FastATA 4000 MK-VI - maxtransfer and filesystems

[Amon_RA]

I'm having some random freezes during CWB boot with my FastATA 4000 MK-V... and I was wondering if I might have setup my CF wrongly.

• Do you still need to change the maxtransfer to 0x1FE00 when setting up a CF, or will the FastATA controller take care of that?
• On the Elbox site it is mentioned : "may be used with >4GB HDDs independently of the installed filesystem". Does this mean you can just use the default FFS for partions >4GB? So no need for a custom filesystem and patched scsi.devices anymore?
• I assume I need to load the fastATA.driver in the startup-sequence AFTER the usage of BlizKick (loading a custom kickstart ROM) ?

[daxb]

I guess with "may be used with >4GB HDDs independently of the installed filesystem" they mean the Split feature (each 2GB partitions).

Originally Posted by Amon_RA

• I assume I need to load the fastATA.driver in the startup-sequence AFTER the usage of BlizKick (loading a custom kickstart ROM) ?

Yes.

[glenn]

Yes they mean the split feature, I use a 20G harddrive splitted in 6(!) partitions of 4G.

The split function split a large harddisk in 4G pieces and then present them to the OS as independent disks that you then can partition with hdtoolbox.

And even if people claim 2G is the max limit the real limit is 4G per partition with standard FFS, the only resason that C= recommended a max size of 2G is that there is a limit of 2G per file, and there is no system to stop you from accidenticaly create a file larger than 2G, therefore it is a potential danger to have partitions larger than 2G.

I have used kick 3.1 and 4G partitions about 20 years now..

Yes you probably should load the fastata driver after blizkick

Regarding the maxtransfer I only change it when its needed, and thats not very often so I don't know.

[Amon_RA]

Thanks for the replies.

I have 3 partitions, a boot partition of about 500MB, and 2 other partitions of around 7GB. All formatted with PFS3 and maxtransfer 0x1FE00.
So no need to use the SPLIT function of FastATA I guess.

[daxb]

You can use Thomas its MaxTransTest Tool to check if the MaxTransfer value is safe to use or not.

If you use SFS/PFS or a new proper FFS version, then you can avoid the Split option of the FastATA. With a large HD split into 2GB wouldn`t make much sense.

[thomas]

Originally Posted by Amon_RA

[*]Do you still need to change the maxtransfer to 0x1FE00 when setting up a CF, or will the FastATA controller take care of that?

The FastATA should not have any MaxTransfer issues. But 0x1fe00 is a standard value which can be set on all IDE harddrives. It does not harm to set it but it can harm to forget it.

[*]On the Elbox site it is mentioned : "may be used with >4GB HDDs independently of the installed filesystem". Does this mean you can just use the default FFS for partions >4GB? So no need for a custom filesystem and patched scsi.devices anymore?

The FastATA comes with its own scsi.device. A patch for Commodore's scsi.device does not have any effect on the FastATA.

AFAIK the FastATA driver supports all available command sets, so in combination with a suitable file system you can partition the harddrive as you like. Even one big 20GB boot partition will work.

[*]I assume I need to load the fastATA.driver in the startup-sequence AFTER the usage of BlizKick (loading a custom kickstart ROM) ?

BlizKick will remove every resident modules from memory, so it should always be the first command in startup-sequence. All other commands like IDEfix or fastATA.driver should come next.

Originally Posted by Amon_RA

I have 3 partitions, a boot partition of about 500MB, and 2 other partitions of around 7GB. All formatted with PFS3 and maxtransfer 0x1FE00.
So no need to use the SPLIT function of FastATA I guess.

That's perfectly ok.

Originally Posted by Amon_RA

I'm having some random freezes during CWB boot with my FastATA 4000 MK-V... and I was wondering if I might have setup my CF wrongly.

A wrong harddrive setup should not cause system freezes. Usual symptoms are that libraries or devices cannot be found although the file is there or that pictures cannot be shown or sounds not be played although the files seem ok.

If the harddrive or IDE bus freezes, you usually can use the system just normally only disk accesses are not responded to. For example you can close Workbench windows but not open new ones.

[Amon_RA]

Thanks a lot for your clear explanation @thomas, it's much appreciated!

Originally Posted by thomas

The FastATA comes with its own scsi.device. A patch for Commodore's scsi.device does not have any effect on the FastATA.

This means that I can load any scsi.device in ROM, or in a custom ROM loaded by BlizKick, before the FastATA.driver is called in the startup-sequence, correct?
Strangely enough I don't see a difference in "version scsi.device" before or after loading the FastATA.driver?

