On Tue, 28 Jan 1997, Steven Newbury wrote:

> The RapIDE support makes a incredible difference to the performance!
> However I have noticed that the kernel sometimes hangs on booting a exiting.
> I assume that as the driver is beta-release that the stability is going to 
> improve, but will it get any faster?

Yes things will get better.

The hangin on rebooting to RISCOS is not a function of the rapide driver.
THis is a behaviour that has become more frequent and appears to be
related to interrupts that are pending but masked out of the IOMD when
RiscOS is restarted. The solution is probably to write shutdown handlers
for all device drivers to make sure that interrupts are disable on chip.
 
> Is the DMA enabled, the RISC OS driver hasn't got it!  If so, does this mean
> that drive transfer time will be better in RiscBSD than it is currently in RISC 
> OS?
Nope this is uses PIO xfers. We are looking at various ways of improving
it further, potentially with DMA thought as yet I have not information on
using DMA with the RapIDE card.

> By the way is there a new parameter for setting up the ramdisc since it's become
> md rather than rd.

As mentioned in another email there will be a new bootloader to take care
of this.
The 'ramdisc=' options is 'memory disk=' in the kernels that went online
the other day but it should be 'memorydisk=' as it will be in any future
kernels. This means that there are a few hiccups with the 4956 kernel and
the current bootloader.
A new bootload will go online that will sue the correct md device etc. and
will work fine with all new kernels. In the meantime use of the other
options field in the bootloader is needed.

Cheers,
				Mark