It should now be obvious to anyone who has read through this waste of web space where this and every other thread Modus gets involved with goes. His view is the only one possible. Everything else is wrong. If his point makes no sense he will make something up to try and validate it (i.e. registry values reverting to default). As I said before, he will drag this thing as he always does, to over 100 posts until he gets the last word. I'll admit, I grow weary of the debate after awhile. No its not because, "I'm taking the high road" as you call it. Its because I've made my points, and am done. You don't agree with them. I don't agree with yours.
Ok, Modus, you win. You are right. I am wrong as is everyone else who has a different opinion than yourself. People should only use AMD systems with winmodems, and drives with one partition. They should not waste their time with defragmenting the drive. This is the only acceptable configuration since Modus says so.
Ok, Modus, you win. You are right. I am wrong as is everyone else who has a different opinion than yourself. People should only use AMD systems with winmodems, and drives with one partition. They should not waste their time with defragmenting the drive. This is the only acceptable configuration since Modus says so.
I think Modus is just really annoyed with people trying to refute his statements by ignoring his facts. I would drag this out too if someone kept ignoring what I said.
It is a simple fact that Modus's posts are obsessively long and tenacious. It suggests pathology of personality that someone would care to steam over a keyboard in such a way for basically meaningless purpose in front of a screen of computer text. Modus will not see his own pathology and that is part of what mind pathology is. Scoring virtual points and slaying imaginary dragons is all basically in his own mind. What others see is basically contentiousness. Anal, obsessive or even paranoid, whatever, it's peculiar.
Remnant2,
<<Ok, here's a question for all of the one-big-partition proponents.>>
A question that has been raised and subsequently dismissed as moot, over ten times in this thread! Yet people like you, too lazy to read the entire thing, jump in and act as if you're giving us a Revelation. Please, no more Defrag arguments. This is the last one, OK? I'm getting really tired of writing the same thing over and over for people who can't be bothered to scroll up and inform themselves.
<<How the hell do you defrag those puppies in any reasonable time?>>
First of all, Defragmentation yields best results when done in Safe Mode, free from the rude interruption of other programs and Windows itself. There, you will find that it takes just as long to defrag a 1G and 3G partition as it takes to defrag a 4G partition. Same difference.
Add to this the fact that according to Microsoft, defragmentation is not supposed to be done more than once every two weeks, and yields no measurable performance benefit is done more often. It is *always* a walk-away job done during normal system downtime, which makes its performance completely irrelevant. So bending your entire drive management scheme around Defrag performance is a bit silly.
JellyBaby,
<<I've rarely witnessed such odd argumentation in a thread. I'm afraid this one has become quite convoluted. It's really funny. . . Very odd and pointless.>>
Much like your post.
<<Again, to each his own and there are several correct answers to this thread's question.>>
We're not picking ice cream flavors here. There's only one best answer.
<<Anyway, you can make arguments for or against partitioning. But at least one person in this thread seems to be in a quest to refute any pro-partitioning stances. I'm not sure why. It's almost like trying to convince someone to change his favorite color.>>
A favorite color is an emotional, irrational choice that has no practical bearing on anything and therefore isn't a matter for debate. But people who partition do it ostensibly because they claim it provides several key benefits. By explaining why these benefits are either worthless or non-existant, we force them into blanket statements of preference: "To each his own," "It helps me stay organized," "There are many valid schemes." This exposes the root motivation behind partitioning and other useless system tweaks -- it makes people feel better.
oldfart,
<<If his point makes no sense he will make something up to try and validate it (i.e. registry values reverting to default).>>
That's the second time you've brought that registry thing up and I still have no idea what you're talking about. I think you must have misread some earlier statement of mine.
Rigoletto,
<<It is a simple fact that Modus's posts are obsessively long and tenacious.>>
Passages beginning with the words "it is a simple fact that. . ." usually have nothing worthwile to say, and yours is just another example.
<<It suggests pathology of personality that someone would care to steam over a keyboard in such a way for basically meaningless purpose in front of a screen of computer text.>>
And what is suggested by a lack of coherent grammar? A childhood educational trauma? What of resorting to bold type in a desperate bid to have one's writing noticed? Subconcious feelings of phalic inadequacy, perhaps?
Modus
<<Ok, here's a question for all of the one-big-partition proponents.>>
A question that has been raised and subsequently dismissed as moot, over ten times in this thread! Yet people like you, too lazy to read the entire thing, jump in and act as if you're giving us a Revelation. Please, no more Defrag arguments. This is the last one, OK? I'm getting really tired of writing the same thing over and over for people who can't be bothered to scroll up and inform themselves.
<<How the hell do you defrag those puppies in any reasonable time?>>
First of all, Defragmentation yields best results when done in Safe Mode, free from the rude interruption of other programs and Windows itself. There, you will find that it takes just as long to defrag a 1G and 3G partition as it takes to defrag a 4G partition. Same difference.
Add to this the fact that according to Microsoft, defragmentation is not supposed to be done more than once every two weeks, and yields no measurable performance benefit is done more often. It is *always* a walk-away job done during normal system downtime, which makes its performance completely irrelevant. So bending your entire drive management scheme around Defrag performance is a bit silly.
JellyBaby,
<<I've rarely witnessed such odd argumentation in a thread. I'm afraid this one has become quite convoluted. It's really funny. . . Very odd and pointless.>>
Much like your post.
<<Again, to each his own and there are several correct answers to this thread's question.>>
We're not picking ice cream flavors here. There's only one best answer.
