Is FAT32 slower than FAT16?
hhl...@sprintmail.com
clean install of Win98. Now, I feel the system is running slower especially
when I access big files. Western Digital claims that the system will run
slower with FAT32 than FAT16 because of smaller clusters. Is that true? Does
anybody have the same experience with Seagate or Matrox hard drive?
Howard Lung
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum
Ron Reaugh
hhl...@sprintmail.com wrote in article <6pkc1e$f1l$1...@nnrp1.dejanews.com>...
> I formated my new 6.4G Western Digital hard drive to FAT32 format and did
> clean install of Win98. Now, I feel the system is running slower
especially
> when I access big files.
Yep. Using "FORMAT x: /Z:16" would make it fast again but the wasted
cluster slop would also increase.
> Western Digital claims that the system will run
> slower with FAT32 than FAT16 because of smaller clusters. Is that true?
Yep. Using the above format command will make the clusters bigger and
things are back fast again.
Tim Moore
below, or just a bit larger than the cluster size and there are a lot of
them. The tables in Partition Magic that tell you how much is wasted
are true only if you hit the above parameters. The larger average file
size, the smaller the relative waste.
For example, FAT16 1 GB partition is 16 KB, 2 GB is 32 KB. If average
file size is 8 KB, the rest of the cluster (8 or 24 KB in this case) is
unused. If average file size is 150 KB, average waste is 6 or 22 KB per
file. In a FAT32 2 GB partition at 4 KB clusters a 150 KB file would
waste 2 KB/file, or 6 KB/file if 8 KB clusters.
The trade off is how many clusters need to be tracked, allocated,
deallocated etc. A 2 GB partition has 65505 clusters to track for 32
KB, 262020 for 8 KB and half a million for 4 KB.
W98 is a year late from its original target date. It has the
outstanding features of rearranging your disk for frequently used files,
a slower file system and an integrated web interface, all of which slow
down your machine.
A keen observer will realize the frequency/load rearranging was
developed as an answer to slow disk controller subsystems and
processor/bus speeds, basically predating PIO4/UDMA, FW SCSI and
5400+RPM drives.
FAT32 might have been a good idea to save space when disk drives of
either religion were considered huge at 1.2GB, but when $250 gets 6GB,
do you really care if you lose 10 MB to cluster slop? It's 1/600 of the
drive (or the waste for 1,747,626 of our average 150 KB files).
As for the seamless desktop/internet interface, personally I don't want
any application constantly running in the background that can maintain
an open connection to the Internet. There are too many examples of
downloading bugs, silent uploading of information about you and your
machine, automatic driver upgrades that are unwanted or break other
things. Do you like the idea of corporate entities having an open
window into your machine (eg- personal life)? Won't mention the memory
overhead.
But then, you did willingly give MS another $90, and Intel gets to sell
you more CPU's, and Toshiba/Hitachi et al get to sell you more memory.
Too bad what you actually got was a slower pile of year-old, not really
useful shit in return.
Thank you for you continued support.
[nothing personal to the repliee, BTW]
> > I formated my new 6.4G Western Digital hard drive to FAT32 format and did
> > clean install of Win98. Now, I feel the system is running slower
> especially when I access big files.
>
> Yep. Using "FORMAT x: /Z:16" would make it fast again but the wasted
> cluster slop would also increase.
>
> > Western Digital claims that the system will run
> > slower with FAT32 than FAT16 because of smaller clusters. Is that true?
>
> Yep. Using the above format command will make the clusters bigger and
> things are back fast again.
--
[Replies: xxx -> ]
"Everything is permitted. Nothing is forbidden."
WS Burroughs.
Ron Reaugh
Tim Moore <timothy...@bigfoot.com> wrote in article
<35BEB1F5...@bigfoot.com>...
> Larger clusters only "waste" space when your average file size is either
> below, or just a bit larger than the cluster size and there are a lot of
> them. The tables in Partition Magic that tell you how much is wasted
> are true only if you hit the above parameters. The larger average file
> size, the smaller the relative waste.
>
> For example, FAT16 1 GB partition is 16 KB, 2 GB is 32 KB. If average
> file size is 8 KB, the rest of the cluster (8 or 24 KB in this case) is
> unused. If average file size is 150 KB, average waste is 6 or 22 KB per
> file. In a FAT32 2 GB partition at 4 KB clusters a 150 KB file would
> waste 2 KB/file, or 6 KB/file if 8 KB clusters.
