| |
Compyx
Registered: Jan 2005
Posts: 631 |
Release id #212349 : VICE 3.6
Since there are some questions in the comments section of VICE 3.6, I figured it would be better use the forum for any questions.
So to start things off:
Quoting The Human Code MachineBut how can I save the monitor window position on GTK3 Vice Windows x64? After exiting it always opens the same small window.
Currently you can't, window resizing/repositioning support is different across OSes/WMs, so making this work properly for all configurations may never work. That said, I'm working on a "fix" to make it at least work on some systems. |
|
| |
Remdy
Registered: Feb 2019
Posts: 26 |
is there a changelog? |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
its even pasted into production info.... |
| |
Compyx
Registered: Jan 2005
Posts: 631 |
I've implemented storing/restoring the monitor window position and size. Works on my Debian box with Gnome and Wayland, no idea if it will work on other systems, but I'm not using any OS/WM-specific hacks (yet), just plain Gtk calls.
Windows builds are here: https://github.com/VICE-Team/svn-mirror/releases/tag/r41413
Linux, BSD and MacOS users will have to build VICE themselves from current trunk. |
| |
The Human Code Machine
Registered: Sep 2005
Posts: 110 |
Thx a lot Compyx for this nice Christmas gift! Tested on Windows 11 and seems to be working as it should. It was the one feature I missed from the old native Windows Vice. |
| |
hedning
Registered: Mar 2009
Posts: 4596 |
This version seems to be as good as 2.4 for Windows! This is a good thing. The flow, sound and everything seems about right. Everything above 2.4 was a pain until now. Good work! |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
It should be much better than any 2.4 ever was actually :) |
| |
Frantic
Registered: Mar 2003
Posts: 1627 |
@gpz: when ppl say this they obviously refer to the user experience, including the interface. Not the accuracy of the emulation itself. |
| |
Raistlin
Registered: Mar 2007
Posts: 559 |
Frantic, I think most people’s problem with versions above 2.4 were that the SID emulation wasn’t quite right and that the framerate just wasn’t smooth.. so while emulation was more accurate, technically, the loss of smooth 50Hz playback spoilt all of that effort.
For me, working on demos, I would always run on latest VICE regardless… if I heard an audio “pop” or “hiss”, or an effect didn’t look smooth (tiny circle scrollers look awful when not run at a smooth framerate, for example), I would check on other emulators - HOXS, Denise etc.
Everything I’m hearing about 3.6 is good … so perhaps I won’t need to revert to other emulators in future.
(I do also test on final hardware of course - but I develop using VICE) |
| |
Compyx
Registered: Jan 2005
Posts: 631 |
Quoting The Human Code MachineThx a lot Compyx for this nice Christmas gift! Tested on Windows 11 and seems to be working as it should. It was the one feature I missed from the old native Windows Vice.
You're welcome! Good to know it works on Windows 11. It also works on my Windows 7 VM, so there's a decent chance it works on other Windows version as well. |
| |
hedning
Registered: Mar 2009
Posts: 4596 |
Quote: Frantic, I think most people’s problem with versions above 2.4 were that the SID emulation wasn’t quite right and that the framerate just wasn’t smooth.. so while emulation was more accurate, technically, the loss of smooth 50Hz playback spoilt all of that effort.
For me, working on demos, I would always run on latest VICE regardless… if I heard an audio “pop” or “hiss”, or an effect didn’t look smooth (tiny circle scrollers look awful when not run at a smooth framerate, for example), I would check on other emulators - HOXS, Denise etc.
Everything I’m hearing about 3.6 is good … so perhaps I won’t need to revert to other emulators in future.