<<Anyway, you can make arguments for or against partitioning. But at least one person in this thread seems to be in a quest to refute any pro-partitioning stances. I'm not sure why. It's almost like trying to convince someone to change his favorite color.>>
A favorite color is an emotional, irrational choice that has no practical bearing on anything and therefore isn't a matter for debate. But people who partition do it ostensibly because they claim it provides several key benefits. By explaining why these benefits are either worthless or non-existant, we force them into blanket statements of preference: "To each his own," "It helps me stay organized," "There are many valid schemes." This exposes the root motivation behind partitioning and other useless system tweaks -- it makes people feel better.
oldfart,
<<If his point makes no sense he will make something up to try and validate it (i.e. registry values reverting to default).>>
That's the second time you've brought that registry thing up and I still have no idea what you're talking about. I think you must have misread some earlier statement of mine.
Rigoletto,
<<It is a simple fact that Modus's posts are obsessively long and tenacious.>>
Passages beginning with the words "it is a simple fact that. . ." usually have nothing worthwile to say, and yours is just another example.
<<It suggests pathology of personality that someone would care to steam over a keyboard in such a way for basically meaningless purpose in front of a screen of computer text.>>
And what is suggested by a lack of coherent grammar? A childhood educational trauma? What of resorting to bold type in a desperate bid to have one's writing noticed? Subconcious feelings of phalic inadequacy, perhaps?
Modus
This is ridiculous, I don't know what to believe. However, I do have 3 OS's and they work fine when I have partitioned them on my 30GB HD. I use FAT32 for everything except Linux.
I could care less for optimum performance or wasted space or....whatever crap you can come up with. I haven't even used half of the HD so I really don't care.
Anyways I partition to organize my data. I also like the fact that I can run scandisk and defrag one partition at a time. Also too many partitions can make daily life tedious, so be carefull.
You guys sure have alot of time to waste on the holidays!
STOP THE ARGUING, it's useless.
I could care less for optimum performance or wasted space or....whatever crap you can come up with. I haven't even used half of the HD so I really don't care.
Anyways I partition to organize my data. I also like the fact that I can run scandisk and defrag one partition at a time. Also too many partitions can make daily life tedious, so be carefull.
You guys sure have alot of time to waste on the holidays!
STOP THE ARGUING, it's useless.
I see the civility of this thread has done nothing but improve since last I checked in.
"I highly doubt it. As Modus pointed out, your average wasted space on a drive regardless of cluster size is
0.5 * ClusterSize * FileCount = WastedSpace"
This equation is terribly inaccurate and not a good method for estimating wasted space. Simple example, using the numbers from one of my previous posts, if you calculated the file count for 4k and 32k cluster, one gave a result of 9,000, the other one over 22,000 files. The larger answer has 2.5 times as many files and neither answer being close to the actual number. I have no where near 700,000+ files on those 2 drives, that's a ridiculously large number of files.
The "real" equation to calculate wasted space is:
(cluster size - average file size per cluster) * number of files
"You know, I was thinking the same thing, and I did the math myself to figure out what in the name of Richard Simmons this guy was storing, but when I discovered his average file size, I decided I'd better not say anything. After all, he deserves his dignity."
It's none of you business what I have on my drives, but that obviously doesn't seem to stop you from incinuating lewd comments and character attacks on other chatters. That being said, I'll tell you what I actually have on my drives. Roughly 10GB of apps and games, 2 copies of my win2k c: partition (don't worry why, I'm just lazy), mp3's, my shared (98&2k) internet temp folder, all my drivers, and a full installation of MS Visual Studio, which is where the majority of the wasting goes, as it is basically 3 CD's of dll's, text files, and HTML pages. If you pull your head out of the gutter, you may be able to think more clearly and develop more sound arguements instead of insults that tick off the rest of the people in the thread.
Descend492
As Radboy stated, win2k is FAT32 compatible which is reason enough to upgrade from WinNT. NTFS is definitely more advanced than FAT16, whether it is better depends on the person. A lot of the NTFS exclusive features aren't really that useful to the average home user. Things like disk quotas and file encryption aren't typically utilized at home. If your drive has more than like 8GB, you are basically required for sanity's sake to use NTFS or you will run out of letters in the alphabet assigning drive letters. Probably the biggest benefit of using FAT16, is it is easier to try and recover a corrupted FAT16 partition than an NTFS partition. Using a win98 boot disk and Norton or other disk utility is easier than trying to figure out the NT recovery process.
"I highly doubt it. As Modus pointed out, your average wasted space on a drive regardless of cluster size is
0.5 * ClusterSize * FileCount = WastedSpace"
This equation is terribly inaccurate and not a good method for estimating wasted space. Simple example, using the numbers from one of my previous posts, if you calculated the file count for 4k and 32k cluster, one gave a result of 9,000, the other one over 22,000 files. The larger answer has 2.5 times as many files and neither answer being close to the actual number. I have no where near 700,000+ files on those 2 drives, that's a ridiculously large number of files.
The "real" equation to calculate wasted space is:
(cluster size - average file size per cluster) * number of files
"You know, I was thinking the same thing, and I did the math myself to figure out what in the name of Richard Simmons this guy was storing, but when I discovered his average file size, I decided I'd better not say anything. After all, he deserves his dignity."
It's none of you business what I have on my drives, but that obviously doesn't seem to stop you from incinuating lewd comments and character attacks on other chatters. That being said, I'll tell you what I actually have on my drives. Roughly 10GB of apps and games, 2 copies of my win2k c: partition (don't worry why, I'm just lazy), mp3's, my shared (98&2k) internet temp folder, all my drivers, and a full installation of MS Visual Studio, which is where the majority of the wasting goes, as it is basically 3 CD's of dll's, text files, and HTML pages. If you pull your head out of the gutter, you may be able to think more clearly and develop more sound arguements instead of insults that tick off the rest of the people in the thread.