>
> The trade off is how many clusters need to be tracked, allocated,
> deallocated etc. A 2 GB partition has 65505 clusters to track for 32
> KB, 262020 for 8 KB and half a million for 4 KB.
Yep, the issue is speed vs space efficiency.
> W98 is a year late from its original target date. It has the
> outstanding features of rearranging your disk for frequently used files,
> a slower file system
No, it's a little faster and more robust.
> and an integrated web interface, all of which slow
> down your machine.
False.
> A keen observer will realize the frequency/load rearranging was
> developed as an answer to slow disk controller subsystems and
> processor/bus speeds, basically predating PIO4/UDMA, FW SCSI and
> 5400+RPM drives.
Nonsense. It was developed to reduce seek overhead which is purely a drive
phenomena and is independent of SCSI vs UDMA.
> FAT32 might have been a good idea to save space when disk drives of
> either religion were considered huge at 1.2GB, but when $250 gets 6GB,
> do you really care if you lose 10 MB to cluster slop? It's 1/600 of the
> drive (or the waste for 1,747,626 of our average 150 KB files).
Nonsense. FAT32 provides support for the 10, 14 16GB drives that are now
available.
Rafal
measure difference with benchmarks. When they noticed this they fixed the
problem. They slowed down FAT16! Now they are almost equal.
Ron Reaugh
slightly faster. See: www.storagereview.com
Ron Reaugh
Rafal <ra...@rafal.com> wrote in article <uZFdBb2u9GA.263@upnetnews05>...
Nonsense. FAT32 was never slower than FAT16. Disk I/O processing is
slower on a volume with more clusters to handle. So, a 4K cluster FAT32
partitions will be slower than a FAT16 partition of the same size using
larger clusters. The FAT32 partition will be much more space efficient.
If one formats the FAT32 partition using :FORMAT x: /Z:n" where n is the
cluster size in sectors and makes the FAT32 partition have big clusters
then it's as fast and space inefficient as FAT16.
Rafal
http://sysdoc.pair.com/hdd.html#File Systems :
"There is a noticeable decrease in performance with FAT32 drives compared to
FAT16 drives! I first noticed this when I defraged my harddisk - the
operation took much longer than usual. I wanted to be sure, so I tested the
harddisk with Winbench. The DiskWinmark96 for an EIDE harddisk decreased by
about 3% with Windows95 build 1111 FAT32 vs. FAT16. Under Windows95 build
1111 with FAT32 compared to Windows95 build 950a with a FAT16-drive of same
size, the DiskWinmark96 decreased about 6%. The accessing of many small
files was recognizably slower. Not only FAT32 is slower (which is due to the
larger amount of clusters), the new build of Windows95 is slower with
harddisk access in general! This doesn't mean that FAT32 is necessarily
slower, this means that Microsoft did something with harddisk access in
their new Windows95 OSR 2 that decreases harddisk performance in some way.
It is not the EIDE driver; I also measured and saw performance decrease with
other EIDE-drivers."
It looks more detailed than your "Nonsense!".
Rafal
This is full:
"There is a noticeable decrease in performance with FAT32 drives compared to
FAT16 drives! I first noticed this when I defraged my harddisk - the
operation took much longer than usual. I wanted to be sure, so I tested the
harddisk with Winbench. The DiskWinmark96 for an EIDE harddisk decreased by
about 3% with Windows95 build 1111 FAT32 vs. FAT16. Under Windows95 build
1111 with FAT32 compared to Windows95 build 950a with a FAT16-drive of same
size, the DiskWinmark96 decreased about 6%. The accessing of many small
files was recognizably slower. Not only FAT32 is slower (which is due to the
larger amount of clusters), the new build of Windows95 is slower with
harddisk access in general! This doesn't mean that FAT32 is necessarily
slower, this means that Microsoft did something with harddisk access in
their new Windows95 OSR 2 that decreases harddisk performance in some way.
It is not the EIDE driver; I also measured and saw performance decrease with
other EIDE-drivers.
The application benchmark Winstone96 wasn't affected very much by this
behavior. I always measured lower values with the newer build and also with
FAT32, but that could have been from measuring errors. The difference was
too small to say it is affected. Unfortunately, the BAPCo doesn't run on the
new build of Windows95, but I guess it also won't show much difference in
performance. That is due to the very small impact of the harddisk on these
benchmarks. It would play a greater role if you only had 8 MB RAM.
