| View previous topic :: View next topic |
| Author |
Message |
Groupie in Training

Joined: 29 Dec 2005 Age: 32 Posts: 112 Location: Helsinki
|
Posted: Fri Aug 01, 2008 2:53 pm Post subject: V19 released |
|
|
I have released V19.
It is an optimization and finetuning of the work that went into V18, and some new ideas and resurrected old ideas, as described in my previous posting.
I have yet work cut out to properly tune this patch and experiment with a lot of combinations of various codepaths and the tunables. V19 contains the parts that I'm relatively sure of. Testing and listening with all the variations is likely to take a long time, so here is V19 in the meantime.
Compared to V18:
- reworked distortion algorithm. A must. Makes it possible to enjoy fist 2 cave, snake disco and hard track at the same time.
- tuned the state variable mixup code. I remove bp-lp link and weakened hp-bp link. This is the most unreliable part at the moment, but it will have to do. I need something to keep certain harsh distortion effects at bay.
- added output strip mixing effect, although I kept it weaker than it is on the real chip in order to not shock you. Real chip sound is impossibly dirty in places.
- added 3 dB to the filter bass output level. |
|
| Back to top |
|
 |
Groupie


Joined: 11 Feb 2002 Posts: 265
|
Posted: Fri Aug 01, 2008 5:40 pm Post subject: Re: V19 released |
|
|
What makes the drum beats in Snake Disco sound so "flat" in the recording from a real R4AR? As far as I can hear, this effect is not emulated at all. _________________ C64C + 1541-II, or Hoxs64 as an accurate equivalent.
Visit hvsc.c64.org for the most comprehensive SID collection. |
|
| Back to top |
|
 |
Groupie in Training

Joined: 29 Dec 2005 Age: 32 Posts: 112 Location: Helsinki
|
Posted: Fri Aug 01, 2008 7:22 pm Post subject: Re: V19 released |
|
|
| Magnate wrote: | | What makes the drum beats in Snake Disco sound so "flat" in the recording from a real R4AR? As far as I can hear, this effect is not emulated at all. |
A broken noise generator. :(
I know I should change to a working 6581 chip. I have two with broken noise generators now, and a third that should still work 100 %. I just try to keep this in mind when I compare the recordings. |
|
| Back to top |
|
 |
Groupie in Training

Joined: 29 Dec 2005 Age: 32 Posts: 112 Location: Helsinki
|
Posted: Sat Aug 02, 2008 1:43 pm Post subject: Problems in V19 |
|
|
ZrX-oMs dug up some problems in V19. To address these issues, V19 can be changed as I describe. (The patch on my server has already been updated, so in case you have applied it, here is the instructions to update your patched copy to match.) In the Filter::clock() function you have this:
| Code: |
float diff_hp_bp = (Vhp - Vbp) * material_cft * 2.f;
|
Replace by:
| Code: |
float diff_hp_bp = (Vhp - Vbp) * material_cft;
float diff_lp_bp = (Vlp - Vbp) * material_cft;
|
20 lines down or so you have this:
| Code: |
Vhp -= diff_hp_bp;
Vbp += diff_hp_bp;
|
Add after those two lines these two lines:
| Code: |
Vlp -= diff_lp_bp;
Vbp += diff_lp_bp;
|
Then update DistortionRate from 8.2e-3 to 8.4e-3, and DistortionCFThreshold from 6e-5 to 1e-4. (I suspect both of these values should be higher yet, but I'm afraid to change these quickly because there is a real danger of regressing somewhere.)
Observant among you identify that these changes are exactly on the part which I described as "weakest" in my previous posting. I was also able to increase distortion a bit, and I also strongly increased the output strip backmixing, which produces horrendous but quite chiplike results for many a song. (I really hate this effect, but it is on the chip so what can you do.) |
|
| Back to top |
|
 |
Groupie


Joined: 10 Dec 2007 Age: 37 Posts: 221 Location: Berlin
|
Posted: Sun Aug 03, 2008 6:36 pm Post subject: |
|
|
Next Beta is available:
Biggest feature is the PSID64 integration!
- context menu in the HVSC/CGSC tab to
1. add files directly to the playlist
2. convert files to executable c64 programs (PSID64 is now in)
- context menu in the Favorites tab to
1. move/copy files from one playlist to another
2. convert files to executable c64 programs (PSID64)
This can be used for all selected favorites at one time.
- Java memory consumtion is displayed in the status line
@Nata: What conversions are needed else, you said something about stripping the header in the past, what was that?
Last edited by kenchis on Tue Aug 05, 2008 7:36 am; edited 1 time in total |
|
| Back to top |
|
 |
Groupie in Training


Joined: 18 Apr 2002 Posts: 105 Location: Hertford, Herts
|
Posted: Mon Aug 04, 2008 10:47 pm Post subject: Mobiles |
|
|
Hey.. fantastic little proggy, not that I can get it running in Windows...but..
just wondering if it could re-tooled and made available for Symbian OS mobiles S60v3 ? like the Nokia N95 and such like... _________________ C128D King |
|
| Back to top |
|
 |
Groupie


Joined: 10 Dec 2007 Age: 37 Posts: 221 Location: Berlin
|
Posted: Tue Aug 05, 2008 7:35 am Post subject: |
|
|
I think there could be some difficulties to get it running on a mobile (size, performance), but it can be used on a IEEE pc.
Perhaps someday there is a version for Android, too (not planned yet).
What was the problem for windows? |
|
| Back to top |
|
 |
Groupie in Training

Joined: 29 Dec 2005 Age: 32 Posts: 112 Location: Helsinki
|
Posted: Wed Aug 06, 2008 11:27 am Post subject: Re: Mobiles |
|
|
| Stainless wrote: | | Hey.. fantastic little proggy, not that I can get it running in Windows...but.. just wondering if it could re-tooled and made available for Symbian OS mobiles S60v3 ? like the Nokia N95 and such like... |
At the moment the performance demands are forbidding. The code requires FPU with IEEE floating point, which eliminates most of mobile platforms outright, which generally do not have hardware FPU units.
The C++ version consumes about 25 % of single core of AMD64X2 4200+, 33 % of single core of Turion TK-55, 50 % of a P4M 2 GHz, and 20 % of E6750. Depending on the CPU architecture, fully third to half of the time is spent on resampling, the rest is wasted on the analog/digital emulation, which clocks the chip parts at 1 MHz clock.
The Java version here consumes about 3x the time of C++ version at comparable output quality, although I have not yet tried kenchis's suggestion of java server mode. At any rate, it is likely that the desktop CPUs run java better than mobile platforms will, and full port to Symbian might require substantial work. I am not familiar with Symbian, but it is probably much more work than just a cross-compile.
I believe that I could reduce the analog simulation CPU drain by clocking it at a reduced rate, say, 250 kHz instead of 1 MHz. The impact on quality ought to be quite neglible, too, even if I do worry a bit about aliasing effects arising from the distortion code's inevitable spectral expansion. (At any rate, such effects should be controlled even on today's version by employing some type of low-pass filter before the data enters the filter proper, and I'm not doing that now.) |
|
| Back to top |
|
 |
Groupie in Training


Joined: 18 Apr 2002 Posts: 105 Location: Hertford, Herts
|
Posted: Wed Aug 06, 2008 7:49 pm Post subject: Re: Mobiles |
|
|
| alankila wrote: | | Stainless wrote: | | Hey.. fantastic little proggy, not that I can get it running in Windows...but.. just wondering if it could re-tooled and made available for Symbian OS mobiles S60v3 ? like the Nokia N95 and such like... |
At the moment the performance demands are forbidding. The code requires FPU with IEEE floating point, which eliminates most of mobile platforms outright, which generally do not have hardware FPU units.
The C++ version consumes about 25 % of single core of AMD64X2 4200+, 33 % of single core of Turion TK-55, 50 % of a P4M 2 GHz, and 20 % of E6750. Depending on the CPU architecture, fully third to half of the time is spent on resampling, the rest is wasted on the analog/digital emulation, which clocks the chip parts at 1 MHz clock.
The Java version here consumes about 3x the time of C++ version at comparable output quality, although I have not yet tried kenchis's suggestion of java server mode. At any rate, it is likely that the desktop CPUs run java better than mobile platforms will, and full port to Symbian might require substantial work. I am not familiar with Symbian, but it is probably much more work than just a cross-compile.
I believe that I could reduce the analog simulation CPU drain by clocking it at a reduced rate, say, 250 kHz instead of 1 MHz. The impact on quality ought to be quite neglible, too, even if I do worry a bit about aliasing effects arising from the distortion code's inevitable spectral expansion. (At any rate, such effects should be controlled even on today's version by employing some type of low-pass filter before the data enters the filter proper, and I'm not doing that now.) |
It sure would be nice to have a capable sid player on SymbianOS... we are starved at the moment... I`m running scenetone.. bit rubbish at sid... good at mod, xm and such like.. _________________
C128D King |
|
| Back to top |
|
 |
Groupie in Training


Joined: 18 Apr 2002 Posts: 105 Location: Hertford, Herts
|
|
| Back to top |
|
 |
Groupie


Joined: 10 Dec 2007 Age: 37 Posts: 221 Location: Berlin
|
Posted: Wed Aug 06, 2008 8:12 pm Post subject: |
|
|
Call for console Version is at least:
"java.exe" -classpath jsidplay2.jar sidplay.ConsolePlayer
But to run in the fastest mode you should change sampling method in the ini file after the first start:
~/sidplay2.ini:
Sampling=0 |
|
| Back to top |
|
 |
Groupie in Training


Joined: 18 Apr 2002 Posts: 105 Location: Hertford, Herts
|
|
| Back to top |
|
 |
Groupie in Training

Joined: 29 Dec 2005 Age: 32 Posts: 112 Location: Helsinki
|
Posted: Fri Aug 08, 2008 1:34 am Post subject: |
|
|
| Stainless wrote: | | kenchis wrote: | Call for console Version is at least:
"java.exe" -classpath jsidplay2.jar sidplay.ConsolePlayer
|
I`ve tried this from the run box... nothing.. a dos box pops up for a second or two and then nothing :( |
Open the command prompt proper, maybe you can tell us what it says when it quits. If there is an error, it would be quite enlightening to hear what it is. |
|
| Back to top |
|
 |
Groupie in Training


Joined: 18 Apr 2002 Posts: 105 Location: Hertford, Herts
|
Posted: Fri Aug 08, 2008 5:52 pm Post subject: |
|
|
Microsoft Windows XP [Version 5.1.2600]
(C) Copyright 1985-2001 Microsoft Corp.
C:\Documents and Settings\JZD>"java.exe" -classpath jsidplay2.jar sidplay.Consol
ePlayer
Exception in thread "main" java.lang.NoClassDefFoundError: sidplay/ConsolePlayer
Caused by: java.lang.ClassNotFoundException: sidplay.ConsolePlayer
at java.net.URLClassLoader$1.run(Unknown Source)
at java.security.AccessController.doPrivileged(Native Method)
at java.net.URLClassLoader.findClass(Unknown Source)
at java.lang.ClassLoader.loadClass(Unknown Source)
at sun.misc.Launcher$AppClassLoader.loadClass(Unknown Source)
at java.lang.ClassLoader.loadClass(Unknown Source)
at java.lang.ClassLoader.loadClassInternal(Unknown Source)
C:\Documents and Settings\JZD>
Thats what I get ... _________________ C128D King |
|
| Back to top |
|
 |
Groupie


Joined: 10 Dec 2007 Age: 37 Posts: 221 Location: Berlin
|
Posted: Fri Aug 08, 2008 5:59 pm Post subject: |
|
|
Try:
"java.exe" -classpath "C:\Documents and Settings\JZD\jsidplay2.jar" sidplay.ConsolePlayer
It works, if the file jsidplay2.jar is located in that directory. |
|
| Back to top |
|
 |
Groupie in Training


Joined: 18 Apr 2002 Posts: 105 Location: Hertford, Herts
|
|
| Back to top |
|
 |
Groupie


Joined: 10 Dec 2007 Age: 37 Posts: 221 Location: Berlin
|
Posted: Fri Aug 08, 2008 6:21 pm Post subject: |
|
|
Now you can try to get the jsidplay2 java certificate,
simply try to start with java server vm, the proper memory settings and the gui version with additional jar files  |
|
| Back to top |
|
 |
Groupie in Training


Joined: 18 Apr 2002 Posts: 105 Location: Hertford, Herts
|
Posted: Fri Aug 08, 2008 6:26 pm Post subject: |
|
|
No idea what you mean... start java server?
...think I`m just going to go back to the windows version...
.. _________________ C128D King |
|
| Back to top |
|
 |
Groupie


Joined: 10 Dec 2007 Age: 37 Posts: 221 Location: Berlin
|
Posted: Fri Aug 08, 2008 6:29 pm Post subject: |
|
|
I was just kidding
Your last try looks quite good, simply add the file you wanna play to the command line. |
|
| Back to top |
|
 |
Groupie in Training


Joined: 18 Apr 2002 Posts: 105 Location: Hertford, Herts
|
Posted: Fri Aug 08, 2008 6:37 pm Post subject: |
|
|
Ok.. so I have to do that for each and every tune I want to play ?
Don`t I get a gui where I can select my sids? _________________ C128D King |
|
| Back to top |
|
 |
Groupie


Joined: 10 Dec 2007 Age: 37 Posts: 221 Location: Berlin
|
Posted: Fri Aug 08, 2008 6:54 pm Post subject: |
|
|
Try this URL in the browser:
This is the beta version with a GUI:
http://jsidplay2.sourceforge.net/jsidplay2.html |
|
| Back to top |
|
 |
Immortal Grandmaster of C64


Joined: 06 Dec 2004 Age: 40 Posts: 5801 Location: Toronto, CANADA
|
|
| Back to top |
|
 |
Groupie


Joined: 10 Dec 2007 Age: 37 Posts: 221 Location: Berlin
|
Posted: Fri Aug 08, 2008 7:04 pm Post subject: |
|
|
| gklinger wrote: | | Stainless wrote: | | Microsoft Windows XP [Version 5.1.2600] |
There's your problem.  |
What what what??? I didnt get it.
Please help me out
Edit: Hehehe now i got it it seems i did not see the smiley |
|
| Back to top |
|
 |
Groupie


Joined: 10 Dec 2007 Age: 37 Posts: 221 Location: Berlin
|
Posted: Sat Aug 09, 2008 12:29 pm Post subject: |
|
|
Here comes a bugfix release:
- Refactoring of the applet (internal changes)
- Save all settings automatically on exit
(Button "Save Config" is removed)
- sidid updated
- selected filter and distortion enable is now saved correctly
- distortion settings were not rounded
(a GUI bug)
- Favorite Tabs are saved now (and restored on next start-up)
- Change settings before loading the first tune was buggy |
|
| Back to top |
|
 |
Groupie


Joined: 11 Feb 2002 Posts: 265
|
Posted: Sat Aug 09, 2008 2:14 pm Post subject: |
|
|
Can you please put a download link to the JAR file in the first post? Thanks. _________________ C64C + 1541-II, or Hoxs64 as an accurate equivalent.
Visit hvsc.c64.org for the most comprehensive SID collection. |
|
| Back to top |
|
 |
|