Descend492
As Radboy stated, win2k is FAT32 compatible which is reason enough to upgrade from WinNT. NTFS is definitely more advanced than FAT16, whether it is better depends on the person. A lot of the NTFS exclusive features aren't really that useful to the average home user. Things like disk quotas and file encryption aren't typically utilized at home. If your drive has more than like 8GB, you are basically required for sanity's sake to use NTFS or you will run out of letters in the alphabet assigning drive letters. Probably the biggest benefit of using FAT16, is it is easier to try and recover a corrupted FAT16 partition than an NTFS partition. Using a win98 boot disk and Norton or other disk utility is easier than trying to figure out the NT recovery process.
BoberFett
Lifer
- Oct 9, 1999
- 37,562
- 9
- 81
Wrong, wrong, wrong. You're completely confused Pariah, which I wouldn't normally care about, but I don't want to see any computer newbies confused by your bogus numbers. I also don't like being told I'm wrong by somebody who obviously doesn't have a clue.
<< (cluster size - average file size per cluster) * number of files >>
Where did you come up with that complete BS? The equation I have is the correct one. Let's think about it logically.
For every file stored on your hard drive, there will be one and only one cluster that is not completely filled by that file. If the file is smaller than your current cluster file, that will be the first cluster. Anything larger than your cluster size will completely fill all but the last cluster. So you take an average which when dealing with 2 points is halfway between those points: 0.5 * ClusterSize. With 32K cluster, some files will fill 1KB in the last cluster they use. Some files will fill 31K of the last cluster they use. Some will fill 6K of their last cluster, some will fill 26K. The average is 16K, or 0.5 * ClusterSize. Unless you've recently proven mathematical concepts such as averages to be false, people should stick with my equation. Take some logic classes Pariah, they'll do you a world of good.
<< (cluster size - average file size per cluster) * number of files >>
Where did you come up with that complete BS? The equation I have is the correct one. Let's think about it logically.
For every file stored on your hard drive, there will be one and only one cluster that is not completely filled by that file. If the file is smaller than your current cluster file, that will be the first cluster. Anything larger than your cluster size will completely fill all but the last cluster. So you take an average which when dealing with 2 points is halfway between those points: 0.5 * ClusterSize. With 32K cluster, some files will fill 1KB in the last cluster they use. Some files will fill 31K of the last cluster they use. Some will fill 6K of their last cluster, some will fill 26K. The average is 16K, or 0.5 * ClusterSize. Unless you've recently proven mathematical concepts such as averages to be false, people should stick with my equation. Take some logic classes Pariah, they'll do you a world of good.
Have to admit, this thread is entertaining. Obvious that no side is gonna convince the other side of anything. What's surprising is that partitioning is such a minor & personal subject. The newby may not know yet, but it's futile to convince another person what works best for *them* if they weighed the pro's and con's for themselves .. & have plenty of pwersonal experience w/ hard drives. Anybody can say what works best for them, and they can explain their reasoning to the newbie, but no more.
Pragmatically, for the person who doesn't know how to use FDISK, Modus' way is easier, cuz they wouldn't have to learn the app .. but it's not very difficult, so that a weak argument at best.
Thot it was common knowledge that the leading/outer edge of a hard drive offered better performance .. which is why disk utility apps like Norton put the swap file there. For similar reasons, I want my OS & apps there .. not at the end of a 45-gig drive (45-gigs is a big drive). As you can plainly see here, data thru-puts get progressively lower/worse as you move toward the inner part of the disk .. (cuz of slower linear velocity).
If I'm doing it wrong, I really wanna know .. but to me, seems clearly obvious that a well-partitioned drive - done right the first time .. to serve a specific purpose - is the way to go. Not trying to talk anybody into my methodology. That's how I feel. Nothing I read has convinced me otherwise .. not even close.
Pragmatically, for the person who doesn't know how to use FDISK, Modus' way is easier, cuz they wouldn't have to learn the app .. but it's not very difficult, so that a weak argument at best.
Thot it was common knowledge that the leading/outer edge of a hard drive offered better performance .. which is why disk utility apps like Norton put the swap file there. For similar reasons, I want my OS & apps there .. not at the end of a 45-gig drive (45-gigs is a big drive). As you can plainly see here, data thru-puts get progressively lower/worse as you move toward the inner part of the disk .. (cuz of slower linear velocity).
If I'm doing it wrong, I really wanna know .. but to me, seems clearly obvious that a well-partitioned drive - done right the first time .. to serve a specific purpose - is the way to go. Not trying to talk anybody into my methodology. That's how I feel. Nothing I read has convinced me otherwise .. not even close.
Jeezus Descend. Look what you have started. Man now I know why Modus is Elite. I don't even bother reading it cuz it's so danm long.
Just like the winmodem threads, if you can't beat them with facts, bore them into silence with repeated cut and pasted "facts"
Modus, realize that there are other people out there with as much or more experience then you.
I have read enough of this thread to realize that on my next system I will still partition my harddrive. Fdisk is not difficult to use. Partitions help ME organize my data in a logical fashion. I like my OS in a small 2 gig partition at the beginning of my drive etc.
It works for me, I like it that way. End of my discussion in this thread.
Modus, realize that there are other people out there with as much or more experience then you.
I have read enough of this thread to realize that on my next system I will still partition my harddrive. Fdisk is not difficult to use. Partitions help ME organize my data in a logical fashion. I like my OS in a small 2 gig partition at the beginning of my drive etc.
It works for me, I like it that way. End of my discussion in this thread.
BoberFett
Lifer
- Oct 9, 1999
- 37,562
- 9
- 81
<< bore them into silence with repeated cut and pasted "facts" >>
*chuckle*
Just because I don't feel like letting this thread die...
The guy asked for opinions. Modus proceeded to give his usual, uh, verbose answer. Pariah proceeded to tell him he was wrong. It has since been established that Pariah doesn't know what the hell he's talking about, so I find no fault in Modus here.
