Hextracker (updated to 0.849B + BPM preview)

[DarkLord]

Grabbed the E and F file.

The E version locks up, just like the previous non-working releases, mouse pointer
frozen on screen.

The F version appears to work normally. I loaded up a couple of MODs, played them,
exited back to the desktop, no problems.

Thanks!

[ljbk]

Hi !


Thanks for the tests, again.

The F solution has been included in the main package so hopefully next time your machine will nor need a special version.


Thanks again,
Paulo.

[ljbk]

Hi !

Hextracker 0.848 is now available with the new and much faster STE multiplex mode strategy and a speed-up increase for all output types and modes with gains from 0 to 2% CPU time.
STE owners can now choose between 25 KHz pure mode, 25 KHz oversampled at 50 KHz mode and 25 KHz multiplexed at 50 KHz mode for almost the same CPU time, depending on the sound they like the most, when 50 KHz are not possible.

Have a nice Easter !

Paulo.

PS: DarkLord, this build should now work on your special STacy, i hope ...

[DarkLord]

Hi !
PS: DarkLord, this build should now work on your special STacy, i hope ...

Yep! 0.848 worked just fine. Unpacked the archive, loaded up some mods, played them
and exited back to the desktop, all with no problems.

Thanks for all the hard work!

[ljbk]

Hi !
PS: DarkLord, this build should now work on your special STacy, i hope ...

Yep! 0.848 worked just fine. Unpacked the archive, loaded up some mods, played them
and exited back to the desktop, all with no problems.

Thanks for all the hard work!

Nice !

This means that from now, all packages will work on your machine.
This also means that if anyone ever uses the provided HMINxxx.s source to replay the musics then it will also work on your machine.

Thanks again for the tests.

Have a nice Easter,
Paulo.

[ljbk]

Hi !


So, here comes another Hextracker update now to version 0.849.

First change is to use a C note as reference instead of a D# note to be in line with other major programs on this and other platforms.
So the reference sample frequency is now 8287 for loaded Protracker MODs.
You can change it to any value you like (C2 REF). If you want the NTSC behaviour just change this to 8363.
If you saved a MOD in Hextracker format, you will have to update the old value by a 0.8408964 factor.
If you used the 9855 value, just replace it with 8287 or 8363 if you prefer.

Again speed was increased a little bit for all. Any handled 16 voices MOD can now be played on YM at 9.75 KHz. I started at 9.17 KHz with version 0.825. The 10 KHz barrier is still far but closer ...
Some important bugs have been solved like the program crashing if the number of voices was not multiple of 4 or below 4, in 25X50 multiplex mode.
Small new features have been added like the click on pattern data to go one pattern data location and the stereo toggle on STE by clicking on MONO or STEREO.
Two major new features are:
- platform test for CPU speed was added and Sound DMA speed test was improved to get a better accuracy during replay;
- HB replay mode was added to get a faster true 8 bit samples replay but requiring a bit of memory;

This last feature allows the replay of any handled 4 voices MOD at 50 KHz in Stereo on a 8 MHz STE using 8 bit samples. I don't know if anyone ever achieved this before.

For you to have an idea about the memory requirements, the sample used area doubles but the generated code area decreases because it is easier to replay compared to the old 8B replay.
Here is an example for the ART_HXT.MOD:
@16.7 KHz:
- 7B base as reference: 0;
- 8B + 67842 bytes are required;
- HB + 125330 bytes are required;
@50 KHz:
- 7B base as reference: 0;
- 8B + 80314 bytes are required;
- HB + 125330 bytes are required;
As you can see the memory difference with the 7B case is identical @16.7 KHz and @50 KHz: it corresponds to the total samples size. On the other hand, between 8B and 7B, the difference increases with frequency because the 8B rout is more complex, so the more updates we have the bigger it turns out to be.


Enjoy it,
Paulo.

[FedePede04]

Thanks for the update

[ljbk]

Auto repeat keys feature added with 0.849A version update.