Conclusion: If you can, use build 950 (version 950a) of Windows95 and use
partitions close to 512 MBytes with 8 KByte clusters. If you have build 1111
(version 950b) of Windows95 you can also use FAT32 drives with 4 KByte
clusters to save space, especially with bigger partitions. Or to get rid of
all these problems use another operating system which uses a different file
system ;) "
John Welter
North West Group
Badbri wrote in message <35BF61C3...@europa.com>...Ron Reaugh wrote:
Tim Moore <timothy...@bigfoot.com> wrote in article
> W98 is a year late from its original target date. It has the
> outstanding features of rearranging your disk for frequently used files,
> a slower file systemNo, it's a little faster and more robust.
> and an integrated web interface, all of which slow
> down your machine.False.
Sorry Ron but I have done my own tests with OSR 2.1 and Win98final using the same two machines:
P. Pro 200 with SCSI +128MB ram 4Gb w/Fat32 and an
MMX Intel 166 with IDE +64Mb ram 2Gb w/Fat32
and no matter how I play with Windows 98 it boots slower
and runs slower. W98 took 3 (yes 3) times the hard disk space
as OSR 2.1 when the same things are installed without any other
applications (just the Windows OS with IE 4). IE 4.01 installed
into OSR 2.1 ran faster than IE 4.01.in Win98 where it is
included/integrated. These tests I conducted were witnessed by
two other Small business owners that wanted to see the differance
between the two OS. And both declined to upgrade to Win98 on
their company machines because of the "sluggishness" and
"excess disk waste" that Win98 creates.
I really hoped that I could get the contracts to upgrade their
machines ( $$$ :) but couldn't persuade them after the tests were
completed. So no one could ever persuade me to believe that
Win98 is a faster more efficient OS than Win95. IE 4 runs
faster with less pukes in OSR 2 than the Win98 integrated IE4.P.S. Your other hard disk info is right on key.
Stanzen
using HPFS many years ago. The performance gain from formatting Fat16
to HPFS was
amazing. My statistical packages ran two to three times as fast for
disk intensive
analyses in HPFS and HPFS supposed didn't fragment the drive as much as
Fat. HPFS was an early, superior disk format, running under OS/2.
Microsoft allowed NT to read HPFS drives because many business servers
were running OS/2. Microsoft never permitted that capacity in W95 or
W98. Microsoft has always opted for proprietary software for market
domination, sacrificing efficiency.
I'm glad that in America, a multi-billion dollar software company can
finally write a disk format that isn't designed for a 8086 processor
running a 400 Mb Hard drive,
15 years after. {A note of sarcasm}
John Welter wrote:
> I'd rather believe the countless other tests I've seen showing 5% to
> 10% better speed with W98 and my own experience on more then 1500 PC's
> that were upgraded to W98 then your tests. W98 is faster, more stable,
> and contains better support for current day hardware then any prior
> version. Nobody can claim that running a 3 year old OS with a bunch
> of patches ontop of it can be as efficient as a clean version of code
> with these changes properly made. When NTFS for NT first came out
> people crapped all over it saying it was slower then FAT, etc, etc.
> Well, funny thing happened - it still took off and was widely used.
> It's now faster then FAT16 in most applications. FAT16 is a dead
> end. Microsoft won't continue to develop and fine tune that code.
> FAT32 is a rising star and MS will continue to optimize it for some
> time. Within a couple rev's FAT32 will easily outperform FAT16 across
> the board. Shortly after that expect to see FAT16 support dropped
> with an automatic conversion process to FAT32 or NTFS mandatory. The
> other thing to consider is reliability. FAT16 was a crude file system
> written for small disks - originally floppies (FAT12). It's obsolete
> in todays world. The added reliability of FAT32 is much more
> important then any perceived performance improvements with FAT16. By
> the end of this decade we'll be seeing 20Gb disks as common place -
> FAT16 isn't up to the task. John Welter, MCSE+ISystems Engineer
--
* Remove NOSPAM from email address to reply to this message*
GA
detail), yet it sounds like you're trying to disagree with him???
Rafal wrote in message <#vHowU8u9GA.315@upnetnews05>...
>I am just trying to read different sources.
