I have an ASUS Eee PC Seashell 1005HA-PU17-BK, 250 GB HDD.
I'm going to put 3 different linux distros on it, like this:
1) 2 GB - xPUD (very fast booting OS, is pretty much just firefox and a music player)
2) 4 GB - Puppy Linux (fast and cute, an actual full OS)'
3) 12 GB - *buntu (for when I need something bigger than the 100 MB puppy linux. I'm not sure whether i should use Ubuntu NBR or Eeebuntu NBR or Xubuntu or Eeebuntu LXDE edition...)
4) 40 GB - Windows 7 Starter (for running .exe's and games and testing some programs I'm currently developing for windows--going to branch out into linux, though. It works pretty good in regular ubuntu right now but the background doesn't load and it can't save or load...)
5) ? GB - Windows 7 recover (the built in recovery thing for windows since the netbook doesn't have a DVD drive for a regular recovery disk. I don't know how big Asus's are, or if they even had one. My HP netbook had an 11 GB one, but i returned that netbook and am getting an Eee PC now, and i don't know what Asus does)

The 6th partition is just a Fat 32 for storing all of my files and stuff for sharing between all OS's. It'll be 150+ GB, whatever is left.
The 7th will be a 2 GB swap partition.

I'm testing this out on a virtual machine first, but just with a 40 GB HDD on the virtual machine, and it will look like this except with larger sizes:
i49.tinypic.com/10y124h.png

Order of partition matters, right? The things that are first run faster? If so, the order will be just like in the screen shot:
Windows first, it needs the extra speed....then Ubuntu, then puppy, then xPUD, then all of my files, then the swap, and lastly, the recovery. I probably won't use the last 2 parititons very much since i don't really expect to hibernate or run into a problem, and if i do, then the performance different will be pretty negligible for just that once that it's accessed.

Wrong.
All the debate about this applied to a disk architecture long since relegated to history.
Windows needs a primary (and maybe the recovery - dunno), all the rest can be logicals - in any order that makes sense. I put the swap first because I reuse it for all distros, and so will never get deleted/moved. The rest (for me) are liable to be deleted resized, so makes sense to put them after the swap.
YMMV.

But...the closer to the edge of the HDD, the more data gets read for each spin, right? it isn't a SDD, it's a 5400 rpm HDD.

Also, I can just do a frugal install of puppy and xPUD onto my ubuntu partition, so it'd only need to be:
Windows 7
*buntu/pupp*/xPUD
Files
Swap
Recovery

or would a separate partition for each be more efficient?
I'm never going to fill up this HDD btw. I could be fine with a 10 GB files partition. I've always only used about 20 GB or so, INCLUDING a Vista or 7 installation, on my PC's (hence why a netbook is ideal for me. I pretty much just do software development and web browsing).
But then i got into games and my bro got a steam account so now i'm using 80 GB on THIS computer, but i won't be playing too many games on the netbook, and those would be part of the 40 GB windows partition anyways....

* The asterisks shows that i'm unsure still whether to throw on Eeebuntu or Ubuntu / Puppy or Puppeee Linux

It is a fixed amount of sectors per track. regardless what track it is on.

Puppy has wifi added to it as an afterthought - WEP works, but WPA is "maybe-ish" at best. Look to Puppeee for a corrected puplet for netbooks. Puppy is designed for OLD hardware - no Fn keys to control processor speed, brightness, and other fancy features, no WPA, etc... it excels on old hardware with wired ethernet connection to your DSL router.

You really should burn a few Linuxes to a Live CD or Live USB, if you are uncertain which to use. Put on the one with most promise, and try the others, trying to beat it.

In the long term, you will be longing for one OS to do it all.

Given you have a netbook, you should look at Peppermint, Knoppix, Puppeee, and Leeenux 3.0 (leeenux-linux.com) - it runs really fast on my eeePC with only 512MB of RAM, and handles like a full-blown Mint-Ubuntu. Anything bigger may be a bit on the slow side on your machine.

I run VirtualBox which is in the repositories of my Mint-Ubuntu, on my main core2duo laptop, in which I run XP just so I can use a softphone app that isn't available for Linux. Other than that, I've weaned myself off of Windows entirely, and am increasingly happy that way.

So, these curved and sloping lines, you are saying that they are flat and straight, right?

Amiejay:
I'm currently running WinXP, Puppy 4.20, & DesktopBSD 1.7 on my Acer Aspire One ZG5, and I haven't had any problems with anything. I used an "Acer Remix 2" for puppy, and DesktopBSD picked up my usb wifi stick, an Intellinet G with sma connector for the antenna. I wasn't interested in using the built-in wifi, which is, in my opinion, dismal without the external sma plug. I like external antennas so much I may hack my netbook and install an sma plug up on one side of the screen. Maybe even one on the other side for bluetooth.
I wanted to use Debian on this netbook, but it had network issues with my netbook, as did Slackware, which I'm using (12.2, not 13.0) on my desktop machine again. I've tried the ubuntu 9.10 on the Acer, which seems to run just fine. Windows is ALWAYS the slowest, even after shutting down all unneeded services (a LOT). It is also always FIRST on all my machines, even if it's not the first GRUB entry. PCLinuxOS, Backtrack 4, Helix, and a few others worked real good on this machine also, and are installed on usb sticks. I may try something else like Leeenux 3.0 to see how that works. My main interest is using what I can show off to other techno-challenged folks that installs easily and weens them off the win-doze junk. When you get yours machine set up, why not post your menu.lst file (or equivalent) here for the benefit of others? It could make someones day.

