For a change, this time I've been writing
little pieces over the last few days; hence what you have here
is a mismash and cocophanical blend of iotas.
Customary disclaimer
Any information on this page about
Anvil, Fantasm 5 or Fanta_C is strictly unofficial! Please remember
I am working on just one part of a larger system and these are
my views only.
190397
First, thanks for all the feedback over the last StuChat. Since that was written,
lots more has been hammered out in Anvil. We've had another test
which was more
satisfactory than the first test. I've gotten hold of the Kula
Shaker CD single (at
last) which is fab, fixed some bugs in the assembly language chroma
coder plugs.
and made some minor modifications to some API specs which now
look more sensible.
Molly the camper is away at the garage getting some welding done
(life is strange with
no transport - the walk back from the garage today nearly killed
me).
220397
Well, as some of you know, I was most keen to get Anvil it's own
MOD player plug.
I'm typing this in Anvil as Anvil is playing my fave MODs. The
amazing thing
is that compared to a stand alone player, this takes less than
half the memory.
Great. I ended up using (after some debate this week) the PlayerPro
library.
I would heartily recommend this to anybody considering a MOD player
for a game
or whatever. The library we've used (4.2) is much improved over
the 4.0 version
and answers most of my problems. The size of the plug comes out
at 60k, that's
with some basic preferences and "play files in a folder"
code. Little UI.
The great thing with Player Pro is it has it's own plugs which
(as long as the file
has a "type") will import most any music file - MIDI,
MOD, XM, S3M etc. Personally
I think that Antoine is completely insane - he should charge money
for this.
Actually implementing the plug took most of today. Nearly 80%
of the time was spent
debugging and correcting Anvils I/F as this is the first native
plug written in C,
and I had some reversed parameters in "switch_to_tool"
:-)
Unfortunately it looks as if (for the time being anyway) as if
this plug (AnvMOD)
is native only - sorry, but I wouldn't let a non interpolating
plug out of the door.
Maybe some Quadras can do it - I wouldn't know being a "Gimmee
G3" man. AnvMOD goes
straight for the "Surround mode" knob and really requires
that 64 bit FPU.
Well, to say I'm chuffed with how this is all falling in to place
would be an
understatement. We've got Fant 4 about half way in - until yesterday
that's what
I was doing, so today (as I was getting maybe a little loopy about
it and my pencil broke)
decided for a bit of fun. Talking of which, we finally got most
of the file cache
hammered out last night on the phone (Err, I had missed some bits,like
sometimes we really
don't need to physically write a file to disk at all, and Rob
had forgotten about the
cached architecture and how not to blow it (PowerPC cache that
is) :-))
We've made great improvements on Anvil's memory manager this week,
and now have quite
a nice design which can act quite intelligently.
Well, here I am, end of another hard week. Newswatcher is raping
alt.binaries.sound.mods,
Mac hasn't crashed all night (no Netscape <BG>) and I'm
happy.
Found an interesting snippet in Byte (Mar 97) about how to calculate
the block size on a
PowerPC processor - goes like this:
for (i= 0; i <
64; i++)
mem[i]=0xffffffff;
dcbz (mem,0);
//place a guard word of 1's at the end of the block
mem[64]=0xffffffff;
//look for first non-zeroed word
for (i=0; mem[i]==0; i++)
/* empty loop */;
block_size=i*4;
Credit to Mike Phillip at Motorola. Actually Byte does seem to
be getting better
these days. In the last couple of years it went a little "arty"
for my liking, but just
recently it appears to be coming back to earth slightly. The number
of Mac articles
is definately on the up as well - maybe that's it?
OK, so who'd buy a WinTel now then eh? Are they in for a shock
or what! Baaaa.
It's going to be great - I can't wait. 1998 - now lets see. Intel
release P7 (RISC). No
emulator. No backup Pentium on motherboard. Hmmm. Thing is, nobody
seems to be publicising this.
And Apple have the new OS running on MacTel and clones, but of
course it runs slower
on 586 because it just hasn't the same oomph as a 604. 620's (now
in redesign as the 604
is so good) will be coming on line high end, and Intel just get
their first RISC chip
out of the door with hopefully no bugs. Intel are trying their
hardest to get a fast
Pentium (say 280 Mhz) that doesn't require compressed gas cooling
whilst 300Mhz 603e's
are reality right now (see below). I wonder what speed they'll
be running at in '98? Did I mention
Exponential :-) It must be nice to be Intel right now, but in
two years time I think
I'd rather be AIM.
