1 GB of RAM has no effect on T41p performance

[Jonathan Cordery]

I have just installed 1 GB of Kingston RAM for my T41p 2373 GGG. The part number is KTM-TP9828/1G which is equivalent to IBM p/n 31P9834. I'm pretty sure it is installed correctly.

The question is should I have to do anything else after installation for it to work? i.e. do I need to activate it in some way? I've noticed that it doesn't appear in the Device Manager list.

The T41p originally had 512 mb RAM and now it has 3 times that so i should have noticed an improved performance. Word documents, for example, open no more quickly than before and graphics intensive software is no quicker either.

Have I been conned?

[christopher_wolf]

Well, look at the memory usage and then see how it responds per application opened. I think a 3x improvement from the get go is a little much and can't quite understand why you would think that as it is neither a direct linear relationship nor consistent across all application loads as they are rather heterogeneous and require different things. Performance depends on many things, but it isn't always so that you get 3x better performance if you get 3x the memory. If the stick works, and if you have been running fine with the Thinkpad knowing it has 1.5GB, then the stick is operational and shouldn't have any errors. You can run a memtest on it if you like though.

[pianowizard]

The T41p originally had 512 mb RAM and now it has 3 times that so i should have noticed an improved performance. Word documents, for example, open no more quickly than before and graphics intensive software is no quicker either.

Have you tried opening a HUGE Word document? You should notice a difference. Also, if you are going to upgrade to Windows Vista in the future, you will notice a difference between 1.5GB and 0.5GB even for doing relatively simple tasks.

[egibbs]

Defrag lately?

That usually makes a bigger difference than anything. Doesn't matter how much RAM you have if the document you are trying to open is fragmented into 300 pieces scattered all over the drive. For that matter, when opening anything the bottlneck is getting the data off the disk, not the amount of RAM you have. A faster drive with bigger cache will do more than more RAM for that type of operation.

Depending on the graphics software and what it is doing your video RAM (which you can't change easily on a laptop) may make more of a difference than system RAM for graphics.

What should feel faster is running multiple programs side by side or working with big documents (once they are open). There will be less paging off to disk, which should result in fewer delays when switching between open applications.

Ed Gibbs

[leoblob]

Unless you were running out of physical memory before (check in task manager) then you may feel no difference. With WIN2K, I have played with 128MB, 256, 512, and 1024 (on my desktop). From 128 to 256 made a big difference. 256 to 512 a little. 512 to 1024, none. Just one guy's experience. I typically have about 160MB in use, so once I got up to 256, more memory only made a difference if I was doing stuff like using Photoshop to edit a 300MB file.

[Initial D]

I've noticed that it doesn't appear in the Device Manager list.

It would actually show up under the General (tab) of your System Properties. (Control Panel -> System... 1.5GB should be listed under Computer: )

[leoblob]

One day after I make the above post, I prove myself wrong. I went onto the NOAA site (weather), and I've never bothered downloading the weather loops (on dial-up - ugh), but now on broadband, why not? I downloaded the "high rez" full USA looping map, and once it was downloaded and running, task manager reported I was using 450 megs. Quite surprised!

[Jonathan Cordery]

Thankyou for your replies. That's certainly cleared things up. I was under the impression that extra RAM would actually allow me to open files up more quickly and now I know it is not the case.

InitialD, yes I checked system properties general tab and it does indeed say 1.50 Gb so the RAM has been correctly installed.

Once again, gracias.

[Dead1nside]

Usually Windows XP can survive with 512MB, but 1GB allows your Firefox to eat up 100MB if it wants to, as you're viewing lots of flash etc. it's a lot better for multi tasking.

In Games and such demanding things you will see an improvement, even these days 1GB should be standard.

[agarza]

Hi Jonathan, do you live in Mexico because of work or it's your residence?

Yes, having RAM is one thing and having a faster defraged HDD is another.

If you buff up a T30 with 1GB of RAM but with a 4200rpm 2MB buffer HDD like I did, running a new game will not impress you because the bottleneck is the HDD. There's even a major difference having a 5400rpm vs 7200 drive.

[Dead1nside]

Indeed the bottlenecks of systems these days is the long term storage, the hard drive. This is why Flash HDDs are being researched, as they have superior access times.

The WD Raptor series has found it's place, being a 10K RPM hard drive it's useful if you really want that extra performance. Unfortunatley these drives do not exist in the laptop space.

The best way to maintain a fast system is minimalism. Don't fill up your hard drive, defrag regularly. Kill unnecessary programs from Add/Remove Programs list, use a registry cleaner, anti-virus and malware up to date and scanning. Keep your startup clean (Run--> msconfig--> [Startup]--> uncheck things like Quicktime).

A lot of people complain about having to reinstall XP, every 6 months. If that's the case you're not managing the system well enough. If you don't want the defrag chores, move to Linux.