>
>http://sysdoc.pair.com/hdd.html#File Systems :
>
>"There is a noticeable decrease in performance with FAT32 drives compared
to
>FAT16 drives! I first noticed this when I defraged my harddisk - the
>operation took much longer than usual. I wanted to be sure, so I tested the
>harddisk with Winbench. The DiskWinmark96 for an EIDE harddisk decreased by
>about 3% with Windows95 build 1111 FAT32 vs. FAT16. Under Windows95 build
>1111 with FAT32 compared to Windows95 build 950a with a FAT16-drive of same
>size, the DiskWinmark96 decreased about 6%. The accessing of many small
>files was recognizably slower. Not only FAT32 is slower (which is due to
the
>larger amount of clusters), the new build of Windows95 is slower with
>harddisk access in general! This doesn't mean that FAT32 is necessarily
>slower, this means that Microsoft did something with harddisk access in
>their new Windows95 OSR 2 that decreases harddisk performance in some way.
>It is not the EIDE driver; I also measured and saw performance decrease
with
>other EIDE-drivers."
>
Ron Reaugh
Rafal <ra...@rafal.com> wrote in article <#vHowU8u9GA.315@upnetnews05>...
> I am just trying to read different sources.
>
> http://sysdoc.pair.com/hdd.html#File Systems :
Winbench96 and build 1111. The page is very old and inaccurate. It's
build 1116 and Winbench98 now.
FAT32 is just as fast as FAT16 at the same cluster and partition size.
OSR2 + REMIDEUP.EXE is just as fast with FAT16 as is W95[a]. W98 is about
the same speed or a little faster.
Ron Reaugh
John Welter <jwe...@nwgeo.com> wrote in article
<35c08...@news.oanet.com>...
>I'd rather believe the countless other tests I've seen showing 5% to 10%
>better speed with W98 and my own experience on more then 1500 PC's >that
were upgraded to W98 then your tests.
>W98 is faster, more stable, and contains better support for current day
>hardware then any prior version. Nobody can claim that running a 3 >year
old OS with a bunch of patches ontop of it can be as efficient as a >clean
version of code with these changes properly made.
The above is prima facie. Why some must have the obvious explained in
detail is beyond me. Thanks for doing so.
Terry R.
>I am just trying to read different sources.
>
>http://sysdoc.pair.com/hdd.html#File Systems :
>
>FAT16 drives! I first noticed this when I defraged my harddisk - the
>operation took much longer than usual. I wanted to be sure, so I tested the
>harddisk with Winbench. The DiskWinmark96 for an EIDE harddisk decreased by
>about 3% with Windows95 build 1111 FAT32 vs. FAT16. Under Windows95 build
>1111 with FAT32 compared to Windows95 build 950a with a FAT16-drive of same
>size, the DiskWinmark96 decreased about 6%. The accessing of many small
>files was recognizably slower. Not only FAT32 is slower (which is due to the
>larger amount of clusters), the new build of Windows95 is slower with
>harddisk access in general! This doesn't mean that FAT32 is necessarily
>slower, this means that Microsoft did something with harddisk access in
>their new Windows95 OSR 2 that decreases harddisk performance in some way.
>It is not the EIDE driver; I also measured and saw performance decrease with
>other EIDE-drivers."
FAT32 is somewhat slower than FAT16, particularly with sequential write
operations. It is usually not more than 5% slower though. The advantage of
FAT32 was never intended to be better performance but rather better
efficiency. By using a smaller cluster size for large partitions you
increase the storage efficiency of the drive (i.e. less wasted space).
For example with FAT16 a >1Gb partition would be only about 75% efficient
when using a 32K cluster size, but a >1Gb partition with FAT32 and a 4K
cluster size is about 97% efficient.
Rafal
slowed down disk access in general.
Rafal
the best I can. I think that it is nice to know about some things especially
when company is not telling about it. Just to know more. I am not trying to
tell that FAT32 is not good. I am using only FAT32 (I'd love to see NTFS
support under Win98).
Ron Reaugh
Rafal <Raf...@email.msn.com> wrote in article
<ur04pcJv9GA.113@upnetnews05>...
> The key is that original FAT16 was faster. Microsoft did something what
> slowed down disk access in general.
That is flat not true. See: www.storagereview.com
Rafal
Ron Reaugh wrote in message <01bdbca1$ad9324a0$4915410c@p133>...
Could you send me this article. I checked this link but only found HDDs
benchmarks. If you found something interesting I'll appreciate.