MMX. Read the book, seen the film, ate the pie. Fact is, MMX barely
caught up in areas
that MMX is supposed to be good at (No, I'm not referring to wiping
the FPU context :-)).
Financial whiz kids see MMX as a "Yawner". And now we
have 300Mhz
PPC's coming up, not to mention 1/2 Ghz 704's just on the horizon.
The truth about
PPC verses Pentium is that at a given clock speed a 603 can be
up to 87%
faster,
never mind a 604!
Are corporate execs aware of all these facts? I'd put good money
on it.
Oh dear. Anvil's gone mad and played the last track five times
over...Hmmm.
240397
Well, Anvil progress has slowed quite dramatically. This isn't
the result of any
slackness on our part. What's happened is that because it is so
modular, rather than
just clump everything in one huge App (which would be easier from
our point of view at
this specific moment in time), we're slowly but surely moving
code into plug-
ins. That way one can get a much smaller memory footprint at the
expense of facilities
which is the users choice (as it should be).
What was supposed to happen for phase three is that Build would
be simply copied out
of Fantasm and into Anvil. This isn't happening now as we're thinking
of putting it into
a plug-in and rather than have a quick "kludge" translator
from Anvil project file to
what the old Build expected (i.e. a BCF) it's better to just rewrite
it and get it over
and done with as Build II. This delay's Anvil testing quite a
lot. Sorry, but it's worth
it in the long run.
250397
Well, I'm so far behind schedule now, it's not even funny any
more. I know Rob is
hammering away at the compiler. This is a bit ironic actually,
as only fourteen days
ago I was hassling Rob for it to test Anvil with - now it looks
like I'm the one behind.
Oh dear. Well, you know what they say - "If you can't make
it, reschedule!".
I'm behind because I keep going back to things that are supposedly
finished and recoding
them because I've found a new way of doing it. For example, tonight
as well as working on
Build I went back to the "stress-o-meter" and looked
more into R2D2 not to mention removing
some more code from Fantasm. The good news is the project window
no longer does the
"death act from Swan Lake" scene anymore, but still
no trash for Rob :-(
Still, I have managed to say "finished" to a few areas,
which means that instead of having
to work on eight things simultaneously, I just have four now.
It's the twins 30th this weekend, so we're off to Wales for a
party in Cardiff. Well, I say
we are but currently Molly has no brakes. Not my fault. I ordered
replacement brake
hoses last week. Today they came. So I sit down, remove one hose,
compare it with the
new one and it's about 50% too short. Grrrr. So now I have a van
with air where the
brake fluid should be and no brake hose.
Anyway, a party in Wales is not to be missed for anything (a 604
would do it thou...),
so hose or not we're going.
I was talking to Rob tonight about "when we were lads"
- you know the sort of
reminiscing; "when micro's were micro's" kind of chat.
Anyway, we got to talking about
the various electronic and computer clubs attended. Somehow or
another the
conversation moved onto "what chip did you have in your tie?"
(don't laugh).
What you do is get the MOST expensive chip you can find (preferably
not an Intel) and stick it onto your tie like a tie pin. A 604
would be cool (and would also probably get you mugged).
Anyway, I started off with a CD4011 (Quad NAND gate - guaranteed
dead 'cause it was a
nylon tie). My pet (sorry) project at the time was messing about
with 6502's and I
really wanted some sound out of it. So, I slaved in a cafe for
two weeks earning fifteen
notes to buy an AY-8-910 for my 6502 breadboard. Finally scraped
enough cash together
and plugged it in. Not five minutes had passed before it did the
old "Roman Candle"
trick and went up in a puff of smoke shortly followed by real,
genuine tears.
"Gutted" was the word. But it made a dead good tie chip
and survived there for three
years after! It finally got killed in a Latin class by some dick
head with mushy peas
where his brain should be. Got him back by wiring up the big brass
doorknob
to his room to the mains. A couple of years later I got him back
again during a deep
sea fishing trip.
290397
Is it me, or is MIDI limited? Ok, let me put some context on that.