"Where did you come up with that complete BS? The equation I have is the correct one. Let's think about it logically."
Here you go:
http://www.compuclinic.com/osr2faq/index.html#fat32
I suggest you email these people and tell them their formula is incorrect and why. Your formula can potentially be correct if by some miracle you find a drive that has an average of half of every cluster filled, which you will never find in the real world. The formula I posted calculates the exact amount of slack under any condition without error.
You obviously didn't plug the numbers in from my earlier post. So here I have done it for you, using your formula:
Files on 4K cluster
1,053,912,432 bytes stored
1,072,504,832 bytes used
18,592,400 bytes wasted
Files on 32K cluster
1,053,912,432 bytes stored
1,428,488,192 bytes used
374,575,760 bytes wasted
0.5 * ClusterSize * FileCount = WastedSpace
Filecount = (18,592,400 / 0.5) / 4000 (4K cluster)
Filecount = 9296
Filecount = (374,575,760 / 0.5) / 32000 (32K cluster)
Filecount = 23,411
Now since you are the logic ace here bober, you point out the math error I made here, and I will admit I am wrong and we'll be done with this retarded argument.
Here you go:
http://www.compuclinic.com/osr2faq/index.html#fat32
I suggest you email these people and tell them their formula is incorrect and why. Your formula can potentially be correct if by some miracle you find a drive that has an average of half of every cluster filled, which you will never find in the real world. The formula I posted calculates the exact amount of slack under any condition without error.
You obviously didn't plug the numbers in from my earlier post. So here I have done it for you, using your formula:
Files on 4K cluster
1,053,912,432 bytes stored
1,072,504,832 bytes used
18,592,400 bytes wasted
Files on 32K cluster
1,053,912,432 bytes stored
1,428,488,192 bytes used
374,575,760 bytes wasted
0.5 * ClusterSize * FileCount = WastedSpace
Filecount = (18,592,400 / 0.5) / 4000 (4K cluster)
Filecount = 9296
Filecount = (374,575,760 / 0.5) / 32000 (32K cluster)
Filecount = 23,411
Now since you are the logic ace here bober, you point out the math error I made here, and I will admit I am wrong and we'll be done with this retarded argument.
Modus : You didn't read, or didn't understand, my point.
Having to boot to safe mode and switch off virtual memory to defrag a single partition is a serious pain in the ass. With multiple partitions, I can defrag without any special worries, since I know windows won't try to write to the drive I'm defragging (of course, defragging the C drive means you DO have to be more careful, but if you make sure you don't store too much on it, its not a bit deal.)
And I can tell you for a fact that a defrag that gets interrupted by disk activity takes far longer than one that doesn't. A defrag of extended partitions has no problem with this; a single partition often does, as you cannot get windows to completely quiet down its virtual mem swapping. That is my point. It is a point backed up by just about every defrag I've done of many, many, win95/98 systems.
And no, I don't defrag more than every few months typically. A single partition means that you need to do the whole damn HD. I usually defrag before I install a large app, in which case I like my defrag to take 20mins instead of 20 hours, by only defragging the partition that needs it.
So although in theory what you say is true, in practice its far easier to simply partition your drive -- and gain free organization, speedier and easier defrags, and slightly more space (squeezing a couple hundred more MBs out of my drive I won't sneeze at all).
Having to boot to safe mode and switch off virtual memory to defrag a single partition is a serious pain in the ass. With multiple partitions, I can defrag without any special worries, since I know windows won't try to write to the drive I'm defragging (of course, defragging the C drive means you DO have to be more careful, but if you make sure you don't store too much on it, its not a bit deal.)
And I can tell you for a fact that a defrag that gets interrupted by disk activity takes far longer than one that doesn't. A defrag of extended partitions has no problem with this; a single partition often does, as you cannot get windows to completely quiet down its virtual mem swapping. That is my point. It is a point backed up by just about every defrag I've done of many, many, win95/98 systems.
And no, I don't defrag more than every few months typically. A single partition means that you need to do the whole damn HD. I usually defrag before I install a large app, in which case I like my defrag to take 20mins instead of 20 hours, by only defragging the partition that needs it.
So although in theory what you say is true, in practice its far easier to simply partition your drive -- and gain free organization, speedier and easier defrags, and slightly more space (squeezing a couple hundred more MBs out of my drive I won't sneeze at all).
hehe...paldo...whoops
at least this means I'll now have more space for my server...none of this crappy "but I only have 300 megs left"
at least this means I'll now have more space for my server...none of this crappy "but I only have 300 megs left"
Mule,
<<You guys sure have alot of time to waste on the holidays!>>
Isn't that the definition of a holiday?
Remnant2,
<<Having to boot to safe mode and switch off virtual memory to defrag a single partition is a serious pain in the ass.>>
Well call it what you like, but if you want to see the most benefit from defragmentation, not to mention optimize your permanent swap file, you'll have to. Partitioning or not. So it's a moot point.
<<With multiple partitions, I can defrag without any special worries, since I know windows won't try to write to the drive I'm defragging>>
Not if the one you're defragging is C: (and since partitioners seem to be obsessed with defragging, that's what they'll want to defrag the most.)
<<A defrag of extended partitions has no problem with this; a single partition often does, as you cannot get windows to completely quiet down its virtual mem swapping.>>
I don't know what's wrong with your system (perhaps too many background apps running) but the vast majority of Windows machines can and do get through a defrag of their OS partition in reasonable time, regardless of its size. If you can't even complete a defrag, you need to take a serious look at your system and figure out what's constantly thrashing the disk when it shouldn't be, because that's not normal.