Ron Reaugh
Rafal <ra...@rafal.com> wrote in article <u4TYdhPv9GA.77@upnetnews05>...
www.storagereview.com "Features" "Windows 95 vs Windows 98: A
Perspective From Disk Performance "
Rafal
Slash wrote in message <35c7fc37...@nntp.alaska.net>...
>On Fri, 31 Jul 1998 10:23:15 -0500, "Rafal" <Raf...@email.msn.com>
>wrote:
>
>>The key is that original FAT16 was faster. Microsoft did something what
>>slowed down disk access in general.
>>
>
>Not microsoft, per se, just simple numbers. With 4k clusters on an 8
>gig drive, you've got literally millions of them versus just over
>260,000 for normal 32k FAT16 clusters (multiple 2.1GB partitions).
>That's a lot of overhead.
>
Steven Mercer
Badbri wrote:
Yes on big drives Fat32 is slower than Fat16.
http://www.winmag.com/win98/5.htmAs far as your Win98 concerns:
Gee John read Ziff-Davis tests tests too !!!
http://www.zdnet.com/products/content/newrev/pcmg/win98test/overall.html
and they didn't find any "performance gain" from Win98.
Here is a "non-partial" review test setup:
http://users.dwx.com/hellmanm/
Most recommend a "new install" and not an upgrade
because of the buggs.. HD probs, games not working, etc.Of course those that "poo-poo" Win98 don't get paid by or
make much money from Microsoft products so they can be
more truthful in what it's really like to move to Win98.Hey John is that a Microsft Cert near your name ???
Your a Microsoft Certified Engineer ?? Well of course
you'll say Win98 is the greatest, fastest best thing to happen.
You make money from Microsoft products and I bet you love
those Win98 problems cause it justifies your salary !!!Why all the questions about hardware problems under Win98,
just look at this Asus newsgroup ??
You ought to check other groups too, there are lotts of buggy
issues from fairly knowledgable people that can't get things
to work properly. Here is site that tracks Win98 annoyances:
http://www.annoyances.org/win98/_index.htmlI like the idea of paying MS $90 for a big Win95 service
pak that still doesn't fix half the stuff it should and causes
more problems than it fixes.........................
and I personally bought over $5000 of MS software in the
last 3 years but won't touch Win98 until a service pak is out.
I stick with NT 4.0 and Win95 OSR 2.1 and so will most
business that need a reliable OS.John Welter wrote:
I'd rather believe the countless other tests I've seen showing 5% to 10% better speed with W98 and my own experience on more then 1500 PC's that were upgraded to W98 then your tests. W98 is faster, more stable, and contains better support for current day hardware then any prior version. Nobody can claim that running a 3 year old OS with a bunch of patches ontop of it can be as efficient as a clean version of code with these changes properly made. When NTFS for NT first came out people crapped all over it saying it was slower then FAT, etc, etc. Well, funny thing happened - it still took off and was widely used. It's now faster then FAT16 in most applications. FAT16 is a dead end. Microsoft won't continue to develop and fine tune that code. FAT32 is a rising star and MS will continue to optimize it for some time. Within a couple rev's FAT32 will easily outperform FAT16 across the board. Shortly after that expect to see FAT16 support dropped with an automatic conversion process to FAT32 or NTFS mandatory. The other thing to consider is reliability. FAT16 was a crude file system written for small disks - originally floppies (FAT12). It's obsolete in todays world. The added reliability of FAT32 is much more important then any perceived performance improvements with FAT16. By the end of this decade we'll be seeing 20Gb disks as common place - FAT16 isn't up to the task. John Welter, MCSE+ISystems Engineer
RodneyIII
FAT32,FAT16 comparison starts on pg. 123 and has different scenarios
of test environments. It has some good info. Check it out on the
web, http://www.zdnet.com/pcmag/features/win98_tests/test1.html and
follow the links for the charts of tests.
On Tue, 28 Jul 1998 11:18:06 GMT, hhl...@sprintmail.com wrote:
>I formated my new 6.4G Western Digital hard drive to FAT32 format and did
>clean install of Win98. Now, I feel the system is running slower especially
Rafal
Win95A or first build of Win95 OSR2, and later versions of Win9x. But it is
not so important now. Just do not yell on me when I am answering. Just read
very carefull and than, if it doesn't interest you, just ignore.