PFS3 All-in-One released

[jbenam]

Originally Posted by Toni Wilen

It can't be "fixed" because already existing buffer settings of 30 would become real buffer setting of 30 and someone will complain that it broke something because it suddenly become slower..

But then you will have users complain that PFS3 is somehow using too much RAM and then they would go back to FFS

Fixing the bug would just restore the normal behaviour of the setting. If someone wants more speed they should just set their buffers to MAX in the first place - it's not like you should expect super-fast speeds if you're setting just 30 buffers.

That's just my opinion, I think nobody would complain if it was plainly written in the changelog. But you're the big boss so it's up to you

Originally Posted by Toni Wilen

There is no need to repartition anything, just change the buffer value in hdtoolbox and save changes. It is 100% safe. (But only if you used hdtoolbox to partition it originally!)

I used HDInstTools when first partitioning. But I might have used HDToolBox from the 3.1 install disk after I discovered that 300 buffers were killing most of my free RAM.

So I should be able to just change them without losing data using HDInstTools, right?

[Toni Wilen]

Originally Posted by jbenam

But then you will have users complain that PFS3 is somehow using too much RAM and then they would go back to FFS

Fixing the bug would just restore the normal behaviour of the setting. If someone wants more speed they should just set their buffers to MAX in the first place - it's not like you should expect super-fast speeds if you're setting just 30 buffers.

But it isn't a bug. It is by design, it has been in PFS3 since ages ago.

So I should be able to just change them without losing data using HDInstTools, right?

No idea. I consider hdinsttools broken. (EDIT: because it does not handle RDB geometry info correctly)

[jbenam]

Originally Posted by Toni Wilen

But it isn't a bug. It is by design, it has been in PFS3 since ages ago.

I see - a bit misleading for the end user, but if it's by design then I can't complain.

I don't remember seeing anything about it in the readme, though!

Originally Posted by Toni Wilen

No idea. I consider hdinsttools broken. (EDIT: because it does not handle RDB geometry info correctly)

Never knew about that - I saw a lot of recommendations for it, especially for bigger drives and/or CF cards.

So the idea is to use OS3.9's HDToolBox, as it's the most updated around, or is the 3.1 one okay?

[daxb]

If PFS3aio is not differnt to original PFS3:
The PFS3 docs say buffers have to be between 70 and 500. PFS3 cannot work right below 70. So if a User use lower 70 he is resposible.

[Toni Wilen]

Originally Posted by daxb

If PFS3aio is not differnt to original PFS3:
The PFS3 docs say buffers have to be between 70 and 500. PFS3 cannot work right below 70. So if a User use lower 70 he is resposible.

PFS3 have following checks. My only change was to allow lower values (MIN_BUFFERS to 10 from 70) because small values do work fine (perhaps it was problem with very old versions) in normal usage and allow much lower memory usage. PFS3 will show error requester if it runs out of buffers and so far I haven't got any reports. (Technically this can be changed to automatically allocate some buffers more when needed)

	/* sanity checks. If HDToolbox default of 30, then 150,
	 * otherwise round in range 70 -- 600
	 */
	if (i==30) i=150;
	if (i<MIN_BUFFERS) i = MIN_BUFFERS;
	if (i>MAX_BUFFERS) i = MAX_BUFFERS;

EDIT: AFAIK only 30 and 50 are hdtoolbox defaults. 50 is already large enough, 50 vs 70 should not make any noticeably difference but 30 vs 70 is a bit too large. Any other value means user manually edited it and it is user's fault if he didn't RTFM before editing it

[Mrs Beanbag]

this might be a silly question, but if PFS3 were in ROM could you use it on floppy disks and boot from them?

[ancalimon]

Originally Posted by Mrs Beanbag

this might be a silly question, but if PFS3 were in ROM could you use it on floppy disks and boot from them?

I guess you can boot from anything if the rom was modified enough.. Still it might require trackdisk.device modification as well but I don't know for sure.

[daxb]

Does it mean that PFS3 docs are not correct or give the wrong advice?

