Page 57 of 60

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 02 Jan 2021 19:00
by arkadiusz.makarenko
terriblefire wrote: 02 Jan 2021 18:36 Just FYI..

here is the times you're getting

bfe001.JPG
800ns? It is not that bad is it?

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 02 Jan 2021 19:22
by terriblefire
arkadiusz.makarenko wrote: 02 Jan 2021 18:59 Could you check this? @terriblefire
There are two changes, KBD_Clock is fixed to one setting Open Drain and never reconfigured. I checked on datawrites and seems to be ok.
second change is simple optimization -OFast insead of -ODebug maybe speed will be a little but better?
Will do. ignore my last. the speed is 7uS to complete a cycle with the ARM. Thats way way too slow.

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 02 Jan 2021 19:27
by terriblefire
terriblefire wrote: 02 Jan 2021 19:22
arkadiusz.makarenko wrote: 02 Jan 2021 18:59 Could you check this? @terriblefire
There are two changes, KBD_Clock is fixed to one setting Open Drain and never reconfigured. I checked on datawrites and seems to be ok.
second change is simple optimization -OFast insead of -ODebug maybe speed will be a little but better?
Will do. ignore my last. the speed is 7uS to complete a cycle with the ARM. Thats way way too slow.
New firmware
bfe001.JPG
this is what it looks like without a mouse plugged in
nomouse.JPG
No change in reset.

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 02 Jan 2021 20:05
by arkadiusz.makarenko
I have trimmed what I could in Interrupt routine quickly, and changed Open Drain to PushPull for KBD_Clock.
So I hope it will force to keep the reset line correctly?

Except that I have changes IRQ priorities, now it gives highest priority for Amiga, and lower for USB.

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 02 Jan 2021 20:18
by terriblefire
I have a suspicion the 020 reset line is getting stuck low. I will investigate.

Will try that tomorrow. I have noticed a few hangs when using a mouse.

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 02 Jan 2021 23:00
by arkadiusz.makarenko
Double checked everything, and now I release CPU as early as technically possible.
I have added macros as well, lets see if this made any difference at all.

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 03 Jan 2021 04:27
by kulamario
I just checked your latest firmware guys it looks like everything is working fine amiga as if it would start faster
keyboard did not start after inserting usb dongle
I had to turn off and on amiga Then everything start working properly

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 03 Jan 2021 11:49
by arkadiusz.makarenko
kulamario wrote: 03 Jan 2021 04:27 I just checked your latest firmware guys it looks like everything is working fine amiga as if it would start faster
keyboard did not start after inserting usb dongle
I had to turn off and on amiga Then everything start working properly
I have removed keyboard sync, this firmware should allow you use of keyboard straight away. But I am 99% certain that your issues are related to your mouse keyboard setup.

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 03 Jan 2021 11:55
by arkadiusz.makarenko
I was thinking how I can speed up things even more.
There very little that can be done inside the routine (I will optimize it even more, later, as there is still room for improvement)

But main issue is using HAL libraries for handling EXTI. There is a lot of overhead there which can be optimized.
Main issue is that I use only one EXTI group per one pin/ INTSIG. I should be able to trim all the checks and few other bits as well, and speed up by setting everything directly on registers, but challenge is that some elements need to stay as they are for USB stack to work.

Re: TF CD32 Riser Revision 2 Design Complete

Posted: 03 Jan 2021 15:34
by arkadiusz.makarenko
All interrupts except RTC are handled directly.
I will now go through disassembly and will look to optimize routines even more.