Pariah,
<<Your formula can potentially be correct if by some miracle you find a drive that has an average of half of every cluster filled, which you will never find in the real world.>>
OK, listen up, I'm going to try to explain this from the ground up. Please, please, try to wrap your mind around it as I don't feel like explaining it again. First, you have to understand how any random access file system (including FAT32 and NTFS) stores a file using clusters. The procedure is pretty simple. Take this analogy:
Imagine pouring water from a jug into many cups. No cup can hold all the water, but if you have enough cups, you can do it. So you pour the water into the first cup. Now, if the jug doesn't have much water in it, you won't even fill up the first cup, leaving some wasted space in the first cup. Fine. But usually, your water will require dozens of cups. So you fill up each cup to the brim, until the jug is emptied. More often than not, the water will finish without filling up a cup perfectly, leaving some empty, wasted space, but only in the last cup. All the other cups are perfectly full, it's just the last cup that has some empty space.
Now, how much space will the average jug of water waste? Half a cup. Why? Probability. Probability dictates that the last cup will sometimes be almost empty and sometimes be almost full and sometimes be half full (half empty?) but on average, the water from the jug will wind up filling the last cup halfway. It's a basic mathematical principle that is the basis for everything from coin tosses to quantum theory.
Now, to apply this to cluster slack. Imagine the water in the jug is your file data, and the cups are clusters. Windows fills up the first cluster completely, fills up the second completely, and so on, until it runs out of file data. At that point, the last cluster is filled with the remaining data. Sometimes it will be mostly full and sometimes it will be hardly full. But, again, probability dictates that the last cluster, on average, will be half full and half wasted. Therefore, the average file will waste half a cluster.
So my forumla is nothing but common sense. It simply says that, if you know your cluster size, and you know how many files you have, you can give a very reasonable estimate of how much space you waste:
Space Wasted = (Cluster Size / 2) * Number of Files
Or we can assume that the Number of Files is equal to the Data Size divided by the Average File Size, yielding
Space Wasted = (Cluster Size / 2) * (Data Size / Average File Size)
Ask anyone who knows anything about hard disk structures and file systems and they'll confirm that this formula is statistically sound.
<<The formula I posted calculates the exact amount of slack under any condition without error. . . The "real" equation to calculate wasted space is: (cluster size - average file size per cluster) * number of files>>
Oh really. Right, then show us. Plug some numbers in that formula and let's see it calculate "the exact amount of slack under any condition without error."
Look, no formula can do that. It's literally impossible. You can't calculate *exact* slack waste without knowing the size of every single file on the drive. You can only estimate, and your forumla is not even a good estimate. In fact, it doesn't make sense at all. I don't care where you got it (though, typical for this kind of "accepted wisdom", the hole-in-the-wall source claims the email address of the formula's author is me@myself.com, credible indeed), it's result is completely useless in every way. Let me explain why:
"Average file size per cluster" is meaningless. There is no such thing. Now, Average File Size is a significant figure. Cluster Size is a useful figure. But "average file size per cluster" doesn't refer to anything at all. That immediately destroys your forumla.
So listen to BoberFett. You haven't the foggiest idea what you're talking about.
Radboy,
<<it's futile to convince another person what works best for *them* if they weighed the pro's and con's for themselves .. & have plenty of pwersonal experience w/ hard drives. Anybody can say what works best for them, and they can explain their reasoning to the newbie, but no more.>>
Yes of course, but the problem is, the "reasoning" of the partitioning crowd doesn't add up. They cannot show any tangible benefit from partitioning. Every one of their pro-partitioning reasons has been shot down:
- Defrag and Scandisk times are irrelevant because an entire drive still takes just as long to process as the same drive split into many pieces, and more importantly because Defrag is a walk-away job done only twice a month during normal system downtime.
- Imaging is no easier with multiple partitions because a restored image of an intital clean OS partition will necessitate reinstallation of all applications, drivers, and settings added since the image was made, meaning that the only time saved by the image was the time needed to copy in the OS itself (which is really no slower when done by its own SETUP program) and the person's data files (which must always be backed up externally for proper precaution, making a data backup partition redundant).
- Cluster slack is no longer an issue. With FAT32 and the relatively large digital media files that fill up today's large hard drives, it typically hovers below 2% for normal mass storage -- insignificant. This has been proven both theoretically and by an informal survey in this thread.
- File organization is not assisted by multiple volumes and a slew of drive letters. In fact, it is hindered, because the best solution is a simple nested folder tree with descriptive names, which never needs to be FDISK'ed or Partition Magic'ed at all. Just drag and drop.
- Performance gains from putting your OS and swap file near the beginnning of the drive while exiling your data and applications to the end, are dubious at best and have never been proven by any sort of real world benchmark. In fact, they could actually decrease performance by forcing the drive to seek back and forth much farther across the disc surface during normal multitasking than what would be necessary with a typical single-partition configuration.
<<If I'm doing it wrong, I really wanna know>>
The problem is not that you're doing it wrong, it's that you're doing it at all. It's like asking if you're using the right method to toothbrush the floor -- you might be great with a toothbrush, but why not grab a mop?
Modus
<<You guys sure have alot of time to waste on the holidays!>>
Isn't that the definition of a holiday?
Remnant2,
<<Having to boot to safe mode and switch off virtual memory to defrag a single partition is a serious pain in the ass.>>
Well call it what you like, but if you want to see the most benefit from defragmentation, not to mention optimize your permanent swap file, you'll have to. Partitioning or not. So it's a moot point.
<<With multiple partitions, I can defrag without any special worries, since I know windows won't try to write to the drive I'm defragging>>
Not if the one you're defragging is C: (and since partitioners seem to be obsessed with defragging, that's what they'll want to defrag the most.)