73

Wrong. It's only true for ancient HDDs where physical sector layout is identical to logical one. Reality is that bit density is fixed, not sectors per track. If you look at HDD speed measurements, they will always tell that HDD speed goes lower as the test gets to the center of the disk.
So, Not the order of partitions in MBR matters, but the order of their physical location matters. I.e. you can, e.g. define one partition first, another second, but place them first after second on the HDD. In this case second will be faster than first.

Your response sounds a bit confusing.
You change the term from fixed sectors to fixed bits... same difference.
But you do not take position as to whether you can, for example, store 2MB on one outer track, and only 512K on an inner one. I think that was his point... because if you were able to store more data on the outer tracks, then this affects your interleave, and you would need performance parametrics to evaluate that and there may be different results on different hard drive models.
But to the best of my old outdated not recognized by you knowledge ;-p if you have the same amount of bits written to each track, then track location with regards to performance doesn't matter.

With my outdated knowledge, I've been putting information of decreasing importance towards the center - first the swap file, then the OS file, and on the outermost, my data partition - the stuff I can't re-create by simply reinstalling from a CD.
But I don't know if this is even a good idea, for the following reason:
1) for each bit that you write, more ferrite molecules will be in that bit on the outside - this is my reasoning for it - but the amplitude/depth in layers, of each write, will be weaker per molecule / shallower per layer, on the outside... and will this increase the chance of "weak bits" that ultimately lead to bad data?
2) on the inside, my reasoning is, less molecules per bit, means more risk... but is it really so? as the current penetrates deeper than it does when it covers more "geography" for each bit.

So I am kind of on the fence with this one.

Yes, it is confusing to someone who is does not look at the data and is determined to hold on to a flawed understanding of what is going on.

Perhaps you might want to comment on why you wanted to make this statement:

when the data is clear and shows that your position was wrong? Essentially all the test data on rotating magnetic hard disks (SSDs are different, as has been mentioned) shows the same general pattern; the gently sloping performance curve, if by performance we can accept the rate at which data can be read off the magnetic surface.

0 members found this post helpful.

Contrary to you, I was being polite!
I know lots about hardware, but just because I don't agree, I'm not going to drag a guy into the mud over it.
RESPECT!
You make no valid references, offer no proof, and don't display that you have any understanding of this in any way. If you have nothing else to do, go try a distro.

I have 8 on my netbook. It's an Aspire One. I would give you a dump of fdisk -l, but I am not on my netbook. I use Fedora most often. The others don't get touched very often.

So do you. Is this valid proof?

Maybe! I found it confusing, as was commented in the Discussion page where this better explanation was linked. Seems that the limiting factor is bits per linear distance rather than the data rate the head can read or write but the examples given are for 1996 and 2000 disks.

More recently: Robin Harris' blog from 2007.

EDIT: I knew of this phenomenon but did not realise the outer tracks' sustained transfer rate is ~2 times faster than the inner -- as is shown on the Tom's Hardware tests linked by salasi.

Just because you disagree with the evidence that I pointed out way back in post #5 (not difficult to find such evidence, you'll find similar on a variety of testing sites and a variety of modern hard disks) and made no reference to the data, you do not seem to want to address the issue of why your theory disagrees with actual measurements from real hard disks.

Obviously, this statement is completely wrong. I have already pointed out that I have offered evidence, the issue is that you have ignored it.

In fact the person who has offered no evidence in this debate isn't me at all. But here is some more.

This is from the older version of storage review's Benchmark Suite, as the later version doesn't seem to analyse the data in this way, just giving an average across the surface, perhaps because this figure is easier to compare between hard disks and SSDs.

The data on the inner and outer edges is contained in rows three and fofour of the data ('WB99 Disk/Read Transfer Rate - Begin' and 'WB99 Disk/Read Transfer Rate - End'). These are just the thirteen fastest hard disks in the database and in every case the outer edge is faster.

Here is some more; but I'll also quote from earlier in the article:

If you look unselectively at the latest hard disk tests, you'll find that there are SSD results mixed in and and those have to be excluded from this argument.

But the bigger question is not only why you want to slag off someone who did offer data for not offering data -obviously, the implication is that you think that data has some worth- but you haven't, so far, chosen to offer any data yourself.

I am sure that you don't want people disagreeing with you, but this is something else that will not help.

You forgot to edit out the reference to FLOPPIES on the wikipedia page before submitting it as proof. ;-p