Yerzmyey will be happy !

[yerzmyey]

Thanx a lot man, I do appreciate this, as I use the program more and more often. Even some people from Amiga (hehe) started to ask about it after the latest YM-Digital.

Meanwhile I've got some stuff to report, I hope it will be helpful.

We were talking about speed differences.
Well, the loop isn't too good/straight but I don't have any other, at the moment.

OCTALYSER (STE) - http://ym-digital.i-demo.pl/OCTA_STE.mp3
This tempo is proper. (Yes, sounds like it's too fast in some moment but it's my fault because I'm stupid and lazy, so anyway - that's exact how it goes from PC, from MilkyTracker).

HEXTRACKER 844 (my king and master ) - http://ym-digital.i-demo.pl/HEX844.mp3
(don't mind the YM noise - it the emulator what generates it, it sounds much better from real ST).
As You can notice, the tempo is _slightly_ faster.
Or the samples are playing is a bit slower?
Anyway it "eats" small part of the loop.

HEXTRACKER 849a (the newest one) - http://ym-digital.i-demo.pl/HEX849error.mp3
The loop is destroyed, also seems like it's being played much faster, as a sample, btw.

All of this happens just after the song's loaded, no tinkering involved.
I noticed the "sample frequency is now 8287" but it's confusing for users to have non-compatible loops with other trackers, just after the HEX is loaded. The default setting, I mean.

Another thing that is disturbing to me, as an YM user, is a loss of quality:
this song on HEX844 works with quality of 22755 Hz.

The same song played on HEX849a stops with the same 22755 Hz quality, reporting CPU's overload.

If You need a test-material, the song is in this pack http://ym-digital.i-demo.pl/YM_DIGITAL2.zip (Alley of Stars).

I hope it's helpful.

[ljbk]

Thanx a lot man, I do appreciate this, as I use the program more and more often. Even some people from Amiga (hehe) started to ask about it after the latest YM-Digital.

Meanwhile I've got some stuff to report, I hope it will be helpful.

We were talking about speed differences.
Well, the loop isn't too good/straight but I don't have any other, at the moment.

OCTALYSER (STE) - http://ym-digital.i-demo.pl/OCTA_STE.mp3
This tempo is proper. (Yes, sounds like it's too fast in some moment but it's my fault because I'm stupid and lazy, so anyway - that's exact how it goes from PC, from MilkyTracker).

HEXTRACKER 844 (my king and master ) - http://ym-digital.i-demo.pl/HEX844.mp3
(don't mind the YM noise - it the emulator what generates it, it sounds much better from real ST).
As You can notice, the tempo is _slightly_ faster.
Or the samples are playing is a bit slower?
Anyway it "eats" small part of the loop.

HEXTRACKER 849a (the newest one) - http://ym-digital.i-demo.pl/HEX849error.mp3
The loop is destroyed, also seems like it's being played much faster, as a sample, btw.

All of this happens just after the song's loaded, no tinkering involved.
I noticed the "sample frequency is now 8287" but it's confusing for users to have non-compatible loops with other trackers, just after the HEX is loaded. The default setting, I mean.

Another thing that is disturbing to me, as an YM user, is a loss of quality:
this song on HEX844 works with quality of 22755 Hz.

The same song played on HEX849a stops with the same 22755 Hz quality, reporting CPU's overload.

If You need a test-material, the song is in this pack http://ym-digital.i-demo.pl/YM_DIGITAL2.zip (Alley of Stars).

I hope it's helpful.

Hi !

Thanks for the reporting.

Let's go step by step.

First the C2 REF (previously FSAMPL).
If the MOD was made on an european (PAL) Amiga, then it should be played with C2 REF = 8287.
If the MOD was made on an american (NTSC) Amiga or if you want to compare with Winamp or Milky, then you should use C2 REF = 8363.
These values are well known in the MOD community.
If you want to have simple 16 byte loops, then you should use values from 8354 to 8363.
8356 / 16 = 522.25 Hz => C note !
8356 is the value that works best for the complete set of 12 semi-tones (and not only the C).