<<A defrag of extended partitions has no problem with this; a single partition often does, as you cannot get windows to completely quiet down its virtual mem swapping.>>
I don't know what's wrong with your system (perhaps too many background apps running) but the vast majority of Windows machines can and do get through a defrag of their OS partition in reasonable time, regardless of its size. If you can't even complete a defrag, you need to take a serious look at your system and figure out what's constantly thrashing the disk when it shouldn't be, because that's not normal.
Pariah,
<<Your formula can potentially be correct if by some miracle you find a drive that has an average of half of every cluster filled, which you will never find in the real world.>>
OK, listen up, I'm going to try to explain this from the ground up. Please, please, try to wrap your mind around it as I don't feel like explaining it again. First, you have to understand how any random access file system (including FAT32 and NTFS) stores a file using clusters. The procedure is pretty simple. Take this analogy:
Imagine pouring water from a jug into many cups. No cup can hold all the water, but if you have enough cups, you can do it. So you pour the water into the first cup. Now, if the jug doesn't have much water in it, you won't even fill up the first cup, leaving some wasted space in the first cup. Fine. But usually, your water will require dozens of cups. So you fill up each cup to the brim, until the jug is emptied. More often than not, the water will finish without filling up a cup perfectly, leaving some empty, wasted space, but only in the last cup. All the other cups are perfectly full, it's just the last cup that has some empty space.
Now, how much space will the average jug of water waste? Half a cup. Why? Probability. Probability dictates that the last cup will sometimes be almost empty and sometimes be almost full and sometimes be half full (half empty?
) but on average, the water from the jug will wind up filling the last cup halfway. It's a basic mathematical principle that is the basis for everything from coin tosses to quantum theory.Now, to apply this to cluster slack. Imagine the water in the jug is your file data, and the cups are clusters. Windows fills up the first cluster completely, fills up the second completely, and so on, until it runs out of file data. At that point, the last cluster is filled with the remaining data. Sometimes it will be mostly full and sometimes it will be hardly full. But, again, probability dictates that the last cluster, on average, will be half full and half wasted. Therefore, the average file will waste half a cluster.
So my forumla is nothing but common sense. It simply says that, if you know your cluster size, and you know how many files you have, you can give a very reasonable estimate of how much space you waste:
Space Wasted = (Cluster Size / 2) * Number of Files
Or we can assume that the Number of Files is equal to the Data Size divided by the Average File Size, yielding
Space Wasted = (Cluster Size / 2) * (Data Size / Average File Size)
Ask anyone who knows anything about hard disk structures and file systems and they'll confirm that this formula is statistically sound.
<<The formula I posted calculates the exact amount of slack under any condition without error. . . The "real" equation to calculate wasted space is: (cluster size - average file size per cluster) * number of files>>
Oh really. Right, then show us. Plug some numbers in that formula and let's see it calculate "the exact amount of slack under any condition without error."
Look, no formula can do that. It's literally impossible. You can't calculate *exact* slack waste without knowing the size of every single file on the drive. You can only estimate, and your forumla is not even a good estimate. In fact, it doesn't make sense at all. I don't care where you got it (though, typical for this kind of "accepted wisdom", the hole-in-the-wall source claims the email address of the formula's author is me@myself.com, credible indeed), it's result is completely useless in every way. Let me explain why:
"Average file size per cluster" is meaningless. There is no such thing. Now, Average File Size is a significant figure. Cluster Size is a useful figure. But "average file size per cluster" doesn't refer to anything at all. That immediately destroys your forumla.
So listen to BoberFett. You haven't the foggiest idea what you're talking about.
Radboy,
<<it's futile to convince another person what works best for *them* if they weighed the pro's and con's for themselves .. & have plenty of pwersonal experience w/ hard drives. Anybody can say what works best for them, and they can explain their reasoning to the newbie, but no more.>>
Yes of course, but the problem is, the "reasoning" of the partitioning crowd doesn't add up. They cannot show any tangible benefit from partitioning. Every one of their pro-partitioning reasons has been shot down:
- Defrag and Scandisk times are irrelevant because an entire drive still takes just as long to process as the same drive split into many pieces, and more importantly because Defrag is a walk-away job done only twice a month during normal system downtime.
- Imaging is no easier with multiple partitions because a restored image of an intital clean OS partition will necessitate reinstallation of all applications, drivers, and settings added since the image was made, meaning that the only time saved by the image was the time needed to copy in the OS itself (which is really no slower when done by its own SETUP program) and the person's data files (which must always be backed up externally for proper precaution, making a data backup partition redundant).
- Cluster slack is no longer an issue. With FAT32 and the relatively large digital media files that fill up today's large hard drives, it typically hovers below 2% for normal mass storage -- insignificant. This has been proven both theoretically and by an informal survey in this thread.
- File organization is not assisted by multiple volumes and a slew of drive letters. In fact, it is hindered, because the best solution is a simple nested folder tree with descriptive names, which never needs to be FDISK'ed or Partition Magic'ed at all. Just drag and drop.
- Performance gains from putting your OS and swap file near the beginnning of the drive while exiling your data and applications to the end, are dubious at best and have never been proven by any sort of real world benchmark. In fact, they could actually decrease performance by forcing the drive to seek back and forth much farther across the disc surface during normal multitasking than what would be necessary with a typical single-partition configuration.
<<If I'm doing it wrong, I really wanna know>>
The problem is not that you're doing it wrong, it's that you're doing it at all. It's like asking if you're using the right method to toothbrush the floor -- you might be great with a toothbrush, but why not grab a mop?
Modus
Just got home. What happened to the last two pages of posts? Still have the pages open .. can copy-n-paste here.
I said this:
No one will say you're not persistent.
Modus says - Imaging is no easier with multiple partitions because a restored image of an intital clean OS partition will necessitate reinstallation of all applications, drivers, and settings added since the image was made .. yada, yada.