[tomh009]

A lot of people complain about having to reinstall XP, every 6 months. If that's the case you're not managing the system well enough. If you don't want the defrag chores, move to Linux.

Note that no filesystem -- whether NTFS, FAT, FFS, ReiserFS, JFS or any other one you can think of, on Windows, Mac OS, UNIX or Linux -- is immune to fragmentation. If you do enough writing (and especially creating and deleting files) to a disk, you will get fragmentation.

One big variable is the applications you run on each OS. The second is the set of tools for defragmentation, and whether those are visual or automated.

[christopher_wolf]

Exactly; there seems to be a preconception that moving to an S like Linux frees you from the responsibilities of proper maintenance and up-keep of your system such as defragging, being responsible about what files you put on, and the like. Which couldn't be farther from the truth.

[Dead1nside]

No it is just not as pinnacle and prevalent as it is on a Windows XP box.

[Jonathan Cordery]

It seems I'm no longer getting the email reminders letting me know that people have replied to my initial query hence the intermittent responses from me.

benottomex: I both live an work here in Mexico and I'm really enjoying it.

I got out of the habit of defragmenting the hard disc as usually it says that it is not required. Having said that I have just done one and it was definitely required so I will now do them more often again. I've also just done a 'regclean' so perhaps these two things will help speed things along a bit.

Vamos a ver.

[pjm99au]

For what its worth, I read a magazine article a few months ago where they performance benchmarked different operating systems with different levels of RAM and found that performance under XP ramped up progressively up to 1gb of RAM but then actually dropped with greater than 1gb of RAM installed. Can't remember the explanation they gave as to why.

[smvp6459]

I'd really like to know where you read that over 1GB of RAM in XP causes a decline in performance.

[Dead1nside]

I guess having a lot of RAM can actually start a decline in performance. i.e. depending on what latency it is, and the configuration.

If you achieved 2GB with 4-DIMMs then you'd pay the penalty for it. Ideally you want it in dual-channel. So just two sticks, both of 1GB. However that is for desktops.

[pjm99au]

Regards the article it was in the Australian PC User magazine. As is Murphy's law I had a clean out last weekend and threw my back issues out. From memory the issue was associated with the windows page file and a link between the size of the file and the amount of RAM installed. You might be able to track the article down via the net. The performance decrease wasn't dramatic but certainly droped after 1gb

[davidspalding]

Are you sure they weren't saying that performance increases tapered off with > 1 GB RAM?? I'm highly skeptical that performance regressed with > 1 GB RAM.

Jonathan, you might not see increases in speed opening a Word doc, but with an increase you would see a performance boost doing the following.

• Open Word.
• Open Outlook.
• Open Photoshop.
• Open Firefox. Browse for a while.
• Close all apps.
• Reopen the apps mentioned above.
• Logout.
• Login.
• Open all those apps again.

If you don't notice faster load times on those apps with more RAM on the second and third instances, I'll eat my hat.

I can also assure you that apps like Photoshop, handling files 19 MB or larger, will certainly speed up when you move to 1GB RAM or greater. That So-and-so's Hardware Guide did a test of some apps with single stick 512 MB, single stick 1 GB, and dual-stick 1 GB and found clear performance boosts with certain apps with 1 GB RAM (either configuration). Moving from 1GB to 2GB showed some performance improvement, but not twice as much. Hence, my assertion that overall performance increases taper off, but do not regress.

[pjm99au]

I clearly recall they found a reduction in performance using XP with greater than 1gb of Ram installed. (i.e. the system performed better with 1gb than with more than 1gb). However, I agree more memory intensive tasks can only benefit from having more memory. Without the article anymore I can't provide the specifics but under "standard" operations the problem as I recall it is that windows sets the page file as a multiple of the amount of Ram installed and the page file gets inefficent at too large a size

[KristianJ]

Regards the article it was in the Australian PC User magazine. As is Murphy's law I had a clean out last weekend and threw my back issues out. From memory the issue was associated with the windows page file and a link between the size of the file and the amount of RAM installed. You might be able to track the article down via the net. The performance decrease wasn't dramatic but certainly droped after 1gb

What issue was it? I might have a PDF of it from one of the earlier free CDs/DVDs where they had a heap of back issues on the disc, or I could have the mag itself. I'm fairly skeptical as well, because I recall an article saying the opposite - it said that performance for either 98 or ME decreased for anything over 512MB. Plus I think I've read that XP thrives on greater than 1GB...

[pjm99au]

It's in the Australian PC User magazine March 2006 issue - page 35. (Your post reminded me that I have the back issues in pdf. )The pdf is copy protected but it says in part

"The most interesting aspect of our tests was the speed drop we saw when jumping from 1gb to 2gb - yes our test system actually got slower. We think the reason is the extra addressing time required"

[davidspalding]

I'd like to read the whole article, as how they tested this is pretty critical to understanding their results. If you can mail it to me (davidbspalding at yahoo), I'd like to read it. You have my word I won't redistribute the PDF.