(I do also test on final hardware of course - but I develop using VICE)
Yes. Exactly. User experience. I want the emulator to work as a C64 would. I don't care much for the start cable exoplanetary matrix half pixel cycle exact cartridge code monitors, or wide border magma chamber mumbo jumbo code monkey settings. I just want a smooth and great C64 experience when emulating. And I haven't had that since Vice 2.4. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
except 2.4 never had as stable framerate or as low audio/video latency as 3.6 has? :) (and i dont buy the "user experience" thing myself. i find the menu structure on 2.4 truely horrible straight from hell, i'd ditch it for the central config dialog alone.) |
| |
hedning
Registered: Mar 2009
Posts: 4596 |
Quote: except 2.4 never had as stable framerate or as low audio/video latency as 3.6 has? :) (and i dont buy the "user experience" thing myself. i find the menu structure on 2.4 truely horrible straight from hell, i'd ditch it for the central config dialog alone.)
I just meant that the feel of Vice emulation on Windows got worse after 2.4 (sid not keeping up, farts started, jerkiness, extremely slow and so on) and now it seems fixed. Great! Menus in 3.1 was nice. Haven't gotten used to these new ones yet, but seems logical and nice. :) |
| |
JackAsser
Registered: Jun 2002
Posts: 1989 |
This might be a stupid question, but how do you pass cmd line parameters to VICE these days on MacOS?
/Applications/vice-x86-64-gtk3-3.6.0/x128.app/Contents/MacOS/x128 --help does exactly nada, netither does open -a /Applications/vice-x86-64-gtk3-3.6.0/x128.app --args -help |
| |
Krill
Registered: Apr 2002
Posts: 2844 |
There's a "bin" folder next to the app folders, i believe, containing the CLI incarnations of the programs.
Worked for me, although something was always quite off with the working directory. |
| |
Frantic
Registered: Mar 2003
Posts: 1627 |
Quote: This might be a stupid question, but how do you pass cmd line parameters to VICE these days on MacOS?
/Applications/vice-x86-64-gtk3-3.6.0/x128.app/Contents/MacOS/x128 --help does exactly nada, netither does open -a /Applications/vice-x86-64-gtk3-3.6.0/x128.app --args -help
@Jackasser: Yes, what Krill says. This changed in one of the more recent versions. I had the same problem at first. |
| |
JackAsser
Registered: Jun 2002
Posts: 1989 |
Quote: @Jackasser: Yes, what Krill says. This changed in one of the more recent versions. I had the same problem at first.
Thanks, that worked! |
| |
Jammer
Registered: Nov 2002
Posts: 1289 |
GTK Vice 3.6 starts with disabled sound output regardless of saved settings. Any idea why? 3.5 didn't manifest anything like this. |
| |
Compyx
Registered: Jan 2005
Posts: 631 |
Might have something to do with the audio code disabling sound on buffer underruns when the buffer size and fragment size are too 'tight', I think on MacOS these settings are even tighter than on Windows/Unix.
So you could try to increase these settings, especially the buffer size, save settings and try running the emu again.
Edit: make sure you've re-enabled sound in the UI before saving. |
| |
Dr.Science
Registered: Oct 2011
Posts: 39 |
Same sound issue as Jammer here also with GTK Vice 3.6 (Win11). |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
please try latest build from here: https://github.com/VICE-Team/svn-mirror/releases and see if the sound problem is gone |
| |
Jammer
Registered: Nov 2002
Posts: 1289 |
I can confirm that 64bit build 3.6.0 r41424 (Win 10) got rid of aforementioned sound issue. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
great!
now go to sound settings, set fragment size to smallest, and buffer to something like 3ms (or less!). i cant believe how small the latency can be now - even on my 8 year old box |
| |
Jammer
Registered: Nov 2002
Posts: 1289 |
With dx driver and smallest fragment size I can go down to 11ms of latency. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
yeah, i just learned that this 1ms isnt quite what i thought due to how the linux sound code works right now. mea culpa :) |
| |
Dr.Science
Registered: Oct 2011
Posts: 39 |
Thanks, sound-issue gone also here! |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
Did anyone experience crashes with this release? Jazzcat reported something like this but is busy with other things now apparently - having someone else who can reproduce it would be nice :) |
| |
Mace
Registered: May 2002
Posts: 1799 |
Is there a way to "virtually" attach a harddisc as IEC device to VICE? |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
? |
| |
wil
Registered: Jan 2019
Posts: 42 |
Thanks for making Vice 3.6, it looks very promising!
