home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
AmigActive 6
/
AACD06.ISO
/
AACD
/
System
/
FBlit
/
FBlit245a
< prev
next >
Wrap
Text File
|
1998-06-25
|
7KB
|
221 lines
FBlit 2.45a
-----------
'FBlit' is © Stephen Brookes 1997-98
Blah
----
This software is experimental, incomplete and fundamentaly dangerous, so
don't use it! I will take no responsibility for any undesirable effects
resulting from the use of any software or information in this archive.
Distribution
------------
'FBlit' is freely distributable.
Required
--------
020+ CPU, the faster the better.
Fast RAM
V39+
The GUI requires MUI, but the GUI itself is not essential. Indeed, you are
advised not to go anywhere near it!
Installation
------------
Use the Install script. (this does not alter your startup-sequence)
Or...
Put 'FBlit' and 'FBlitGUI' in 'C:'
And note that this version contains some new tasks in the 'included tasks'
list. If you were already running FBlit, delete 'envarc:fblit.cfg' and
reboot to get the new tasks. Or, if you don't want to lose your current
config, add them yourself (see FBlitGUI.doc). The tasks are...
'AsyncLayoutDaemon' (Multiview)
'AWebJFIF decoder'
'AWebGif decoder'
'AWebPng decoder'
In any event, if you were running a previous FBlit, you will need to reboot
after installing this one.
Usage
-----
Type 'run >nil: fblit', or launch from the workbench.
For a permanent installation. Add the line 'run >nil: c:fblit' to your
's:startup-sequence' soon after 'env:' and 'envarc:' have been set up.
See separate doc for GUI usage. 'Cancel' is probably the only gadget you
want to use :)
What's That?
------------
If you skipped the 'Blah' section, go and read it now!
FBlit patches some OS functions that normally use the blitter, such that
they can use the CPU, and consequently may function outside chip mem.
The patched functions are also faster and will suffer less from colour
flicker on non-interleaved data.
In addition, FBlit allows non-displayable bitmaps to be allocated in fast
mem for named tasks.
The current version of FBlit with default config will do the following.
- Minor OS rendering speed up.
- Reduction in colour flicker.
- Allow NewIcons to be stable in RTG mode on a standard Amiga (if you can
live with corrupt icons while dragging them, or are running DopusMag)
- Forces the following programmes to use fast ram for some graphics data.
Voyager 288
IBrowse
AWeb
IPrefs (workbench/window backdrops)
Multiview
FastIPrefs also gets promoted because it pretends to be IPrefs. This may
depend on the version of FIP you run.
Voyager295 (perhaps other 29x versions also) does not get promoted, but it
uses fast mem for images itself anyway.
IBrowse1.2 apparently uses fastmem now itself as well.
Also, it seems AWeb3.1 may not get promoted unless you add the new tasks
mentioned in 'Installation'. (these tasks are already in the default cfg,
so you probably don't need to worry about it)
Note that the only version of AWeb3.1 I have (CUCD version) appears to still
use 'AWebIP' for allocating images, so I have not tried these myself.
The Multiview promotion is only partial. If you use it to display pictures
in a window, chipmem use will be reduced slightly, and scrolling arround
large images will be significantly faster.
FBlit must be run early in the startup sequence if default wbpatterns are to
be promoted, and similarly if NewIcons is to be launched in its RTG mode.
Trouble And Strife
------------------
FBlit appears to function perfectly on the majority of systems, but it can
never be 100% compatible. There is much scope for other software to cause
problems and some certainly does.
Unfortunately, none has been identified at this time. If FBlit causes
corrupt graphics on your system you will have to either forget about FBlit,
or find the software causing the problems and then decide which is more
important to you. (and please tell me about it. The last (2.4) aminet release
only generated 2 'this doesn't work' comments. One, 'FastIPrefs' was cured in
2.43+, the other was related to the screen depth gadget, but I get strange
behaviour there with or without fblit...)
Other software that patches the same functions as FBlit may also cause
trouble. MCX, MCP, CPUBlit etc... though I run MCP here without problems.
Short Tech Bit
--------------
FBltBitMap is fully functional and the most complete/tested patch.
FBltClear is also fully functional, but probably pretty pointless.
FBltTemplate only intercepts and discards operations in Fast RAM.
FBltPattern will only work in Fast RAM with JAM2 and no mask/area pattern.
FBitMapScale is fully functional, but a very early version.
FAllocMem and FAllocBitMap are of no interest.
FSetRast may be fully functional.
The fact that 'BltBitMap' is patched also effects a large number of other
OS functions. I don't know exactly how much of the OS can now operate in
Fast RAM, 'DrawGList' (BOBs etc.) probably won't, 'BltTemplate' (text) and
'BltPattern' certainly won't. 'BltBitMapRastPort', 'ClipBlit',
'BltMaskBitMapRastPort' and 'DrawImage' are useful ones that will. Super
bitmap stuff also appears to work.
In general, you are likely to have problems if a fast mem bitmap is
attached to a rastport.
More techy nonsense is in the FBlitGUI.doc and Programmers_Note.doc
Past
----
There has been loads of that, and I'm not going into it here.
Future
------
Is under consideration....
PPC has been mentioned. That's a possibility, but I guess (given that I
know nothing about WarpOS/PPC.lib or whatever might replace them) that
patching the 68k OS with PPC routines will be very slow. Also fblit is
largely involved in accessing chip mem, so any speed up might be marginal,
or even negative, given what I said in the previous sentence.
Built With Amiga?
-----------------
Development machine is...
A1200 040/40 V40 etc.
Please Crash My Computer
------------------------
Thanks to some people who like corrupt displays so much, they tried this
software...
Evan Tuer <et@enterprise.net>
http://homepages.enterprise.net/et
(Endless Testing and icon)
Gary Colville <gary@empire.u-net.com>
2:250/366.12
http://www.personal.u-net.com/~empire
(Testing)
James L Boyd <N/A>
(More testing & AWeb info)
Colin Wenzel <colstv@pronet.net.au>
http://www.pronet.net.au/~colstv
(IBrowse testing)
Luca Longone <llong@tin.it>
(FastIPrefs & GUI stack harrassement)
Rick Pratt <rickprat@iswt.com>
(Additional code and impetus)
Thanks also to everyone who mailed me about fblit. All comments are
appreciated :)
Contact
-------
Stephen Brookes <sbrookes@tpec.u-net.com>
Fido currently unavailable