Meantime, I think I'll shrink my pagefile(s) down to 1 GB, and see what that does. I can appreciated that on a slower hard drive accessing a 2GB pagefile could cause a delay. Likewise, if the pagefile is fragmented (which it ought not to be).

[KristianJ]

Biggish file (16MB) - I'll try and see if I can get it through to you tomorrow when I can get to my uni and take advantage of their much, much faster upload speeds than here at home.

[smvp6459]

If you wanted to share it someplace like here: http://rapidshare.de/ I''d love to see it as well.

[davidspalding]

The copyright holder will probably NOT like their PDF shared on the Web without permission. Hence, my suggestion to e-mail it.

Kristian, if the file is rejected by yahoo, you can send it to my domain, contact.2004 (at) korova (dot) com. I believe my own server is set to accept all attachment sizes.

[christopher_wolf]

One thing to note about copyright is that there is a fair use clause such that letting another person read it for non-commercial purposes and simple sharing of knowledge isn't a problem.

[KristianJ]

If you wanted to share it someplace like here: http://rapidshare.de/ I''d love to see it as well.

I was planning to upload it via Yousendit, but not post that download link on this thread. I'll email the magazine and ask them what their response would be if I was to post the link as opposed to privately emailing it. In the meantime, I can PM members with a YSI link.

[davidspalding]

Thanks, Kristian, found the article on page 34.

Well, their article did find that moving from 1GB to 2GB increased a couple of the tests by up to 9 seconds (out of 4 1/2 minutes). I'm skeptical of their test as opposed to the ones at Tom's Hardware Guide (do I have the name right?) in which they tested single- and dual-channel setups with real world apps like Photoshop (running an intensive script), Office, WinZip, Acrobat, Office, among a dozen or so apps. (Tom's also tested with two or more applications chewing data simultaneously.)

This mag uses their own benchmarking app, which is consistent, but may not reflect real world applications.

... Our new UserBench Video 2006 test uses VirtualDub and TMPGEnc Plus, both comparatively small applications in terms of memory requirements. [emphasis added] As you can see from the results in our table below, we saw only a small increase of less than 7% by jumping from 256MB to 1GB. But if you're starting with just 128MB, the gains are 17% for MPEG-2 encoding and 24% for DivX encoding. More important here is the processor and hard disk subsystem.

I have to point out that other, real world application might show a significant increase in performance moving up from 256 MB. (Example: running a complex series of filters on 10 large images all open in the application.) An app that shows marked increases from 128MB to 256MB is not one that I would expect to benefit from changing from 1GB to 2GB. Here's how their video tests played out: 128MB => 5:17 | 256MB => 4:40 | 512MB => 4:39 | 1GB => 4:22 | 2GB => 4:28.

They did not explain their "image" test, only posted the numbers. I have to point out though, that it is clearly documented that when Photoshop has several large files open, and has to start paging out to the scratch disk, performance for some operations slows. Considering that my Camera RAW images run to 19MB each, opening a series of 10 images and manipulating them can eat up RAM considerably.

Here's how their image tests played out: 128MB => DNF | 256MB => 5:10| 512MB => 4:02 | 1GB => 3:39 | 2GB => 3:48.


The results of their VeriTest Business Winstone 2002 showed 47.2 (minutes, or seconds, they don't say) with 1GB, and 48.2 for 2GB. I would agree that 2GB is excessive for running Office 2003. It is NOT excessive for running Visual Studio 2006, or Virtual Server 2005 R2. In fact, if you're running virtuals, 4GB (if supported) is recommended by moi.

Now to the meat of their assertion. In a sidebar labeled Don't Buy Too Much, they state:

The most interesting aspect from our tests was the speed drop [sic] we saw when jumping from 1GB to 2GB of memory -- yes, our test system actually got slower. We think the reason is the extra addressing time required. There also aren't any applications that need 2GB of memory, so in our view it really is throwing around $140 down the drain.

Well, I have to counter this with, You used a benchmarking application, using components that you admit are "small," not something that is specifically known to benefit from more RAM, like Photoshop. And "test system ... got slower" is not the same as the image test which took 3:39 with 1GB DDR400 and 3:48 with 2GB DDR400. That was a specific test, and not the same as the whole system slowing down. I find their phrasing slightly misleading.

Further, they didn't investigate, e.g. running the 2GB test with smaller virtual memory, no virtual memory, etc. The author(s) just said, "We think the reason is...." I don't find that credible beyond their specific tests.

My own two cents, if you're only doing office apps and web surfing, 1GB is probably fine. If you're doing work with large applications like Visual Studio or Photoshop where the consensus is "load as much RAM as you can," then find a good price and lock 'n load.