I've used MIDI most
successfully when recording, normally with an Atari ST and MIDI
keyboard and some sequencing
S/W. MIDI is great for this kind of composition. BUT, playing
a MIDI file through my
Mac with QTMA is pretty dire. It all sounds very el cheapo synthesiserish,
or is there
something wrong with my Mac? I set it to 16 bit at 44KHz and still
get aliasing (or Nyquist)
distortion. Is it supposed to be this bad? According to the formula,
if I play back at 44KHz,
then I don't get distortion below 22Khz. One would normally expect
to find a brick wall
low pass filter at (sf/2)-((sf/2)*(5/100)) where "sf"
is the playback clock frequency.
i.e. the low pass filter is set at the upper limit minus a few
per cent, where the "few
percent" is dependent on the quality of the filter. The better
the filter, the lower
this figure. I'd say this low pass filter is missing or if present,
not working.
040497
Phew, after much hard work, I can trace right from Build, into
the translator, out again
and back to Build. And nothing gets lost - happy is not the word.
Unfortunately we had to change one of the API calls (A_Send_command)
which means that
everybody with the Anvil kit (phase three t1 or 2) may as well
throw it in the bin now.
It won't work with Build (which sends out lots of these messages).
Sorry chaps. If any
API calls change after t3 I'll shoot myself with a sawn off and
get it over and done
with :-)
It's all very exciting because any day
now, Anvil will be able to build itself. The exciting bit is that
it'll just suddenly happen; nobody can say when it'll happen!
What I mean is, is that we can dry run this to death, and we do,
but it's not until you "let it fly" that you really
know whether it works or not. The "working" bit is invariably
preceded by many crashes, so when it doesn't crash you get a nice
warm feeling, and when it works, you go down the pub and get a
nice "wobbly" feeling.
The party in Wales did not happen. Why? Molly just didn't want
to go. Still no transport.
Problem now is I've destroyed one of the rear stub axles whilst
trying to replace the
bearing. You can't get these things at any price, so currently
we're phoning left right
and centre trying to get one from somewhere. We've rescheduled
the party for next week.
If I don't get Molly going I'm going to be as popular as John
Major 'round about this time
next month.
Ouch, politics; sorry not allowed to do that.
Anyway, hopefully a nice chap will come tomorrow and say "Oh
yes, I have one of those,
give me lots of money" and I'll be happy?
Gripe of the week. Just recently I've noticed in a lot of other
peoples 68k assembler the following:
moveq.l #x,Rx
NO! I hate it. "moveq" affects all 32 bits. All assemblers
don't flinch at the ".l", but it's
just naff. I guess people do it as a reminder that all 32 bits
are being affected...
"addq and subq" are the ones you need to worry about
the size of..naff, naff, naff. Gripe over.
Find of the week is the Weekly
MOD Review site.
You'll find all the latest and greatest MODs here, all reviewed
by many people.
Now I know this is going to sound like a blatant plug, but it
isn't - honest.
About this time last week James finished his book "Halcyon Days"
(I know what Halcyon means now as well (Rob looked it up in a
dictionary - cheers mate) :-)).
Ever since, I've been avidly reading it. We waited a long time
for it, and it was worth it.
Amazing interviews. Good old James.
Some of the interviews are very inspiring, some are very funny,
and some are pretty,
well, sad.
I would recommend it to anyone with even a vague interest in what
makes these people
"tick". It's 'specially nice to know you are not the
first person in the world to kill
yourself over a hot keyboard.
In a similar vein you might like to have a look at
http://www.nvg.unit.no/sinclair/planet/index.html
This has interviews with many early Spectrum programmers such
as Jet Set Willy programmer
Mathew Smith (and I thought I was few sandwiches short of a picnic!),
and an interesting
piece on "Rare" (A bit sad IMO).
I see that 300 MHz 603e's are shipping in the form of Apple's
6500. Actually, you have to
congratulate Motorola on this as the goal was to reach 300Mhz
by the end of 1997. It's
only just turned April! Get the details here.
I've just noticed that Lightsoft WWW
has just passed it's first birthday, as the first
StuChat was posted 28th March 1996.
Wow, that went quick! That must mean that Lightsoft must be just
over a year old, as the company was officially formed on March
1st 1996. And we were so busy we missed our own birthday. Won't
happen next year - somebody remind us!
Oh well, look out for some cool stuff
in our second "official" year!
Well, I really must get back to it.
Happy Birthday Lightsoft.
Send mail to Stu
