#153 Some questions about the Audio (SID) menu in Denise v2.8

[Anonymous]

Hello. I have a question about the SID emulation in Denise V2.8. Since I don't always update to every single new version of Denise, there have now been some changes in the Audio (SID) menu, compared to my older V2.3 version, about which I have a few questions. Maybe you can help me there?

1. The "external filtering" function has now disappeared. Why did it come away, because I always thought it was a good feature and is this function now permanently switched on or off?

2. The "Filter Volume Adjust" function has also disappeared. I assume, this function is now permanently activated, because the SID filters all sound nearly the same loud, when I switch between them. Has it a special reason, why this function was taken out?

3. Is it still recommended, to set the SID8580 filter bias to "-3000", if the normal "ReSID" SID-filter is used, or should it now be set to the middle of the scale (i.e. to "0", or according to that new scale, then to 0.5000)?

4. There are changes in the way, the "sampling intervall" function is shown in the menu now. What intervall in the old scale is it, when I set this function to "medium" now? Is it 7 or 18 or something between these values, of the old scale?

5. What of the three selectable options (average, weak, strong) is recommended, for the new function "wave strength", when the new ReSID-fp SID-filter is used and what this function exactly does, when the user has chosen this new filter type, because the other three filters don't even have this feature?

6. It was always the case in Denise, that you could switch between the SID filters directly on the fly, without that the sound was gone. With these old three SID filters "ReSID, ReSID-Old and Chamberlin" this is still the case for me, but if I switch to ReSID-fp, then the sound is always gone for me and I have to reset and reload the running program first, so that there is sound again. Is this normal and if so, can this be changed, so that you can also switch to this new SID filter on the fly? It was always a great thing, to directly compare the sound of all SID filters and to do so, the best always was, to directly switch between them, in a running song. Therefore it's a bit sad, that this doesn't seem to work with ReSID-fp, as it does with the other three filters. In addition I found out, that all the four selectable SID-filters sound nearly the same to me, in Denise v2.8. Don't know if that's normal? Well, maybe I'll have to try a few special SID tunes, that could possibly cause problems for one or the other filter, maybe then I'll hear a real difference?

7. And the last question, which is about something else. I have noticed in v2.8, that the content in the individual Denise sub-menus appears a bit slower, than in my previous v2.3 version. So it sometimes takes about a second, until I see the content of a newly opened menu, which is sometimes a bit annoying or unsightly, I must say. What could be the reason for this? Or is there perhaps some function, in the menu of Denise, that I haven't found yet, to correct this again?

Sorry for the many questions, but there was a lot, that had accumulated now. lol Thanks in advance, cheers

Clemens

[PiCiJi]

1. always on. It's part of C64 hardware (outside of SID). I have never noticed a tune where an audible difference exists. The filter limits the frequency range to between 16 Hz and 15.9 kHz. There is no reason to want to allow frequencies outside this spectrum.

2. I imagine every user would prefer not to have to adjust the volume when switching between SID types. The goal is to avoid overwhelming the user with too many functions—functions where we already know what the user is going to choose.

3. The new scale is better suited for reSID and residfp. In the new scale, -5000 (old) corresponds to 0, and 5000 (old) corresponds to 1.0 .... -4000 would be 0.1, -3000 would be 0.2 and 0 would be 0.5, 3000 would be 0.8

4. highest: 1 before, high: 2 before, medium: 7 before.

5. Please ask the author of libresidfp about that. These appear to be the characteristics of different SID sub types. It would actually be helpful to know of a tune where a difference is audible ... just to test whether the option in Denise works

6. To make this possible, the internal state of the reSID emulation would have to be converted to that of residfp. The author, Leandro Nini, is unlikely to undertake such an effort. Moreover, the state is never 100% transferable. Typically, the situation resolves itself when a new tune plays; otherwise, a reset is required.
   To spot real differences, you really need to get hold of the right tunes and have some knowledge of which deviations are—or aren't—normal on a real SID chip.

7. As the number of widgets increases, the initialization time goes up. However, this only happens once per selected tab, provided the emulator remains open. Unfortunately, refactoring the code to improve initialization speed would be a major undertaking. I expect this can only be improved if I use more modern frameworks like WPF or QT, instead of the old WinAPI. I also don't see much point in doing so, given that the delay only occurs during the initial UI TAB selection.

[Anonymous]

Thanks for answering my questions, now everything, regarding the Audio menue, is clear again. I think, I should update to every new version of Denise from now on, because otherwise, you suddenly end up with too many changes all at once, in the menues and one partly loses track of things.

Cheers,
Clemens

[Anonymous]

Sorry, now a small question has come up again, about my point #3, so I wanted to double-check this. Back in Denise V2.3, the 8580 bias value for the ReSID filter was preset to -3000 and I never changed anything on this. In Denise V2.6 however, this bias for the ReSID filter was preset to 0.5000 now, which is exactly the middle of the scale and when converted to the old scale, that would essentially be a value of 0 in Denise V2.3 and not -3000. The default setting has changed here.

What is the recommended 8580 bias value now, for the ReSID filter? Is it recommended, to set it back to 0.2000 in Denise V2.6 now, because that would be -3000, which was preset in V2.3's old scale, or should I leave it at 0.5000 in V2.6? Or has something changed on the ReSID filter itself, in the meantime, resulting in a different default setting of this bias value in Denise V2.6, compared to the older V2.3?

Cheers,
Clemens

[PiCiJi]

In addition I found out, that all the four selectable SID-filters sound nearly the same to me, in Denise v2.8. Don't know if that's normal?

Keep in mind that you need a tune that uses the filter. Without the filter, the SID operates largely digitally, and it is normal not to hear any difference in sound. Differences between the models are audible only in tunes that utilize the filter. You can quickly determine whether a song uses the filter by briefly switching it off; if the sound changes, the filter is active.

What is the recommended 8580 bias value now, for the ReSID filter?

Denise is using the latest standard value. I don't have an overview of how it has changed over time.
if you press the button (back) next to the slider, the current default value is activated.

[Anonymous]

Okay, thanks.
Cheers,
Clemens

[Anonymous]

Another user here. Just read this here and I have a hunch, about the likely cause of the problem mentioned your point 7, Clemens, because with the latest versions of Denise, I have exactly the same annoying problem and the menus in the emulator take noticeably longer, to fully appear.

Up until version 2.6, there was a 'Threaded Emulation' function, under options/settings and I always had to disable this feature. Doing so, made the menus load lightning-fast, here on my machine. Unfortunately, that option has disappeared sometime between version 2.6 and 2.7 and I guess, that this 'Threaded Emulation' is presumably always active now. That’s likely the reason of this problem, you mentioned. Unfortunately, some good functions, which allowed good performance on older computers and always ran flawlessly, have been removed from Denise recently and in this regard, the program now seems to be following the same path as VICE, which is pretty sad.

Doc. Oc.

[PiCiJi]

Using a UI thread should not delay the construction of the UI. If it does, the computer must be very old—featuring only a single core or a Celeron processor.
Denise uses a separate UI thread, as is common practice nowadays. Features such as the debugger rely on this. In terms of development effort, it is no longer feasible to offer a single-threaded variant as well.

[Anonymous]

On my computer, this function definitely affected, how quickly the menu contents appeared, when i switched between the menus in Denise. It’s not a Celeron, but 'only' a dual-core machine from 2012. Still, it has always been sufficient for emulation, up to the Playstation1.

You wrote it would be no longer feasible. But what I don’t quite understand here is this - the single-thread feature was already there and worked perfectly. So, why remove something like that? The vast majority of users don’t even use the debugger anyway, because they don't know anything about programming and debugging and only want to run software in the emulator. Besides, there could also be a hint added, indicating that the function needs to be enabled, if someone actually wants to use the debugger (something 95% of users will never do).

Regards,
Doc. Oc.

[PiCiJi]

I’m sorry that the graphical interface’s initialization time is causing you trouble. I imagine emulators like Hoxs or CCS64 are better suited to your system. Well, they don’t get many updates anymore—just like your system.