Originally Posted by thomas

AFAIK the FastATA driver supports all available command sets, so in combination with a suitable file system you can partition the harddrive as you like. Even one big 20GB boot partition will work.

What would be the best File system to go for using an FastATA and a 16GB CF? Maybe just stick with PFS3 so I can always read the CF in other Amigas that don't have a FastATA just in case (although they would require a patched scsi.device to be able to handle >4GB)?

Originally Posted by thomas

BlizKick will remove every resident modules from memory, so it should always be the first command in startup-sequence. All other commands like IDEfix or fastATA.driver should come next.

Indeed, the strange thing however is that if I put "C:fastATA.driver QUIET" after "C:BlizKick DEVS:custom.rom CPUCARD QUIET" workbench just freezes at boot, I can see the icons appear, but then the cursor changes into the little round timer clock and it just freezes.
When I put "C:fastATA.driver QUIET" before "C:BlizKick DEVS:custom.rom CPUCARD QUIET" workbench sometimes starts, sometimes not... I see the same behavior with cold, warm, hot reboots.

Originally Posted by thomas

A wrong harddrive setup should not cause system freezes. Usual symptoms are that libraries or devices cannot be found although the file is there or that pictures cannot be shown or sounds not be played although the files seem ok.

If the harddrive or IDE bus freezes, you usually can use the system just normally only disk accesses are not responded to. For example you can close Workbench windows but not open new ones.

I have the feeling that the issue that I'm seeing in indeed related to libraries or devices or pictures not being found for one or the other reason... Could it be related to the fact that the FastATA is set to PIO 5? Might it be too fast for the CF to follow?

[daxb]

Originally Posted by Amon_RA

This means that I can load any scsi.device in ROM, or in a custom ROM loaded by BlizKick, before the FastATA.driver is called in the startup-sequence, correct?

Yes.

Edit: Only true for FastATA A1200. Not for FastATA A4000!
Until another scsi.device is loaded the verson in ROM is used. That is important to know if you start your Amiga without Startup-Sequence. At that point the limits of the ROM scsi.device are valid. If you do write access to HD start the FastATA driver first. Edit end.

Strangely enough I don't see a difference in "version scsi.device" before or after loading the FastATA.driver?

A "version scsi.device" will show the version in ROM (at least here). Scout should show the right version under "Devices".


Are you sure that the driver for FastATA A4000 MK IV is named "fastATA.driver"? I ask because the driver for FastATA A1200 is named "ata3.driver". However, about your boot problems you may add "set echo on" on top of your startup-sequence to see if something goes wrong. Its also a good idea to remove the "QUIET" switch from at least blizkick and fastATA.driver for testing time. If you want to check if fastATA.driver is running you can start it after booting from a Shell. If already running it should output some information about connected devices.

[Amon_RA]

I finally figured out the issue thanks to the feedback from @daxb and @thomas above. Thank you guys!!

Originally Posted by daxb

If you do write access to HD start the FastATA driver first.

I didn't do this... Whenever I booted without startup-sequence I never manually loaded the fastata.driver before editing the startup-sequence... This seems to be very crucial indeed. Not doing this corrupted the startup-sequence file in a way and stopped the startup-sequence during the workbench boot. I wonder what exactly goes wrong when writing to HD without a FastATA driver loaded... ?

I now booted without startup-sequence, manually loaded the fastata.driver and edited the startup-sequence file. I've put the "C:BlizKick DEVS:custom.rom CPUCARD QUIET" as the first line, and the "C:fastATA.driver QUIET" on the second line, and now everything boots perfectly! Problem solved!

Originally Posted by daxb

Are you sure that the driver for FastATA A4000 MK IV is named "fastATA.driver"?

Yes, it's called fastATA.driver for the A4000 FastATA, on a A1200 FastATA it's indeed called ata3.driver...

Originally Posted by daxb

However, about your boot problems you may add "set echo on" on top of your startup-sequence to see if something goes wrong. Its also a good idea to remove the "QUIET" switch from at least blizkick and fastATA.driver for testing time.

I've done this, and it's indeed very helpful. The only issue is that you get an "intuition wants to reload workbench" error, and the resolution is not ok, but that's not really a problem when troubleshooting.

Really happy to have find a solution to this issue, thanks everyone for the help.

[Amon_RA]

I finally figured out the issue thanks to the feedback from @daxb and @thomas above. Thank you guys!!

Originally Posted by daxb

If you do write access to HD start the FastATA driver first.

I didn't do this... Whenever I booted without startup-sequence I never manually loaded the fastata.driver before editing the startup-sequence... This seems to be very crucial indeed. Not doing this corrupted the startup-sequence file in a way and stopped the startup-sequence during the workbench boot. I wonder what exactly goes wrong when writing to HD without a FastATA driver loaded... ?

