Page 10 of 12

Re: ZXBaremulator 3.2

Posted: Sat Aug 01, 2020 6:27 pm
by TMD2003
Well, it's working, even if I'm limited to using composite video (as I don't have any external speakers that aren't being used for something else, which I'd need if using HDMI). Marvellous! I will test it further, soon enough.

I might also need to not use a Mac keyboard, seeing as it doesn't have a Break key on it...

Re: ZXBaremulator 3.2

Posted: Sat Aug 01, 2020 7:22 pm
by zx81
akeley wrote:
Sat Aug 01, 2020 3:44 pm
zx81 wrote:
Sat Aug 01, 2020 3:27 pm
The ZX-Uno's Spectrum 48k core is basically correct, but the contended memory timings is wrong on all cores, all FPGA's I know. The 128k core isn't complete and the FPGA PLL can't generate the needed CPU freq. for the model (is off by 8 kHz). The +2a is the same core than 128k with the +2a ROMS.
How do contended memory timings affect accuracy? Could you give examples of software which does not work becasue of that?

It's the same for the 128K core, if they are meant in regard to MiSTer. I'm interested in perfecting these cores, so can ask questions about it but need some concrete examples of reproducible errors.
zx81 wrote:
Sat Aug 01, 2020 3:27 pm
Anyway, the image quality is very poor, IMHO.
Both ZX Uno and MiSTer's IQ is 1:1 perfect when compared to my real ZX Spectrums - tried it side by side on CRT TVs. Could you be more specific?
The IR_Contention test don't runs correctly on any FPGA (excluded the Next, I know nothing about it). Many 128k cores runs at a lower speed freq (the ZX-Uno runs at 3.5 Mhz, not 3.5469 Mhz). Many demos (Paralactika, Mescaline, Tiratok, etc) runs bad or fails badly. I own a ZX-Uno, and I know the status on others FPGAs by users, not by my own tests. The Spectrum 48k implementation is a bit better, but far from perfect, IMHO.

Re: ZXBaremulator 3.2

Posted: Sun Aug 02, 2020 1:07 am
by cmal
zx81 wrote:
Sat Aug 01, 2020 9:11 am
Pegaz wrote:
Fri Jul 31, 2020 1:33 pm

I would just like to get rid of this buzzing sound, which we talked about earlier.
I don't know if you had time to investigate it, I just have this problem on each of my 3 pi models (2, 3, 3+) and I haven't been able to eliminate it, except using the old v2.0 version, which doesn't have a GPIO keyboard support.
However, BM has its advantages, such as precise emulation and the ability to easily connect the Spectrum keyboard and its also free.
I don't know which is the next equivalent variant, I guess ZX-Uno with twice the price than Pi, but also with greater potential for sure.
The new Next kickstarter is already around the corner, so everyone has something to choose for themselves...
The problem with the buzzing sound is that I can't reproduce it on my TVs. Could be that some TVs aren't happy with the 32 kHz sampling rate I use.

The ZX-Uno is dead. I own one, sleeping in his box. Nobody develops cores for it. Now, you can choose a MiSTer, Mistica or SiDi. But don't exist a FPGA core that his behaviour will be so precise as many software emulators. And nobody have enough motivation to enhance the cores. The most important core for MiSTer is NeoGeo.

The Next isn't for me, for less money I can have a MisTer with 128MB, I can live without his pretty keyboard.

Pegaz, not sure if I'm hearing the same buzzing sound as you, but I have a feeling it's something different. It's more of a few clicks that I sometimes hear after loading a tape. In some cases the virtual tape doesn't stop after the game has loaded and I have to press F8 to stop the tape. This doesn't bother me because you would have to stop a real tape connected to a real Spectrum anyway.
This was more prevalent on my Pi 2 than it is on the Pi 3 B+ that I'm using now. In fact now it so rarely happens that I don't notice it.

I agree with zx81 on halting development although I am admittedly disappointed. At the same time I do feel ZXBaremulator is "complete" in it's current state and offers a great and true to life emulation experience. I plan to continue using it as my main emulator.

Re: ZXBaremulator 3.2

Posted: Sun Aug 02, 2020 1:18 am
by cmal
TMD2003 wrote:
Sat Aug 01, 2020 6:27 pm
Well, it's working, even if I'm limited to using composite video (as I don't have any external speakers that aren't being used for something else, which I'd need if using HDMI). Marvellous! I will test it further, soon enough.

I might also need to not use a Mac keyboard, seeing as it doesn't have a Break key on it...
I got lucky on a pair of laptop speakers that plug straight into the Pi's 3.5 in socket, with the USB plugging into the Pi for power. They sound great on ZXBaremulator and are solidly constructed and can easily clip onto a small display. It also has volume control. I paid $15 on Amazon.
Here's the link if you're curious:

Re: ZXBaremulator 3.2

Posted: Sun Aug 02, 2020 9:36 am
by Pegaz
@cmal

Unfortunately, it’s a little worse than that.
In my case, it's an unpleasant buzzing sound of about five seconds each time I enter the F1 menu and go back to the emulation.
I have described this in more detail on page 3, in this thread.

Re: ZXBaremulator 3.2

Posted: Sun Aug 02, 2020 12:05 pm
by Pegaz
zx81 wrote:
Sat Aug 01, 2020 7:22 pm
akeley wrote:
Sat Aug 01, 2020 3:44 pm
zx81 wrote:
Sat Aug 01, 2020 3:27 pm
The ZX-Uno's Spectrum 48k core is basically correct, but the contended memory timings is wrong on all cores, all FPGA's I know. The 128k core isn't complete and the FPGA PLL can't generate the needed CPU freq. for the model (is off by 8 kHz). The +2a is the same core than 128k with the +2a ROMS.
How do contended memory timings affect accuracy? Could you give examples of software which does not work becasue of that?

