home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Aminet 22
/
Aminet 22 (1997)(GTI - Schatztruhe)[!][Dec 1997].iso
/
Aminet
/
util
/
cli
/
MemSpeed2.lha
/
MemSpeed2.BLOODYREADME!
< prev
Wrap
Text File
|
1997-06-12
|
5KB
|
109 lines
»»»» MemSpeed 2.0 »»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»»
02-06-97 20:44
By Olly Koenders
4/43 Thomas St.
Ringwood, Victoria
Australia. 3134
First released on GREMLiNS BBS. »» +61 (03)9870 4393
MemSpeed is Freeware and may be distributed far and wide for the benefit of
the Amiga community without charge except for small, unquantifiable amounts
where cost of "magnetic media" is concerned. Please don't alter the docs as
I like them pretty much too! Keep them with the executable when in piracy
mode. Thanks.
**** Disclaimer ****
I'll NOT be held responsible for any "damage" wrongly attributed to MemSpeed
due to misuse or failure to read instructions. MemSpeed will NOT impinge
upon or harrass innocent programs, ROM routines or your mothers left foot
'cos it does exactly the same as the original ROM routine only faster!
----------------------------------------------------------------------------
MemSpeed patches the exec.library AllocMem() function and specifically
targets the "MEMF_CLEAR" routine. When a programmer allocates memory and
specifies it to be cleared with nulls, then instead of the MEMF_CLEAR routine
attempting to fill a given area of memory with nulls (MEGs of it at times)
like this:
...
moveq #0,d1
...
...
bra.b $fc1812
$fc1810 move.l d1,(a0)+ ;fill address in a0 with d1's value ($00000000)
$fc1812 dbra d3,$fc1810 ;do it until d3 = -1
...
...
...which only clears 4 bytes at a time and the next intruction is nearly
useless. I patch a routine that clears significantly more per intruction and
all the clearing instructions are head-to-foot. This allows the clearing
of the allocated memory at about 90% increase in speed (or more depending if
you're running a CPU cache) as compared to the old MEMF_CLEAR routine.
Many programmers wrongly allocate memory with the MEMF_CLEAR bit set only to
copy data to it immediately after. There's no point in doing that as it
only wastes time - though some have their reasons and should know what
they're doing. In either case MemSpeed reduces their mistakes or otherwise
to about half (so to speak).
Applications such as "Audio Engineer" or "DSound" will benefit from this as
they tend to clear the memory either during allocation or before quitting.
**** The difference between version 1.0 and this (2.0) release is that I
included the "FragLess" section of code (10 bytes) in the main patch.
MemSpeed is fully aware of the FragLess patch and if FragLess is run before
MemSpeed then MemSpeed will not patch its function - as the FragLess patch
can't be removed - and the same in the reverse situation as they both patch
the same area. The only way to have the most benefit is to use MemSpeed 2.0
only. See the docs on FragLess for info on what FragLess actually does.
Any program using exec's AllocVec() or AllocEntry() will also be passing
through MemSpeed's patch so the benefits can be seen in more than just one
area overall.
MemSpeed MUST be launched from the CLI/Shell only due to it not replying to
messages from the WorkBench task. If you attach an icon and launch it from
the WorkBench then expect the "Guru" to come visiting within a couple of
nanoseconds. Note that this also applies to the "WBStartup" drawer as
launching MemSpeed from this location requires an icon to be present also and
will return the same result. This is the same for all CLI only programs
and is NOT the result of bad programming.
MemSpeed needs no arguments or parameters as all it does is insert the patch
which is 268 bytes in size and then quits. It doesn't leave a "process"
running in the background so there's no wasted processor clock-cycles. The
most appropriate place for MemSpeed is at the top of the startup-sequence
where it can begin its work early although it can be started at anytime.
The patch will never be removed unless a reset or Guru appears (which will
NOT be the fault of MemSpeed!) because there'd be no point in doing it - what
the hell, it's only 268 bytes...
If anyone's got bright ideas or would just like to say hello then I'm
contactable at the address at top of doc or on the BBS number provided.
Other fast and compact assembler utils by Olly...
modCRUSHER - 31 instrument MOD cruncher - pack rate 60 - 85%! Beats LHArc,
Lzx, Zip, Zoo, Arj, Imploder 4, PowerPacker 4 hands down!!!
CLI/WorkBench versions - All Amigas.
DoubleTake - Locates and lists duplicate files on any Dos-legal device - a
space saver for HD's! - All Amigas.
WinSpeed - Speeds up refreshing of intuition/WorkBench windows by up to
4 times. Several options mix-and-matchable.
Some other benefits too. All Amigas.