I now booted without startup-sequence, manually loaded the fastata.driver and edited the startup-sequence file. I've put the "C:BlizKick DEVS:custom.rom CPUCARD QUIET" as the first line, and the "C:fastATA.driver QUIET" on the second line, and now everything boots perfectly! Problem solved!

Originally Posted by daxb

Are you sure that the driver for FastATA A4000 MK IV is named "fastATA.driver"?

Yes, it's called fastATA.driver for the A4000 FastATA, on a A1200 FastATA it's indeed called ata3.driver...

Originally Posted by daxb

However, about your boot problems you may add "set echo on" on top of your startup-sequence to see if something goes wrong. Its also a good idea to remove the "QUIET" switch from at least blizkick and fastATA.driver for testing time.

I've done this, and it's indeed very helpful. The only issue is that you get an "intuition wants to reload workbench" error, and the resolution is not ok, but that's not really a problem when troubleshooting.

Really happy to have find a solution to this issue, thanks everyone for the help.

[daxb]

Originally Posted by Amon_RA

I didn't do this... Whenever I booted without startup-sequence I never manually loaded the fastata.driver before editing the startup-sequence... This seems to be very crucial indeed. Not doing this corrupted the startup-sequence file in a way and stopped the startup-sequence during the workbench boot. I wonder what exactly goes wrong when writing to HD without a FastATA driver loaded... ?

Your startup-sequence was corrupted?

Edit: Removed wrong stuff. See Thomas note below.

Edit: Following is only true for FastATA A1200. Not for FastATA A4000:

So, for example if you boot without startup-sequence and start a game you want to play from a partition beyound 4GB. The game saves a highscore or config to HD. Then it can happen that the write access is done on boot partition instead of the game one. This can/will cause data corruption you may notice too late.

[thomas]

Originally Posted by daxb

You just need to know the limits of old scsi.device <=V40 what is in use when no fastata.driver is running.

This is not true. FastATA 4000 is a Zorro card with its own boot ROM. A driver in startup-sequence is not needed. FastATA.driver is only there to apply settings you made in FastATAPrefs and maybe to patch some bugs in the ROM driver. But large harddrive support is already present at boot time. Like I said above, you can even make one big boot partition, covering the entire drive (if the file system supports that).

Version of scsi.device does not matter. V40 is not the FastATA driver, it's Commodore's driver for the internal IDE port. It has nothing to do with FastATA.

If another scsi.device is already in the system, then the FastATA driver becomes 2nd.scsi.device or 3rd.scsi.device and so on.

The FastATA driver has nothing in common with the Motherboard driver except a part of the name, and they don't act on each other's hardware.

You should be aware that the FastATA 4000 is a true and properly implemented expansion, not a patchwork like the FastATA 1200.

[daxb]

Thanks for clarification Thomas! I thought FastaATA A1200 and FastATA A4000 work/behave the same or very similar. Two further questions to FastATA A4000:

Does it mean that devices at FastATA and at internal IDE port can be used both parallel?

Will the FastATA initialized/recognized during hardware initialization like for example accelerator boards?

[thomas]

Originally Posted by daxb

Does it mean that devices at FastATA and at internal IDE port can be used both parallel?

Yes, the FastATA adds two IDE channels, it does not occupy the internal IDE port.

Will the FastATA initialized/recognized during hardware initialization like for example accelerator boards?

Yes, you can boot from a drive connected to the FastATA.

OTOH if you remove all drives from the internal IDE port you will suffer from the 10s boot delay.

[idrougge]

Originally Posted by thomas

OTOH if you remove all drives from the internal IDE port you will suffer from the 10s boot delay.

But that is easily solved using an "IDE terminator".

[glenn]

You can even use a splitter on the internal IDE combined with fastATA4000.

In my A4000T I have two harddisks and one CD-burner on internal IDE, and then four harddisks on the FastATA4000.

[klx300r]

can any owners confirm if I can use my DOM with FastATA 4000 MK-VI? Right now its happily working in the MB IDE connector but would be great to speed it up on the FastATA if possible

[johnnybarker]

I had a 16gb hyperdisk DOM working fine on my fastaATA4000.

[klx300r]

Originally Posted by johnnybarker

I had a 16gb hyperdisk DOM working fine on my fastaATA4000.

awesome thanks for the feedback! what speed were you getting from it?

[johnnybarker]

It was about 6-7mg/s I think (sysinfo results). Unfortunately I can't test it again as the fastata went on fire when I accidently fitted it upside down.