home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Turnbull China Bikeride
/
Turnbull China Bikeride - Disc 2.iso
/
STUTTGART
/
UTIL
/
MEMORY
/
UPDATEMEMC
/
ReadMe
next >
Wrap
Text File
|
1995-07-02
|
3KB
|
101 lines
—————————————————————————————————————————————————————————————————————————————
•••••• ••••••
blackICe •••••• ••••••
••• •••
Marko Lukat ••• •••
Venise-Gosnat-Straße 48 ••• •••
14770 Brandenburg an der Havel ••• •••
Germany ••• ••• matrix
—————————————————————————————————————————————————————————————————————————————
UpdateMEMC Version 0.26 (02 Jul 1995)
—————————————————————————————————————
Machines with MEMC1(VL86C110) control their DRAM refresh through
the SWI "OS_UpdateMEMC". After power on it's set to DURING VIDEO
FLYBACK ONLY. During screen blanking it'll change to CONTINUOUS.
But before Service_ScreenRestored is delivered it's reset to its
original value ... but it isn't!!!!!
The MEMC refresh state is set to NONE.
DRAM Refresh Control (MEMC1/VL86C110)
bit 98
—————————————————————————————————————
00 None
01 During Video Flyback Only
10 None
11 Continuous
As long as screen update will be performed there is no reason to
be alarmed. Reading screen memory periodically (video DMA) will
perform at least a worst case refresh.
Sometimes sound and video DMA will be switched off to get more
speed during vital actions (!ChangeFSI, !GZip). This time the CPU
is the only one which performs memory access. And that's not enough
for a worst case refresh ...
NB:The same bug still remains in RISC OS 3.5 but there is no MEMC
at all. The SWI "OS_UpdateMEMC" is kept for compatibility and its
only function is to control the video DMA by examining bit 10 and
perform the appropriate action on the IOMD chip.
How it works
————————————
UpdateMEMC installs itself in the module area, and listens out for
every "OS_UpdateMEMC" call. When one goes off, bit 8 will be forced
to 1 (in R0 and R1) i.e. there is no chance to run without refresh.
How to use it
—————————————
Simply double-click on the UpdateMEMC module. If the original SWI
entry (vector location &00000008) contains no branch instruction
this module refuses to install. Also you can't kill this module.
(Why you should do this?)
This module has been tested under RISC OS 3.10 and 3.50. Please
feel free to send me bug reports.
Credits
———————
Well, some to me :
Marko Lukat
Venise-Gosnat-Straße 48
14770 Brandenburg an der Havel
Germany
Email : Marko.Lukat@E-Technik.TU-Chemnitz.DE
Mail me if you want the source.
NB:The SoundDMA module version 1.21 (RISC OS 3.1) and version 1.36
(RISC OS 3.5) contains some horrible bugs too, but not as important
as the one above (you can live with them).
(I've patched them already and if you are interested ...)
—————————————————————————————————————————————————————————————————————————————
regards
Marko
Date: Sun, 02 Jul 95 17:36:47
—————————————————————————————————————————————————————————————————————————————