[Yulquen74]

Ok,

What is the recommendation then for WB3.1/68000 users, when setting up
partitions >4GB?

I tried using HDInstTools on a 32GB ssd (disk is attached to scsi->ide converter inside a guru rom'ed A590).
It seems HDInstTools reports the correct size of the disk, so I made a partition containing the whole disk.

When launching format from WB, it says the disk is 1.5GB, and after
a quick format WB reports that the disk has 3900 MB's free. Does WB report
the wrong number, or is the partition size indeed 4GB ?

Originally Posted by Toni Wilen

No idea. I consider hdinsttools broken. (EDIT: because it does not handle RDB geometry info correctly)

[demolition]

WB3.1 does not handle >4G partitions properly. They can work, but remaining space will always show as <4GB (it wraps around). If you keep filling it up, it will eventually wrap around when it reaches 0 and then show ~4GB free again. WB3.9 shows the correct free space (maybe also 3.5?).

It can be an issue if it shows very little space free and you are trying to install some software, as it will think it cannot fit. Then you can copy some files to it to make it wrap around, and then the installation will run.

[Bamiga2002]

"http://eab.abime.net/showpost.php?p=921061&postcount=12"
Just grab Don Adans scsi.device v46, put it into memory with LoadModule (in S-S) and DITCH HDInstTools - its buggy trust me. Use HDToolBox instead.

[demolition]

Since he's running on an A590 I think he's not using scsi.device?
Anyway, scsi.device has no influence over what WB writes as remaining space.

[Toni Wilen]

Originally Posted by Mrs Beanbag

this might be a silly question, but if PFS3 were in ROM could you use it on floppy disks and boot from them?

trackdisk.device won't have any filesystem selection code, it most likely just says "use default" = OFS (leaves filesystem field empty).

Originally Posted by demolition

Since he's running on an A590 I think he's not using scsi.device?
Anyway, scsi.device has no influence over what WB writes as remaining space.

Exactly. A590 is true SCSI controller and direct scsi just works, there are no artificial limits.

[daxb]

Originally Posted by Yulquen74

When launching format from WB, it says the disk is 1.5GB, and after
a quick format WB reports that the disk has 3900 MB's free. Does WB report
the wrong number, or is the partition size indeed 4GB ?

Use http://aminet.net/package/util/sys/info instead. Should print correct partiton sizes.

[Mohican]

Originally Posted by Bamiga2002

"http://eab.abime.net/showpost.php?p=921061&postcount=12"
Just grab Don Adans scsi.device v46, put it into memory with LoadModule (in S-S) and DITCH HDInstTools - its buggy trust me. Use HDToolBox instead.

Where can I find this version v46. (checked Classic WB large HD support area and I tried with google but the latest what I found is v44.2) ?

Thank you in advance ...

[jbenam]

Originally Posted by Mohican

Where can I find this version v46. (checked Classic WB large HD support area and I tried with google but the latest what I found is v44.2) ?

Thank you in advance ...

"Official" thread is here on EAB: http://eab.abime.net/showthread.php?t=67067

And the official website is http://wt.exotica.org.uk/test.html

[Mohican]

@jbenam

big thanks mate ... added to my collection.

[Bamiga2002]

Originally Posted by daxb

If PFS3aio is not differnt to original PFS3:
The PFS3 docs say buffers have to be between 70 and 500. PFS3 cannot work right below 70. So if a User use lower 70 he is resposible.

I have always used minimum of 100 buffers with PFS3, and with partitions sized many gigabytes I have always put max 500 buffers. But I've had the fastmem to do that

[ancalimon]

Originally Posted by Bamiga2002

I have always used minimum of 100 buffers with PFS3, and with partitions sized many gigabytes I have always put max 500 buffers. But I've had the fastmem to do that

That's what did (at least 300 buffers for many partions) with my OS3.9 installation. Now that I will install and try os4.1 once 4000T is back home in her shoes again, I'm afraid I'll reduce those buffers :P

[Foul]

Does PFSIII require the scsi.device or is there a way to load pfs3 into memory to replace scsi.device at boot ? Blizkick or Loadmodule maybe ?