Second the CPU overload.
With version 0.849, you have a better accuracy. The CPU speed is really measured. If it was working before with 0.844 and it is not working with 0.849, this means that before the MOD was played a bit too slowly because your CPU speed was assumed to be the most common (8021247) and in fact it is slower. So probably more sample reads or mix calculations have to be done per frame even if it is just 1.
Try this: viewtopic.php?f=15&t=24862
and report your result there if you please.

So forget version 0.844, and try with C2 REF = 8363 if you want to compare with Milky.
8363 or any other value above 8287 will mean more CPU load, even if it is 1%, so possibly less maximum replay frequency.


I will have a look at the "Aleja" MOD.
Can you please help in locating the pattern and voice where you have that drum pattern.
I assume that this is pattern 0 and 1 and the 2nd out of 6 voices, correct ?
For YM recordings, please use Hatari .

By the way, you can have files comptaible between Octalyser STE and PC.
Take and hex editor, go to offset $438 and change from CD61 (Octalyser only tag) to 6CHN that can be read by all: PC, Octalyser STE, Hextracker ...


Paulo.

[ljbk]

Well,

I have tried Aleja on pattern 00 and voice 1 with either 8287 or 8363 and it sounds correctly with version 0.849A on Hatari 1.62.
Here is what i have done:
1- load your mod;
2- set all voices off except CHAN 1;
3- set VQ to 6.0 (or more if you want) because you only have 1 voice;
4- change C2 REF to 8363 (optional);
5- press PATTERN;
6- set Output Type to YM2149A MONO;
7- set F to 22341 Hz; (you can go up to 43885 Hz !!!)
8- disable all the "goodies": DATA, DIST, VU, DIGISCOPE;
9- set Hatari to record sound;
10- press PLAY;
And i get the attached sound converted to 256Kb MP3 with Audacity 1.3 .


About the 8287 / 8363 business, just by doing a quick google search with "8287 8363 mod" you get this:

http://deku.rydia.net/program/sound3.html
http://forums.modarchive.org/index.php? ... 401.5;wap2
http://code.google.com/p/micromod/sourc ... .java?r=52
http://wikipedia.qwika.com/de2en/Fasttracker_II
http://cd.textfiles.com/darkdomain/faqs ... -part2.txt
http://chipmusic.org/forums/topic/4696/ ... io-issues/

I hope this is enough to convince you .
Hextracker offers you what others don't: you can choose the one you like most or any other value up to 7 times faster samples which means a much higher quality.
This allows to be compatible with Octalyser 20K and 25K modes choosing 10000 and 12500 as Octalyser uses C-3 as reference.


Paulo.

[yerzmyey]

Thanks a lot.
Hmm, I wonder what the difference is.

There are two differences:

- I should set the 8363 value every time I load another MOD (before I used 9855 in the recording - that's what automatically appeared just after loading)
- I should use Hatari to test my stuff (before I used Steem).



> Second the CPU overload.
> With version 0.849, you have a better accuracy. The CPU speed is really measured.
-------------------
Hmmm. Should I understand (as a humble notes-gluer, having no knowledge about frequencies and all the weird stuff) that
HEX844's quality of 22755 Hz is exactly the same what HEX849a+22341Hz is?


Greetzzzzz,
Y



___________
PS: A funny thing, the loop works the best in 8487, heh. On Steem anyway.
So it works perfectly well now, it would seem. I suspect, for new users it would be safer for the prog to remain in standard 'tracker' frequency after loading the MOD. I know I know, no new users of Atari trackers. But... You never know.
Still, the question remains - is the HEX844's quality of 22755 Hz exactly the same what HEX849a+22341Hz is?

[ljbk]

Thanks a lot.
Hmm, I wonder what the difference is.

There are two differences:

- I should set the 8363 value every time I load another MOD (before I used 9855 in the recording - that's what automatically appeared just after loading)
- I should use Hatari to test my stuff (before I used Steem).



