

Re: Slow System Performance

Thu Dec 22, 2011 2:32 pm

What applications are you running when it's going slow?
JdL
286DX > Pentium Pro 200 > Athlon 1000 @ 1.4 GHz > Athlon 64 2.2 GHz > Athlon XP 3200+ > Core 2 Duo 3300 @ 4.0 GHz > Core 2 Quad 6600 G0 @ 3.2 > Core i7 3770 @ stock
 
Flatland_Spider
Graphmaster Gerbil
Posts: 1324
Joined: Mon Sep 13, 2004 8:33 pm

Re: Slow System Performance

Fri Dec 23, 2011 9:24 am

Try setting a static page file size. If Windows is managing the page file, it won't allocate the space until it needs to swap pages out of memory, and that will cause more disk writes then needed.

Sometimes, Windows programs will write to the page file just because they want to. It's an odd quirk of the platform, and that could trigger the problem as well.

Also, I'd look in to upgrading to a 64-bit OS or dropping down to 3GB of RAM with 32-bit Windows. I know they say a 32-bit OS can handle 4GB of RAM, but things get weird when you try it in reality.
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Slow System Performance

Fri Dec 23, 2011 10:56 am

Flatland_Spider wrote:
Sometimes, Windows programs will write to the page file just because they want to. It's an odd quirk of the platform, and that could trigger the problem as well.

I wouldn't even say it is a particularly odd quirk. It makes sense to have a certain amount of RAM available that you can reallocate on a moment's notice based on demand. If the virtual memory manager believes that a certain set of dirty pages aren't going to be used for a while, flushing the contents of those pages out to the page file *might* make sense.

Is the VMM always going to guess correctly? No. But it does have the *potential* to improve overall performance in many scenarios. Later versions of Windows (Vista/7) do these sorts of "optimizations" more aggressively than older ones...

Back on topic... to the OP:

This is not a simple issue. The fact that the activity is classified as "paging" does not necessarily mean that the swap file is being accessed. What is your actual swap file usage?

A not uncommon trick in modern OSes/applications is to use "memory mapped files" for I/O. A segment of virtual memory is "mapped" to the contents of an arbitrary file on disk, and the OS/application then relies on the VMM's page fault logic to fetch the data from disk when that RAM is accessed, instead of issuing explicit file I/O requests. Memory mapped file I/O might show up as paging activity (since the VMM is doing the I/O on the application's behalf), without actually using the swap file.

Bottom line is, the only real solution here may be to swap the HDD for an SSD if you need to multi-task disk intensive applications.
Nostalgia isn't what it used to be.
 
Ryu Connor
Global Moderator
Posts: 4369
Joined: Thu Dec 27, 2001 7:00 pm
Location: Marietta, GA
Contact:

Re: Slow System Performance

Fri Dec 23, 2011 2:27 pm

Flatland_Spider wrote:
Try setting a static page file size. If Windows is managing the page file, it won't allocate the space until it needs to swap pages out of memory, and that will cause more disk writes then needed.


The managed pagefile follows a ratio and has a fixed value until the system begins to run out of paging space. There is little reason to shift to a static swap and good reason not to.

just brew it! wrote:
Flatland_Spider wrote:
Sometimes, Windows programs will write to the page file just because they want to. It's an odd quirk of the platform, and that could trigger the problem as well.

I wouldn't even say it is a particularly odd quirk. It makes sense to have a certain amount of RAM available that you can reallocate on a moment's notice based on demand. If the virtual memory manager believes that a certain set of dirty pages aren't going to be used for a while, flushing the contents of those pages out to the page file *might* make sense.

Is the VMM always going to guess correctly? No. But it does have the *potential* to improve overall performance in many scenarios. Later versions of Windows (Vista/7) do these sorts of "optimizations" more aggressively than older ones...


Linux has some more leeway vs Windows with the paging design through the swappiness value, but one of the kernel developers is apparently a firm believer in an aggressive swap design.

Link

The Linux 2.6 kernel added a new kernel parameter called swappiness to let administrators tweak the way Linux swaps. It is a number from 0 to 100. In essence, higher values lead to more pages being swapped, and lower values lead to more applications being kept in memory, even if they are idle. Kernel maintainer Andrew Morton has said that he runs his desktop machines with a swappiness of 100, stating that "My point is that decreasing the tendency of the kernel to swap stuff out is wrong. You really don't want hundreds of megabytes of BloatyApp's untouched memory floating about in the machine. Get it out on the disk, use the memory for something useful."
All of my written content here on TR does not represent or reflect the views of my employer or any reasonable human being. All content and actions are my own.
 