You're assuming the only images people restore are that of "an initial clean OS". But you assume wrong. This is understandable if you if don't make images, which you can't do if you only have a single, large partition (unless you have the newest version & burn, or have multiple drives). Most ppl I know image their boot drive that contains not only the OS, but also all apps (but no games, too big, games go on a separate partition .. which you don't have) .. including all the settings/tweaks. All you need to back up is individual files that you work on between images (about every 2 weeks for me).
If I install new drivers, or something like DX8, which can't be uninstalled reliably .. or anything that hoses my system for *whatever* reason, I can restore an image in *minutes* .. and be back up & running. No sweat. I done this about 10 times in the last couple years. I've *never* (even once) restored an "initial clean image" like you say.
Also, I like to have my OS, apps & swap/page file at/near the leading edge of the drive, where performance is best .. and I like to put my ghost images & files that I keep only for back-up purposes (not used in normal operation) at the end of the drive, where perf relatively sux .. and I can only do this with partitions. Even 3 is fine - beginning/fastest, middle/average, end/slowest. I don't want to waste the best performing part of my drive with 'dead' files that I (hopefully) will never use.
Let's try this one - a real-life example I've seen several times. After using W9x for some period of time, a User wants to try W2K, cuz he hears it's a cool, stable OS. If he has a few partitions, he can clean one out and put it there, dual-booting, no prob. If he doesn't like W2K, he deletes it, no prob, and still has W9x. But if he follows your single/large partitions scheme, he either has to wipe out the data on the drive & repartition with FDISK, and spend his hard-earned money and buy Partition Magic to repartition the disk to make room for W2K.
I said this:
No one will say you're not persistent.
Modus says - Imaging is no easier with multiple partitions because a restored image of an intital clean OS partition will necessitate reinstallation of all applications, drivers, and settings added since the image was made .. yada, yada.
You're assuming the only images people restore are that of "an initial clean OS". But you assume wrong. This is understandable if you if don't make images, which you can't do if you only have a single, large partition (unless you have the newest version & burn, or have multiple drives). Most ppl I know image their boot drive that contains not only the OS, but also all apps (but no games, too big, games go on a separate partition .. which you don't have) .. including all the settings/tweaks. All you need to back up is individual files that you work on between images (about every 2 weeks for me).
If I install new drivers, or something like DX8, which can't be uninstalled reliably .. or anything that hoses my system for *whatever* reason, I can restore an image in *minutes* .. and be back up & running. No sweat. I done this about 10 times in the last couple years. I've *never* (even once) restored an "initial clean image" like you say.
Also, I like to have my OS, apps & swap/page file at/near the leading edge of the drive, where performance is best .. and I like to put my ghost images & files that I keep only for back-up purposes (not used in normal operation) at the end of the drive, where perf relatively sux .. and I can only do this with partitions. Even 3 is fine - beginning/fastest, middle/average, end/slowest. I don't want to waste the best performing part of my drive with 'dead' files that I (hopefully) will never use.
Let's try this one - a real-life example I've seen several times. After using W9x for some period of time, a User wants to try W2K, cuz he hears it's a cool, stable OS. If he has a few partitions, he can clean one out and put it there, dual-booting, no prob. If he doesn't like W2K, he deletes it, no prob, and still has W9x. But if he follows your single/large partitions scheme, he either has to wipe out the data on the drive & repartition with FDISK, and spend his hard-earned money and buy Partition Magic to repartition the disk to make room for W2K.
Then Modus said this:
Radboy,
<<You're assuming the only images people restore are that of "an initial clean OS". But you assume wrong.>>
Well any other imaging scheme, such as the one you described, is counterproductive because you end up imaging practically the entire drive, which you could do anyway with a big partition. Not to mention the fact that you must ALWAYS have an external backup of any image or data file that you consider important enough to rely on. As soon as you do that, partitioning becomes redundant.
The person who simply formats and reinstalls everything from a single-partition image is going to be up and running just as fast as you, except for the games (which he will have legal original discs for anyway).
<<Also, I like to have my OS, apps & swap/page file at/near the leading edge of the drive, where performance is best>>
We've already explained why this doens't help and why no benchmarks have ever shown a performance benefit in real world usage and why it might actually hinder performance. So you guys can drop it now.
Modus
Radboy,
<<You're assuming the only images people restore are that of "an initial clean OS". But you assume wrong.>>
Well any other imaging scheme, such as the one you described, is counterproductive because you end up imaging practically the entire drive, which you could do anyway with a big partition. Not to mention the fact that you must ALWAYS have an external backup of any image or data file that you consider important enough to rely on. As soon as you do that, partitioning becomes redundant.
The person who simply formats and reinstalls everything from a single-partition image is going to be up and running just as fast as you, except for the games (which he will have legal original discs for anyway).
<<Also, I like to have my OS, apps & swap/page file at/near the leading edge of the drive, where performance is best>>
We've already explained why this doens't help and why no benchmarks have ever shown a performance benefit in real world usage and why it might actually hinder performance. So you guys can drop it now.
Modus
Then I said this:
Modus, I'm curious how many images have you created and restored? By your reply, I tend to think none. You say it's counterproductive cuz you end up imaging the whole drive. Showing ur @ss here. That's exactly why you make partitions .. so you *don't* have to image the whole/entire drive.
Let's use some real numbers to illustrate. The guy who posted the original Q asked about a 46GB drive. Let's use that. And let's say we have 40 gigs of data on the drive. Sound reasonable?
First off, your method - the Modus single, 46GB partition - does not allow you to make an image .. as Ghost will not image to a partition from which it's being created. This makes sense, cuz then you'd be changing the contents of the partition as you created the image. This person will either have to burn their image to CD-R/RWs, but it'll take a long-@ss time to burn 40 gigs worth of data, wasting both time & discs. Or this person will have to go out and buy another hard drive - even if they don't need it. And the image will still use up most of the 2nd drive, cuz it will be so big, cuz you don't make partitions to segment data. This makes no sense.