> Second the CPU overload.
> With version 0.849, you have a better accuracy. The CPU speed is really measured.
-------------------
Hmmm. Should I understand (as a humble notes-gluer, having no knowledge about frequencies and all the weird stuff) that
HEX844's quality of 22755 Hz is exactly the same what HEX849a+22341Hz is?


Greetzzzzz,
Y



___________
PS: A funny thing, the loop works the best in 8487, heh. On Steem anyway.
So it works perfectly well now, it would seem. I suspect, for new users it would be safer for the prog to remain in standard 'tracker' frequency after loading the MOD. I know I know, no new users of Atari trackers. But... You never know.
Still, the question remains - is the HEX844's quality of 22755 Hz exactly the same what HEX849a+22341Hz is?

Hi !

First, no you don't have to change to 8363.
Only if you want to have EXACTLY the same behaviour as Winamp and Milky Tracker you should use 8363.
If a guy next to you creates a mod with an european Amiga, then you have to use 8287 to get EXACTLY the same frequencies.
If you take that same mod without changing 1 bit and you play it on an american Amiga, it will sound different. It will sound off-key by a 8363/8287 factor which is 15% of a semi-tone (2.5 finetune steps) to put it in a musician perspective.
If an american guy creates a mod on his Amiga and comes to Europe and then if he tries his mod on a European Amiga without changing 1 bit, the same will happen: the music will sound off-key.
Winamp, Milky Tracker, Scream Tracker, and maybe others, have choosen 8363 Hz as base and do not let you change it as far as i know.
But most of the first mods were born in Europe (Karsten Obarski) and a big part of the Amiga scene is in Europe (look at the Revision 2013 winner demo !), so i choosed the european 8287 value. But anyway, the program lets you change it.
The reason for this "strange" Amiga behaviour has to do with the Amiga to be EXACTLY video compatible with the color schemes used in Europe and in the Americas to allow video edition.
As the Paula (the Amiga sounchip) is taken out of the master Amiga clock, the mod reproduction is indeed linked to the video signal.
I do not know which equivalent value is used by Octalyser STE and all other Atari platform digi-tracker programs.
Regarding Octalyser STE, the replay sources available are bit confusing regarding this.

Hatari has MUCH better YM emulation than STeem. All the rest should be equal.
Please post your real ST HW test results at the hardware topic.

No, 22755 with version 0.844 is not the same as 22341 with version 0.849a.
22755 with any version will be the same. Version 0.849a has a better accurate replay so the result can be slightly different.
I choose to replay at 22341 because you wrote it did not work at 22755 but in fact with Hatari it does.

I don't understand the 8487 value.
The quality with version 0.849a will be always better than with 0.844 namely because a small bug in some frequencies that comes from very long ago, was solved.


Paulo.

PS: To replace the CD61 by a 6CHN, to have a compatible file for all, you can also use Hextracker. Load the mod and save it with Alt P.

[ljbk]

Hi Yerzmyey !


I tried to reproduce the error you refered under STeem 3.2 but i did not manage to.
I attach the several mp3 recorded both with STeem and Hatari in both cases for 8287 Hz base and 8363 Hz base.
As you see there is no mixing error and the results are very similar.
I followed all the steps described previously except F is now 22755 Hz.
I think you might have a config problem for STeem, a performance problem at STeem/PC level or you keep the dialog box where you can choose the recording file open. In that case the PC sound buffer will keep repeating the same sound ...
I also attach my STeem sound config. My main PC is a common 2.4 GHz Core 2 Duo.
By the way, i did reproduce completly Alejaste.mod(from the zip) at 22755 Hz with STeem 3.2 and i had no CPU overload.
I have left 0086 in the top left corner for 8287 base and 0077 for 8363 base with version 0.849A. This means that at peak there was 0.2% more CPU use with 8363 base.
Of course, if you changed something to your mod it might not work at 22755 Hz any more. The refered minimum replay frequency in the manual for 6 voices is 20480 Hz.