LaChupacabra
Gerbil First Class
Posts: 145
Joined: Tue Dec 30, 2008 10:59 pm

Re: Slow System Performance

Fri Dec 23, 2011 3:11 pm

That Dell is one of their business class offerings so my first question is are you using this for personal use, or was it issued to you your job? In either case are you running any type of disk encryption? Depending on the program and how it's set up I've seen applications like that cause the types of issues you describe on 6520's. Another thing that seems to happen with them is the copious amount of pre-installed wireless networking software can really slow everything down. The applications can conflict with each other and cause the machine to thrash. Try picking the one you like and removing the rest.

If neither of those are the case can you think of anything else that might be running on your laptop that is out of the ordinary?
 
Flatland_Spider
Graphmaster Gerbil
Posts: 1324
Joined: Mon Sep 13, 2004 8:33 pm

Re: Slow System Performance

Mon Jan 02, 2012 4:15 pm

just brew it! wrote:
I wouldn't even say it is a particularly odd quirk. It makes sense to have a certain amount of RAM available that you can reallocate on a moment's notice based on demand. If the virtual memory manager believes that a certain set of dirty pages aren't going to be used for a while, flushing the contents of those pages out to the page file *might* make sense.

Is the VMM always going to guess correctly? No. But it does have the *potential* to improve overall performance in many scenarios. Later versions of Windows (Vista/7) do these sorts of "optimizations" more aggressively than older ones...


It does make sense to have some RAM available for newly started programs, especially in low RAM situations like desktop computers. I'm not really up on the finer points of Windows virtual memory management; I just try to hit the disk as little as possible. Vista, at least and probably Win7, are much better about paging then older versions; it's less painful then it used to be.

VMMs have always worked well enough for me. I've never thought one was really stupid. I've always considered the VMM to be limited by storage technology and hampered by user expectations.

A not uncommon trick in modern OSes/applications is to use "memory mapped files" for I/O. A segment of virtual memory is "mapped" to the contents of an arbitrary file on disk, and the OS/application then relies on the VMM's page fault logic to fetch the data from disk when that RAM is accessed, instead of issuing explicit file I/O requests. Memory mapped file I/O might show up as paging activity (since the VMM is doing the I/O on the application's behalf), without actually using the swap file.


This would explain some of the behavior I've classified as odd, and no one could really explain. Like writing 2-3MB of data to the swap file when opening a text editor, or programs requiring a swap file before they will start.
 
Flatland_Spider
Graphmaster Gerbil
Posts: 1324
Joined: Mon Sep 13, 2004 8:33 pm

Re: Slow System Performance

Mon Jan 02, 2012 5:54 pm

Ryu Connor wrote:
The managed pagefile follows a ratio and has a fixed value until the system begins to run out of paging space. There is little reason to shift to a static swap and good reason not to.


There is a very good reason. Systems slow to a crawl for a long period of time when Windows increases the swap file on the fly, and a static swap files ensures a user will hit a hard limit, eventually. If someone is running out RAM and swap, the hardware and software requirements need to be re-evaluated. The automatic expansion could masking the problem, and the user could be trying to be nice and continue to live with the poor performance. Basically, the system to complain, and in turn the user, rather then mask the problem.

It also makes monitoring easier if the swap space is static. 25% of a known 6GB is more meaningful then 25% of ?. (?=whatever the OS felt like setting the swap file to at the moment)

Vista started setting the swap file to a much more reasonable number then past versions, and that has negated this piece of advice somewhat. In this scenario, it's a troubleshooting measure. If the system is swapping heavily and the OS is masking the problem, then causing it to run out of swap space will narrow down the problem. If the system doesn't fail, then something is hogging the disk IO.

Linux has some more leeway vs Windows with the paging design through the swappiness value, but one of the kernel developers is apparently a firm believer in an aggressive swap design.

Link

The Linux 2.6 kernel added a new kernel parameter called swappiness to let administrators tweak the way Linux swaps. It is a number from 0 to 100. In essence, higher values lead to more pages being swapped, and lower values lead to more applications being kept in memory, even if they are idle. Kernel maintainer Andrew Morton has said that he runs his desktop machines with a swappiness of 100, stating that "My point is that decreasing the tendency of the kernel to swap stuff out is wrong. You really don't want hundreds of megabytes of BloatyApp's untouched memory floating about in the machine. Get it out on the disk, use the memory for something useful."


It really depends on what you're trying to do. Setting the VMM to swap out aggressively will provide more usable RAM at the expense of responsiveness. If the user or computer is going to be doing one thing for a while, or be doing one resource intensive thing, it would be a good idea to setup an aggressive swap setting. Responsiveness requires less lag, so setting a less aggressive swap is better. Andrew is probably compiling lots of code and using an IDE the most, so he is a good candidate for a very aggressive swap setup.

I'm personally think a system should be as responsive as possible. RAM should be leaned on hard since it's going to be the most responsive storage medium, and I want to switch quickly between multiple applications. As such, I'm going to favor a less aggressive swap setting like 20 - 30.

Who is online

Users browsing this forum: No registered users and 1 guest
GZIP: On