It's the same for the 128K core, if they are meant in regard to MiSTer. I'm interested in perfecting these cores, so can ask questions about it but need some concrete examples of reproducible errors.
zx81 wrote:
Sat Aug 01, 2020 3:27 pm
Anyway, the image quality is very poor, IMHO.
Both ZX Uno and MiSTer's IQ is 1:1 perfect when compared to my real ZX Spectrums - tried it side by side on CRT TVs. Could you be more specific?
The IR_Contention test don't runs correctly on any FPGA (excluded the Next, I know nothing about it). Many 128k cores runs at a lower speed freq (the ZX-Uno runs at 3.5 Mhz, not 3.5469 Mhz). Many demos (Paralactika, Mescaline, Tiratok, etc) runs bad or fails badly. I own a ZX-Uno, and I know the status on others FPGAs by users, not by my own tests. The Spectrum 48k implementation is a bit better, but far from perfect, IMHO.
Can't find Paralactika demo in tap/tzx format, to test it on 128k models, I only have a trd version.
I saw in soime youtube video "paralactika_zx128" or something like that but the author did not leave a link.
Can you share some download link for that version, which you used when testing on 128k machines?

Re: ZXBaremulator 3.2

Posted: Sun Aug 02, 2020 12:07 pm
by zx81
cmal wrote:
Sun Aug 02, 2020 1:07 am
zx81 wrote:
Sat Aug 01, 2020 9:11 am
Pegaz wrote:
Fri Jul 31, 2020 1:33 pm

I would just like to get rid of this buzzing sound, which we talked about earlier.
I don't know if you had time to investigate it, I just have this problem on each of my 3 pi models (2, 3, 3+) and I haven't been able to eliminate it, except using the old v2.0 version, which doesn't have a GPIO keyboard support.
However, BM has its advantages, such as precise emulation and the ability to easily connect the Spectrum keyboard and its also free.
I don't know which is the next equivalent variant, I guess ZX-Uno with twice the price than Pi, but also with greater potential for sure.
The new Next kickstarter is already around the corner, so everyone has something to choose for themselves...
The problem with the buzzing sound is that I can't reproduce it on my TVs. Could be that some TVs aren't happy with the 32 kHz sampling rate I use.

The ZX-Uno is dead. I own one, sleeping in his box. Nobody develops cores for it. Now, you can choose a MiSTer, Mistica or SiDi. But don't exist a FPGA core that his behaviour will be so precise as many software emulators. And nobody have enough motivation to enhance the cores. The most important core for MiSTer is NeoGeo.

The Next isn't for me, for less money I can have a MisTer with 128MB, I can live without his pretty keyboard.

Pegaz, not sure if I'm hearing the same buzzing sound as you, but I have a feeling it's something different. It's more of a few clicks that I sometimes hear after loading a tape. In some cases the virtual tape doesn't stop after the game has loaded and I have to press F8 to stop the tape. This doesn't bother me because you would have to stop a real tape connected to a real Spectrum anyway.
This was more prevalent on my Pi 2 than it is on the Pi 3 B+ that I'm using now. In fact now it so rarely happens that I don't notice it.

I agree with zx81 on halting development although I am admittedly disappointed. At the same time I do feel ZXBaremulator is "complete" in it's current state and offers a great and true to life emulation experience. I plan to continue using it as my main emulator.
The HDMI sound works differently than PWM sound. In PWM, the "metronome" is the sound (ZXB executes a frame, send the PWM sound using DMA and waits for the end of transfer). For HDMI, the metronome is a system interrupt, and any slight delay sending the samples produces a sound fault. I try to have the minimal samples on the fly, to reduce the sound lag. The samples are sended to VC4 using a mailbox, and is dependent of a very frequent polling of the VC4 answers, in periods lesser than 0.1 ms. As I'd said before, the CPU-0 executes many tasks using a cooperative scheduler. Sometimes, the scheduler comes too late, I fear.

I can't say how impact the firmware version to my program. The PI is a very strange and awful ARM architecture.

The TAP files don't have a "STOP TAPE" block as TZX have. And many TZX don't have a "STOP TAPE" block between different loads. I've modified many TZX to add such block.

In my initial wishes, I should add a tape browser, support for SAVE virtual tapes, load and save snapshots, +3 FDC & floppy emulation, configurable options to some features....

ZXB is far from complete, IMHO. But, have any sense to add it, when the users are starting to migrate to FPGA based systems?

Re: ZXBaremulator 3.2

Posted: Sun Aug 02, 2020 12:10 pm
by zx81
akeley wrote:
Sat Aug 01, 2020 3:44 pm
Can't find Paralactika demo in tap/tzx format, to test it on 128k models, I only have a trd version.
I saw in soime youtube video "paralactika_zx128" or something like that but the author did not leave a link.
Can you share some download link for that version, which you used when testing on 128k machines?
Yes, I have, on my home. But I'm not at home today. Tomorrow afternoon, I'll put a link to it.

Re: ZXBaremulator 3.2

Posted: Mon Aug 03, 2020 10:50 am
by Pegaz
zx81 wrote:
Sun Aug 02, 2020 12:07 pm
ZXB is far from complete, IMHO. But, have any sense to add it, when the users are starting to migrate to FPGA based systems?
Can’t speak on behalf of others, but I certainly won’t migrate and stop using BM.
I have been following your emulator from the beginning and I will always be a big fan of it, regardless of all the new fancy hardware achievements.

Re: ZXBaremulator 3.2

Posted: Mon Aug 03, 2020 1:51 pm
by Mort
i have a next and still like the home made zxbaremetal machines almost as much. All i want is the option to save state my games etc and all would be fine by me :D