But how can I view the disassembly window and the processor status window? On version 3.2 (Win32), there are several buttons on top of the monitor window for enabling these views, but I could not find these buttons on the 3.6 version. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
That was a feature that only existed in the old Windows UI |
| |
Mace
Registered: May 2002
Posts: 1799 |
@Groepaz: I'm testing a C64 ool that allow the possibility to attach a harddisc, but I have no idea what technical specs it assumes, so I wanted to try. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
VICE emulates the CMD HD, IDE64 and some IEEE harddisk - you gotta know what you need and set it up accordingly :) (You'll need the ROMs for it) |
| |
Mace
Registered: May 2002
Posts: 1799 |
Thanks, when I get to further testing, I'll see how I'll get the ROMS and how it's set up. No priority at this point. |
| |
Youth
Registered: Aug 2003
Posts: 40 |
On macOS I have no video drivers available for recording media (GTK version). I have ffmpeg installed with 'brew install ffmpeg'. Is there a way to make video recording work?
Logs:
Graphics Output: Error - Requested graphics output driver FFMPEG not found.
Warning - Trying to read value from unknown resource `FFMPEGFormat'.
Warning - Trying to read value from unknown resource `FFMPEGVideoCodec'.
Warning - Trying to read value from unknown resource `FFMPEGAudioCodec'.
Warning - Trying to read value from unknown resource `FFMPEGVideoBitrate'.
Error - failed to get current value for resource 'FFMPEGVideoBitrate', defaulting to 0.
Warning - Trying to read value from unknown resource `FFMPEGVideoBitrate'.
Error - failed to get value for resource 'FFMPEGVideoBitrate'
Warning - Trying to read value from unknown resource `FFMPEGAudioBitrate'.
Error - failed to get current value for resource 'FFMPEGAudioBitrate', defaulting to 0.
Warning - Trying to read value from unknown resource `FFMPEGAudioBitrate'.
Error - failed to get value for resource 'FFMPEGAudioBitrate'
Warning - Trying to read value from unknown resource `FFMPEGVideoHalveFramerate'.
Error - invalid resource name 'FFMPEGVideoHalveFramerate' |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
looks like dqh forgot to enable ffmpeg in the macos build =P oops |
| |
Compyx
Registered: Jan 2005
Posts: 631 |
All the ffmpeg dylibs appear to be in the Mac dist, so can someone check "Help" -> "Compile time features", and see what it says for HAVE_FFMPEG? |
| |
Rastah Bar
Registered: Oct 2012
Posts: 336 |
r41431:
HAVE_FFMPEG yes
HAVE_FFMPEG_AVRESAMPLE no
HAVE_FFMPEG_HEADER_SUBDIRS yes
HAVE_FFMPEG_SWRESAMPLE yes
HAVE_FFMPEG_SWSCALE yes |
| |
Compyx
Registered: Jan 2005
Posts: 631 |
That's the same as my Linux build. That means VICE is configured for ffmpeg and the video recording tab should work. Odd. |
| |
Youth
Registered: Aug 2003
Posts: 40 |
Sorry wasn't accurate enough, I am using the Apple Silicon Mac (arm64) version. This version does not let me select a video driver (empty dropdown). It says HAVE_FFMPEG yes in the Help section.
I have checked though; this doesn't seem to be an arm64 binary, this is what I did:
lipo -archs x64sc.app/Contents/MacOS/x64sc
The result: x86_64, where it should say arm64
So I tried the Intel Silicon Mac version, this also reports x86_64 architecture, but the video recording works on this version!
For now I will use the Intel (x86_64) version. My two cents from this:
1. The Apple Silicon version on the website it not actually an arm64 binary. It is an Intel x86_64 binary. It will run through the Rosette emulation layer and not make use of the arm64 advantages (speed/stability).
2. You could offer a universal binary with both architectures in it. I did this for SIDFactory II, maybe I can be of some help. |
| |
dqh
Registered: Jun 2019
Posts: 46 |
The binary is ARM - but the wrapper/launcher .app binary is not. The wrapper is generated with platypus, which currently always generates x86 apps. However this is is a launcher, the real binary is tucked away inside VICE.app.
I've replicated the ffmpeg thing on my M1 - I'll look into it and fix it for VICE 3.6.1. |
| |
dqh
Registered: Jun 2019
Posts: 46 |
Quote: There's a "bin" folder next to the app folders, i believe, containing the CLI incarnations of the programs.
Worked for me, although something was always quite off with the working directory.
I've fixed the working directory thing btw. The bin/ launchers will now preserve the current pwd, and the .app launchers will set it to your home directory instead of a folder within VICE.app :)
These changes will be in 3.6.1 but you can get the new behaviour from the r41431 builds at https://sourceforge.net/projects/vice-emu/files/experimental%20.. |
| |
JackAsser
Registered: Jun 2002
Posts: 1989 |
Quote: I've fixed the working directory thing btw. The bin/ launchers will now preserve the current pwd, and the .app launchers will set it to your home directory instead of a folder within VICE.app :)
These changes will be in 3.6.1 but you can get the new behaviour from the r41431 builds at https://sourceforge.net/projects/vice-emu/files/experimental%20..
Thank you for your mac efforts, highly appreciated! 🙏 |
| |
MagerValp
Registered: Dec 2001
Posts: 1055 |
Quoting dqhThe wrapper is generated with platypus, which currently always generates x86 apps.
In case you haven't gotten around to it yet, I just tried Platypus 5.4 r1217 built from the current GitHub master branch and verified that it generates x86_64 + arm64 apps. |
| |
Erhan
Registered: Feb 2005
Posts: 17 |
i'm curious if these bugs only happen in my system...
win10pro 19043.1415 locale TR lang TR,
x64sc.exe,
emulation screen appears in windowed mode as usual, after i click CRT or Mixer controls at lower panel, window is expanded. Unclicking them, gray bars appear above & below emulation screen.
vsid.exe,
buttons below playlist, only
add tune to playlist and clear playlist buttons work. others not. |
| |
Compyx
Registered: Jan 2005
Posts: 631 |
That happens on all systems.
The gray bands appear because Gtk doesn't resize the window after the CRT and/or mixer widgets are hidden. And since both the main menu bar and the status bar are set to non-expanding, it expands the rendering area. I used to resize the window to 1x1 pixels when disabling a CRT or mixer widget, causing Gtk to minimize the space used for the window, but that obviously didn't work well for people who had their window manually stretched to a certain size.
Throw in the fact that window decorations are involved in getting/setting window sizes and Gtk doesn't know about them (the window manager adds those) and things get hairy quickly.
I've added a hotkey to restore the display to its minimum size (Alt+R) to alleviate the problem at little bit, quicker than manually dragging the window to its minimum size.
The VSID playlist is still a work in progress, and low priority. My focus (heh) is on the proper emulators, especially x64sc. I'll work on VSID again after we've released 3.6.1. |
| |
Rastah Bar
Registered: Oct 2012
Posts: 336 |
I tried the sinchai joystick adapter on MacOs Monterey 12.1 with Vice 3.6. The directions work, but the fire button shows a blue square in Vice when pressed, instread of a red square. But fire is not detected by any games.
This may very well be a Sinchai+MacOs hardware problem, but since Vice detects at least something as shown by the blue square, perhaps it can be "fixed" in Vice by changing some threshold setting or something like that? |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
please try latest build from here: https://github.com/VICE-Team/svn-mirror/releases and see if the sound problem is gone :)
If not, please fire up the "gamecontroller" test thing in windows, and tell what it reports (number of axes, buttons etc)
edit: oh it is macos... macos seems kindof broken. still do the above, please :) |
| |
Rastah Bar
Registered: Oct 2012
Posts: 336 |
I didn't notice a sound problem on MacOs.
How do I fire up the "gamecontroller" test thing (on MacOs)? |
| |
Peacemaker
Registered: Sep 2004
Posts: 243 |
cygwin -> ./x64sc -> strg+c does not end x64 anylonger since some releases. please help :> |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
Quote:How do I fire up the "gamecontroller" test thing (on MacOs)?
I dont know. Assuming this is a proper OS - there surely is some kind of settings dialog that lets you setup and test gamecontrollers? |
| |
Compyx
Registered: Jan 2005
Posts: 631 |
Quoting Peacemakercygwin -> ./x64sc -> strg+c does not end x64 anylonger since some releases. please help :>
I don't know what 'strg+c' is, but if you mean control+c, that works when running an emulator from an msys2 shell.
Edit: the shell needs to have focus for this to work of course, otherwise control+c ends up in the emulated machine as CBM+c. |
| |
Peacemaker
Registered: Sep 2004
Posts: 243 |
Quote: Quoting Peacemakercygwin -> ./x64sc -> strg+c does not end x64 anylonger since some releases. please help :>
I don't know what 'strg+c' is, but if you mean control+c, that works when running an emulator from an msys2 shell.
Edit: the shell needs to have focus for this to work of course, otherwise control+c ends up in the emulated machine as CBM+c.
Yeah , strg = cntrl of course. Sorry for that.
Does not work with cygwin anylonger thou. Using msys for windows it works. Lets see if i can get all working toolsto msys then. |
| |
MagerValp
Registered: Dec 2001
Posts: 1055 |
Quote: Quote:How do I fire up the "gamecontroller" test thing (on MacOs)?
I dont know. Assuming this is a proper OS - there surely is some kind of settings dialog that lets you setup and test gamecontrollers?
Controllers Lite: https://apps.apple.com/se/app/controllers-lite/id673660806?mt=12 |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
controllers on macOS should be fixed now (test if you can compile) |
| |
hedning
Registered: Mar 2009
Posts: 4596 |
Nice Canadian review of Vice 3.6 ("VICE 3.6 Released With Improvements (and Bugs)"):
https://www.youtube.com/watch?v=mPUKbtj1pGw |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
It'd be much nicer if it'd actually tell details about the many bugs that are fixed over the 3.5, instead of pointing out two mostly irrelevant details as new bugs :) |
| |
hedning
Registered: Mar 2009
Posts: 4596 |
Quote: It'd be much nicer if it'd actually tell details about the many bugs that are fixed over the 3.5, instead of pointing out two mostly irrelevant details as new bugs :)
He is mostly praising it. And while doing so points out two bugs. :) |
| |
Perplex
Registered: Feb 2009
Posts: 254 |
There seems to be an issue with the REU support in this version.
I have made a small test program which runs fine with 3.5 but fails with 3.6 and the latest snapshot. It mounts a 256KB image with the last page consisting of the bytes 0-255 in order, then tries to copy that page to memory and checks a few addresses for expected values.
https://github.com/lhz/reu-check |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
are you autostarting your .prg? then that is the problem :) try with the github build from https://github.com/VICE-Team/svn-mirror/releases |
| |
Perplex
Registered: Feb 2009
Posts: 254 |
Yes, I am autorunning the prg, which used to work fine. It also fails with the r41604 snapshot. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
prints "ram expansion unit content validated." in green here - how exactly do you run it? i basically just copied the cmdline from your makefile
x64sc -reu -reuimage reu-image.bin -reusize 256 reu-check.prg
edit: try adding -default to the beginning of that cmdline and see if that changes anything |
| |
Perplex
Registered: Feb 2009
Posts: 254 |
Interesting!
From the output of x64sc:
command line was: /home/lars/dev/misc/svn-mirror-r41604/vice/src/x64sc -reu -reuimage reu-image.bin -reusize 256 reu-check.prg
reconstructed commandline options (might be incomplete):
-default -monlog -sdlinitialw "1530" -sdlinitialh "1160" -reu -reusize "256" -reuimage "reu-image.bin" -remotemonitor |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
you need at least https://sourceforge.net/p/vice-emu/code/41609/ :) |
| |
Perplex
Registered: Feb 2009
Posts: 254 |
Works fine here too with r41609. Thanks! :) |
| |
Joe
Registered: Apr 2002
Posts: 224 |
I can swear that in one of the many versions of Vice, RunStop was on the positional "Caps Lock key".
Getting my head around using Esc key feels a bit awkward.
Part from that, still enjoy this one :D! |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
That was very long ago, and it should be obvious it causes nothing but problems. You can always remap it that way if you like broken keyboard :) |
| |
Stone
Registered: Oct 2006
Posts: 168 |
Works on real HW! Back in the day when my Restore key stopped functioning, I hardwired it to the Shift Lock key. Worked like a charm! |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
LOL. feel free to create a stones-broken-keyboard.vkm =) |
| |
Endurion
Registered: Mar 2007
Posts: 72 |
It seems the arguments for true drive have changed.
If passing the old version (-/+truedrive -/+virtualdev) VICE simply doesn't startup without any message, just returning 1.
Please, if you don't like the arguments, show some info message about that. |
| |
Compyx
Registered: Jan 2005
Posts: 631 |
The -virtualdev option has indeed changed to -virtual[8-11], as has -truedrive, which is now -drive[8-11]truedrive.
VICE does print an error message:
Option '+virtualdev' is ambiguous.
Error parsing command-line options, bailing out. For help use '-help'
So I assume your using Windows, where Microsoft in their infinite wisdom decided to not connect stdout/stderr to the terminal. You can still catch the messages by redirecting stdout/stderr to a file or piping to a program like more. |
| |
JackAsser
Registered: Jun 2002
Posts: 1989 |
Thanks for the 3.6.1 update! Btw does cpu tracing eat so much extra power that it can’t be compiled in by default? Not that it’s hard to compile yourself ofc, just asking out of curiosity. I could also just measure myself… :) Follow up question, I have not digged in that code for a long time but remember I patched in cycle/line tracing also for some project I did, but would it be complicated to have enabling/disabling cpu tracing as a runtime option? |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
"Btw does cpu tracing eat so much extra power that it can’t be compiled in by default?"
It takes almost zero extra effort actually, its just a cyclic buffer, 3 lines of code per CPU cycle. So yeah, the option to disable it can probably be safely removed alltogether :) |
| |
JackAsser
Registered: Jun 2002
Posts: 1989 |
Quote: "Btw does cpu tracing eat so much extra power that it can’t be compiled in by default?"
It takes almost zero extra effort actually, its just a cyclic buffer, 3 lines of code per CPU cycle. So yeah, the option to disable it can probably be safely removed alltogether :)
Thanks for the quick answer GP! Consider that a request then. :) |
| |
Erhan
Registered: Feb 2005
Posts: 17 |
win10pro 19043.1415 locale TR lang TR,
x64sc.exe, both 3.6.0 and 3.6.1
monitor activated, there, it is impossible to type # and $ from keyboard. Shift+3 Shift+4 AltGr+3 AltGr+4 respectively types ^ + <none> <none>. (typing # $ was also a problem in main emulator - default settings - but after editing gtk3_pos.vkm it's normal there) |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
try the build from github, alt-gr in the monitor was fixed. |