Now let's say we use the Radboy method make a nice, small 3GB partition at the leading edge of the drive, which offers us the best performance. 3GB's is plenty of space for Windows & all your apps, which includes all settings and system configs, excluding games (which go on another partition). This means, when you restore this image (if you don't know), you don't have to re-install all the apps & re-config the system like you indicated in an earlier post. Ghost offers different compression settings, but say I have 2.5 gigs of data (about what I have). Using 'FAST' compression setting will result in an image of ~1.5 gigs. Lots better than the ~30 gig image your method would yield. Your way would have us imaging everything on the drive - MP3s, jpegs, movies, etc. .. *if* at all (your method would require a burner and a stack of disks .. or a separate hard drive, which would be mostly used by the image itself).
So, as you can plainly see I'm *not* (as you say) "imaging practically the entire drive" .. am I? O the contrary, *your* method will image practically the entire drive.
Let me also add here that, of the 10 or so times I've had to restore images, never once was it b-cuz the drive died. Every time was cuz something in Windows got screwed up .. like a new version of DX didn't work in my system, or Windows just wouldn't boot from some reason. In everyone of these cases, it was okay to have the image on the same drive (different partition, tho). But if the drive were to die, then I'd lose the image, too. But this has never happened to me (knock on wood).
I actually store my images on different hard drives from the one with the partition I created them from, but not everyone has multiple drives like me. And not everyone needs another hard drive to receive the image .. unless they follow your partitioning scheme.
You say, "The person who simply formats and reinstalls everything from a single-partition image is going to be up and running just as fast as you". It takes me less than 10 minutes to be up & running with a restored Ghost image, whereas it takes me the better part of a week to "reinstalls everything from a single-partition" (I have a *lot* of apps).
You say, "We've already explained .. why no benchmarks have ever shown a performance benefit in real world usage and why it might actually hinder performance. So you guys can drop it now."
Maybe you've never heard of HD Tach, which clearly shows HERE that performance is better at the leading edge of the drive. Anyone with eyes can see this is true. Do you not see that thru-puts at the leading edge are ~40MB/s, and only 18MB/s at the end of the disk? This is even a graph of a 45GB drive - almost identical to what the original poster asked about. This is why disk optimazation utilities, like Norton, place the swap file at the leading edge of the drive .. where performance is best. Maybe you should send a note to the folks at Norton and tell them they're doing it wrong.
Modus, I'm curious how many images have you created and restored? By your reply, I tend to think none. You say it's counterproductive cuz you end up imaging the whole drive. Showing ur @ss here. That's exactly why you make partitions .. so you *don't* have to image the whole/entire drive.
Let's use some real numbers to illustrate. The guy who posted the original Q asked about a 46GB drive. Let's use that. And let's say we have 40 gigs of data on the drive. Sound reasonable?
First off, your method - the Modus single, 46GB partition - does not allow you to make an image .. as Ghost will not image to a partition from which it's being created. This makes sense, cuz then you'd be changing the contents of the partition as you created the image. This person will either have to burn their image to CD-R/RWs, but it'll take a long-@ss time to burn 40 gigs worth of data, wasting both time & discs. Or this person will have to go out and buy another hard drive - even if they don't need it. And the image will still use up most of the 2nd drive, cuz it will be so big, cuz you don't make partitions to segment data. This makes no sense.
Now let's say we use the Radboy method make a nice, small 3GB partition at the leading edge of the drive, which offers us the best performance. 3GB's is plenty of space for Windows & all your apps, which includes all settings and system configs, excluding games (which go on another partition). This means, when you restore this image (if you don't know), you don't have to re-install all the apps & re-config the system like you indicated in an earlier post. Ghost offers different compression settings, but say I have 2.5 gigs of data (about what I have). Using 'FAST' compression setting will result in an image of ~1.5 gigs. Lots better than the ~30 gig image your method would yield. Your way would have us imaging everything on the drive - MP3s, jpegs, movies, etc. .. *if* at all (your method would require a burner and a stack of disks .. or a separate hard drive, which would be mostly used by the image itself).
So, as you can plainly see I'm *not* (as you say) "imaging practically the entire drive" .. am I? O the contrary, *your* method will image practically the entire drive.
Let me also add here that, of the 10 or so times I've had to restore images, never once was it b-cuz the drive died. Every time was cuz something in Windows got screwed up .. like a new version of DX didn't work in my system, or Windows just wouldn't boot from some reason. In everyone of these cases, it was okay to have the image on the same drive (different partition, tho). But if the drive were to die, then I'd lose the image, too. But this has never happened to me (knock on wood).
I actually store my images on different hard drives from the one with the partition I created them from, but not everyone has multiple drives like me. And not everyone needs another hard drive to receive the image .. unless they follow your partitioning scheme.
You say, "The person who simply formats and reinstalls everything from a single-partition image is going to be up and running just as fast as you". It takes me less than 10 minutes to be up & running with a restored Ghost image, whereas it takes me the better part of a week to "reinstalls everything from a single-partition" (I have a *lot* of apps).
You say, "We've already explained .. why no benchmarks have ever shown a performance benefit in real world usage and why it might actually hinder performance. So you guys can drop it now."
Maybe you've never heard of HD Tach, which clearly shows HERE that performance is better at the leading edge of the drive. Anyone with eyes can see this is true. Do you not see that thru-puts at the leading edge are ~40MB/s, and only 18MB/s at the end of the disk? This is even a graph of a 45GB drive - almost identical to what the original poster asked about. This is why disk optimazation utilities, like Norton, place the swap file at the leading edge of the drive .. where performance is best. Maybe you should send a note to the folks at Norton and tell them they're doing it wrong.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter