Poster | Thread |
T_Bone
|  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 10:43:50
| | [ #21 ] |
|
|
 |
Elite Member  |
Joined: 11-Sep-2003 Posts: 3043
From: here To: there | | |
|
| @broadblues
Quote:
broadblues wrote:
I wonder what else I'll find if I actually read the manual!
|
You can learn more about Amiga filesystems by reading the Filesystem Wars in CSAA from a few years back (where SG argues Amiga filesystems are superior to everything on the planet) than you can from any manual. _________________ "If you want to tell people the truth, make them laugh, otherwise they'll kill you." - Oscar Wilde |
|
Status: Offline |
|
|
poweramiga2002
|  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 10:58:43
| | [ #22 ] |
|
|
 |
Super Member  |
Joined: 29-Jul-2003 Posts: 1389
From: Flinders NSW Australia | | |
|
| @Mopemen
Just did a test here with SFS 1.268 and moved 990meg in 132 seconds from DH1 to DH4
is that good speed or still slow ? |
|
Status: Offline |
|
|
OldFart
|  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 11:32:06
| | [ #23 ] |
|
|
 |
Elite Member  |
Joined: 12-Sep-2004 Posts: 3071
From: Stad; en d'r is moar ain stad en da's Stad. Makkelk zat! | | |
|
| @broadblues
Quote:
I wonder what else I'll find if I actually read the manual! |
Reading the manual? Haven't you got something more pressing to do? Tsk, tsk, tsk, reading a manual. It's probably the last thing I would do!
OldFart_________________ Life is a waste of time. Time is a waste of life. Get wasted all the time and you'll have the time of your life! |
|
Status: Offline |
|
|
OldFart
|  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 11:46:32
| | [ #24 ] |
|
|
 |
Elite Member  |
Joined: 12-Sep-2004 Posts: 3071
From: Stad; en d'r is moar ain stad en da's Stad. Makkelk zat! | | |
|
| @Mopemen
Here's my partition table:
Unit Size Used Free Full Errors Status Name File System CD0: 257M 132.095 0 100% 0 Read Only ADCD_2.1 DH0: 255M 82.319 179.345 31% 0 Read Only OS_4.0 r2 FFS DH1: 255M 82.996 178.668 32% 0 Read Only OS_4.0 r3 FFS DH2: 255M 93.128 168.536 36% 0 Read Only OS_4.0 r4 FFS DH3: 255M 182.500 79.164 70% 0 Read/Write ExtendedWorkBench FFS DH4: 2.047M 538.200 510.136 51% 0 Read/Write Applicaties FFS DH5: 2.047M 163.570 884.766 16% 0 Read/Write Projecten FFS DH6: 3.071M 736.017 836.607 47% 0 Read/Write Games FFS DH7: 100M 91.581 113.283 45% 0 Read/Write WebServer SFS DH8: 6.000M 2.552.772 3.591.260 42% 0 Read/Write Extracts FFS DH9: 4.000M 2.306.182 1.789.850 56% 0 Read/Write Archives FFS DH10: 4.000M 328.595 3.767.437 8% 0 Read/Write UAE FFS DH11: 1.023M 91.813 2.004.379 4% 0 Read/Write OldWebServer_2 SFS DH13: 127M 247.994 13.702 95% 0 Read Only OS_4.0 Final SFS ENV: 110K 110 0 100% 0 Read/Write Env ICD0: No disk present RAM: 145K 145 0 100% 0 Read/Write RAM Disk TEXTCLIP: 116M 69 122.388.480 0% 0 Read/Write TextClip ZIP3-2-64: No disk present
The partitions add up to some 20Gb on a 120Gb HD thereby leaving some 100Gb for some extensions.
With the advent of OS4.0 Final I moved over to SFS and for some partitions that were created later I also used SFS as the File Sytem of choice. It generally feels faster than its FFS counterpart. Stability is *NOT* an issue at all, due to the FS being in continuous and active developement for a great many years now.
OldFart
EDIT: This really looks awfull doesn't it? Which tags are available for formatting? EDIT: After a pair of <tPRE> & </PRE> tags it looks much better.Last edited by OldFart on 20-Apr-2007 at 11:57 AM. Last edited by OldFart on 20-Apr-2007 at 11:56 AM. Last edited by OldFart on 20-Apr-2007 at 11:54 AM. Last edited by OldFart on 20-Apr-2007 at 11:53 AM.
_________________ Life is a waste of time. Time is a waste of life. Get wasted all the time and you'll have the time of your life! |
|
Status: Offline |
|
|
Chip
|  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 12:34:50
| | [ #25 ] |
|
|
 |
Cult Member  |
Joined: 4-Mar-2005 Posts: 574
From: Budapest, Hungary | | |
|
| @Micam
Quote:
Can someony please tell my the ups and downs with SFS? |
Contra: No recovery tool for SFS Pro: Damn fast, rock stable.
During betatesting OS4 I had no data loss because of SFS at all. Since I discovered its speed I've totally dropped FFS from my HDD.
|
|
Status: Offline |
|
|
Hans
|  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 12:49:08
| | [ #26 ] |
|
|
 |
Elite Member  |
Joined: 27-Dec-2003 Posts: 5123
From: New Zealand | | |
|
| @sundown
I just looked and my SFS partitions have a block size of 1024, buffers = 80 and maxtransfer = 1FE00. Can the buffers and mactransfer be changed without wiping data on the partition? In particular, I think that increasing the buffers might increase the copy rate. You're system has a significntly faster copy rate and the filesystem settings seem to be the major difference (although I have an A1-XE and am using an sii 680 controller).
Hans
_________________ Join the Kea Campus - upgrade your skills; support my work; enjoy the Amiga corner. https://keasigmadelta.com/ - see more of my work |
|
Status: Offline |
|
|
Mopemen
 |  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 13:00:37
| | [ #27 ] |
|
|
 |
Regular Member  |
Joined: 7-Feb-2005 Posts: 147
From: The Netherlands | | |
|
| @OldFart
Wow, it's quite a list!
What do you use for a buffer size? _________________ µA1c OS4.1 (with a dead powersupply (powersupplied)) A500 |
|
Status: Offline |
|
|
poweramiga2002
|  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 13:09:59
| | [ #28 ] |
|
|
 |
Super Member  |
Joined: 29-Jul-2003 Posts: 1389
From: Flinders NSW Australia | | |
|
| @Hans
I have my buffers set @ 8000 |
|
Status: Offline |
|
|
Framiga
|  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 13:17:32
| | [ #29 ] |
|
|
 |
Elite Member  |
Joined: 5-Jul-2003 Posts: 2214
From: Unknown | | |
|
| @poweramiga2002
thats a good amount for SFS.
_________________
|
|
Status: Offline |
|
|
ChrisH
 |  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 14:03:41
| | [ #30 ] |
|
|
 |
Elite Member  |
Joined: 30-Jan-2005 Posts: 6679
From: Unknown | | |
|
| @poweramiga2002 That's 7.5MB/second - not blazingly fast, but since you're actually doing data transfer between two partition on the same HD (?), it's really 15MB/second. Which is better - still not awesome, but possibly as much as you can realistically expect from your machine I guess. _________________ Author of the PortablE programming language. It is pitch black. You are likely to be eaten by a grue... |
|
Status: Offline |
|
|
ChrisH
 |  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 14:08:47
| | [ #31 ] |
|
|
 |
Elite Member  |
Joined: 30-Jan-2005 Posts: 6679
From: Unknown | | |
|
| @OldFart Looking at your 12 partitions I wondered if you'd ever heard of folders? 
Backing-up that system must be a nightmare.... or perhaps you don't do backups? 
My own partitioning goes something like this, and I consider that slightly over-the-top: Workbench: Data: Programs: Games: Temp: WorkbenchBackup:
All SFS of course, albiet under WinUAE (using real partitions, not them stupid file "partitions"). Only the first 4 need backing up. _________________ Author of the PortablE programming language. It is pitch black. You are likely to be eaten by a grue... |
|
Status: Offline |
|
|
OldFart
|  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 14:59:45
| | [ #32 ] |
|
|
 |
Elite Member  |
Joined: 12-Sep-2004 Posts: 3071
From: Stad; en d'r is moar ain stad en da's Stad. Makkelk zat! | | |
|
| @Mopemen
Quote:
What do you use for a buffer size? |
Not much really: my wife has cup B.
Furthermore FFS uses 2048 bytes per block and SFS the recommended 512 bytes.
OldFart
_________________ Life is a waste of time. Time is a waste of life. Get wasted all the time and you'll have the time of your life! |
|
Status: Offline |
|
|
OldFart
|  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 16:05:44
| | [ #33 ] |
|
|
 |
Elite Member  |
Joined: 12-Sep-2004 Posts: 3071
From: Stad; en d'r is moar ain stad en da's Stad. Makkelk zat! | | |
|
| @ChrisH
Quote:
Looking at your 12 partitions I wondered if you'd ever heard of folders? |
Yeah, sure! I've even got some on the Windows box. But you mean drawers? Generically named 'directories'? How would you put 4 different boot partitions into one and still keep the separately bootable? Btw, your question reminds me that some time ago I promised to describe me setup in more detail.
Quote:
Backing-up that system must be a nightmare.... |
Backing-up? What's that?
Quote:
... or perhaps you don't do backups? |
Ah! You figured it out, I see.
For backing up I use ZIP100's, but unfortunately a SCSI-ZIP drive is currently only mounted as Read-Only. Don't know whether it's Stephane Guillard or somebody else addressing this prob, but until that time: sorry, no BU's today!
I really should consider backing up on a more frequent base than the once or twice that it is applied the previous 10 years on the A1200. The last time I did that was when I upgraded from OS3.5 to OS3.9.
OldFart_________________ Life is a waste of time. Time is a waste of life. Get wasted all the time and you'll have the time of your life! |
|
Status: Offline |
|
|
Mopemen
 |  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 16:06:07
| | [ #34 ] |
|
|
 |
Regular Member  |
Joined: 7-Feb-2005 Posts: 147
From: The Netherlands | | |
|
| The SFS partitions were: Blocksize: 512 Buffers: 600 MaxTransfer: 1FE00 Mask: FFFFFFFE
I changed them to: Blocksize: 512 Buffers: 10000 MaxTransfer: FFFFFFF Mask: FFFFFFFE
Then I did the same test but the results are exactly the same. I also tried 8000 buffers.
_________________ µA1c OS4.1 (with a dead powersupply (powersupplied)) A500 |
|
Status: Offline |
|
|
TetiSoft
|  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 17:21:08
| | [ #35 ] |
|
|
 |
Cult Member  |
Joined: 3-Mar-2005 Posts: 585
From: Germany | | |
|
| @Hans
Quote:
Can the buffers and mactransfer be changed without wiping data on the partition?
|
Yes. Media Toolbox will warn you when you try to save a modified config which would destroy the contents of one or more partitions. You'll see such a warning after you changed the blocksize for example.
IIRC the OS4 filesystems ignore MaxTransfer anyway, just because there exists so much broken hardware which doesnt work with transfers >64K because it was never tested outside Windows :)
|
|
Status: Offline |
|
|
olsen
|  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 18:04:14
| | [ #36 ] |
|
|
 |
Cult Member  |
Joined: 15-Aug-2004 Posts: 774
From: Germany | | |
|
| @Samwel
Quote:
Samwel wrote: @Mopemen
128 seconds for two 300MB files???? Damn slow! A PC with Windows would probably do it in less than 15 seconds, not more than 20 seconds anyway.
|
(groan) A PC with Windows doesn't FFS. One of the things that make the FFS slow is that you can't both have crash-safe write operations and fast disk access. I think I've explained this a couple of times. The FFS reimplementation has to put up with a lot of baggage that came with the original data structures, and every single shortcoming follows from this history. |
|
Status: Offline |
|
|
AlexC
|  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 18:07:55
| | [ #37 ] |
|
|
 |
Super Member  |
Joined: 22-Jan-2004 Posts: 1301
From: City of Lost Angels, California. | | |
|
| @Hans
>Can the buffers and maxtransfer be changed without wiping data on the partition?
Yes, no problem there but keep in mind that the buffers are for the dirs/files list caching, not data transfer.
Just don't alter the block size (though if you did, after rebooting or mounting the partition would show up as invalid and you could just go back and revert the block size without data loss).
_________________ AlexC's free OS4 software collection
 AmigaOne XE/X1000/X5000/UAE-PPC OS4 laptop/X-10 Home Automation |
|
Status: Offline |
|
|
sundown
 |  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 19:46:35
| | [ #38 ] |
|
|
 |
Elite Member  |
Joined: 30-Aug-2003 Posts: 5120
From: Right here... | | |
|
| @Mopemen
Quote:
The bigger the better? Or is a very big buffer not advisable? |
Going from 100 to a 500 buffer size will show an increase in speed, increasing to 1000 will show very little gain. I see you have them set to 600, which is just right. Buffers use memory, blocksizeXbuffers, so if you have a lot of partitions, you could run low on memory for other apps. Everything's a trade off & the same situation applies to FFS as well._________________ Hate tends to make you look stupid... |
|
Status: Offline |
|
|
sundown
 |  |
Re: FFS VS SFS speed test OS4 Posted on 20-Apr-2007 20:39:29
| | [ #39 ] |
|
|
 |
Elite Member  |
Joined: 30-Aug-2003 Posts: 5120
From: Right here... | | |
|
| @Hans
A block size of 512 vs 1024 will have the same speed, though 512 is recommended (SFS). Maxtransfer setting will not speed anything up as well. The buffer setting is way to low & will make a big difference in speed if you increase it to 500-1000. Just remember, the more buffers, the more free memory will be used. UDMA settings affect speed & a sii680 should be just as fast as a micro. I have both HD on the primary port as master & slave, my DVD burner is on the secondary port as master. If you only have 1 HD & 1 CDROM, it might help to have both set as masters, 1 on each port.
You can test speed vs buffers in a shell. Addbuffers DHx: will tell you how many buffers are set, +/-#buffers will add or subtract buffers to test copy speed. Once you find the magic number, then set the buffer size in Media ToolBox for all partitions. _________________ Hate tends to make you look stupid... |
|
Status: Offline |
|
|
ChrisH
 |  |
Re: FFS VS SFS speed test OS4 Posted on 22-Apr-2007 9:15:29
| | [ #40 ] |
|
|
 |
Elite Member  |
Joined: 30-Jan-2005 Posts: 6679
From: Unknown | | |
|
| @OldFart who said Quote:
How would you put 4 different boot partitions into one and still keep the separately bootable? |
I wasn't suggesting that, although it's quite possible - I already do that for OS3.1-OS3.9 thanks to TrueMultiAssign & a *lot* of hacking... (I did once describe my system on these forums, I think)
Even ignoring the different versions of OS4, you have 9-10 partitions, which is basically obscene 
P.S. If you don't backup, be prepared to loose everything when you least expect it... Perhaps a HD failure, or worse theft, or else just plain stupidity/carelessness  Last edited by ChrisH on 22-Apr-2007 at 09:16 AM.
_________________ Author of the PortablE programming language. It is pitch black. You are likely to be eaten by a grue... |
|
Status: Offline |
|
|