Enjoy,
Paulo.

PS:
In his available source, Lance uses 7090000 as the Amiga CPU clock frequency. That means his reference value is: 7090000/856(C-2) = 8283 Hz.
Lance players are available as UCDM modules in Octalyser STE: 4 voices/7 bit @ 50 KHz or 8 voices/6bit @ 50 KHz.
But that value is slightly wrong because the Amiga PAL clock is 7093790 and so we get the 8287 value.
This another reason to choose 8287 as default.

[ljbk]

Hi !

A minor update to version 0.849B is available.
It solves some minor but annoying problems with keyboard keys Alt, Control or Shifts after saving or loading using the keyboard.
It also solves a deadlock issue due to the new auto-repeat feature at the end of a replay.

Enjoy,
Paulo.

[yerzmyey]

What I always wanted to do, was to use the Hex-Tracker with my 520ST in a concert.
I suspected the setup will work very well but I've never put it into the test.
Until last weekend.
http://youtu.be/w6L7uMVe3DA
The first song in this short video - it's 520ST with the HEX! (And toy-keyboards of course).

[bullis1]

Very nice performance Yerz!

[yerzmyey]

Only because the soft and the machines were good.

Thx!

[DarkLord]

That was awesome! I D/L'ed it from Youtube so I can have it here to play
whenever I want. Bet it was a real joy to finally get to do that. Were
you controlling the background screen as well, or was that independent?

BTW, can we get those tracks you were playing? I love the start to the
very first one (but all were great!).

Thanks for sharing!

[Scarlettkitten]

What I always wanted to do, was to use the Hex-Tracker with my 520ST in a concert.
I suspected the setup will work very well but I've never put it into the test.
Until last weekend.
http://youtu.be/w6L7uMVe3DA
The first song in this short video - it's 520ST with the HEX! (And toy-keyboards of course).

Very nice setup and performance

Agreed we need the first track to download, I loved it

[yerzmyey]

Heh, thanks a lot and yes - after the gig even some Atari fans approached me and they were asking if it's possible certain songs were gettin' from 520ST. So I was finally happy to show how good the Hex-Tracker is.
Yes, the Atari songs I was playing are here http://ym-digital.i-demo.pl/music.html and the first in the concert was "Time Dilatation". I also played "Quantum Fluctuations", "Alley of Stars" and the "Draconus" cover viewtopic.php?f=19&t=25536 - everything from YM of my 520ST 4Mb.
All the rest was ZX Spectrum and Amiga 1200.
Best wishes.
Yerz

[yerzmyey]

PS: Ah, and the background screen was independent, the "playlist" of demos/games has been prepared earlier. It would be to much to control also this one, heh.

[yerzmyey]

Just wanted to say, that *especially* for the Hex-Tracker I'm buying the Atari TT this month.
I wonder what can 32 Mhz do with quality of sound and number of channels there.

At the moment we just tested the old 6-channels songs, and on ym2149 it gave almost............ 40 khz of the sound-quality!!!! =8-D

Another thing is, that Atari TT has also the DMA, as far as I know, so who knows - it could give most interesting effects!

So who knows, maybe on the next-time-concert I could play something more complex than 6ch only.
http://yerzmyey.i-demo.pl/yerz_setup2.jpg

[DarkLord]

Looking forward to anything you do.

[yerzmyey]

Well, what can I say...
It's aliiiiiiiiiiiiiiiiiiive!

My first ATARI TT / Hextracker song - "aTTraction".

http://yerzmyey.i-demo.pl/Yerzmyey-aTTraction.mp3
(Recorded from real hardware - standard Atari TT, 4Mb of RAM).

16 channels - I knew the machine was powerful - and it surely is.

Hex-Tracker 849a.
Setup: DMA / MONO / 25x50
volume: 6.

The MOD itself: http://yerzmyey.i-demo.pl/aTTraction_16 ... er849a.rar