Musical potential of the NEXT.

This section is for discussing everything about Next hardware and latest updates.
User avatar
sol_hsa
Posts: 82
Joined: Fri Jun 02, 2017 10:10 am

Re: Musical potential of the NEXT.

Postby sol_hsa » Tue Jun 19, 2018 5:27 pm

SevenFFF wrote:
Tue Jun 19, 2018 2:25 pm
If you have any technical suggestions about what you’d like to see the Pi do or what the interface should be like, please give them. I’ll make sure they get passed on to Victor.
Primarily I'm worried that since it hasn't been stress-tested too much, there might be some hardware issues, but since it's on top of everything else that I actually care about, it will be interesting to see what happens..

I can see the pi used in two different ways; either as a pile of coprocessors that can be called from the z80 side (audio, 3d graphics, whatever), or completely taking over the machine being (basically) a small linux computer with a weird keyboard.

What I wouldn't want to see is programs becoming z80 stubs uploading arm code to the pi and then running it there. That feels like an easy way to accelerate "helper functions" (just see what they did with ARM palm machines) but will be abused and most code will run on arm (just see what actually happened with ARM palm machines)..

User avatar
fgeva
Posts: 31
Joined: Wed May 31, 2017 3:20 pm

Re: Musical potential of the NEXT.

Postby fgeva » Wed Jun 20, 2018 8:47 am

SevenFFF wrote:
Tue Jun 19, 2018 2:25 pm
If you have any technical suggestions about what you’d like to see the Pi do or what the interface should be like, please give them. I’ll make sure they get passed on to Victor.
Something I think could be useful would be to expose enough NEXT CPU state to the Pi to allow for efficient full debugger support (single-stepping, memory watches, ...) where the development tools would run on the Pi. I don't have any technical suggestions on what that would involve unfortunately, so I don't know if it's even possible. I do know it would be a lot of work.

In an ideal world where code just appears by magic and no hardware limitations exist, I'd imagine a system where both the NEXT and the Pi can control keyboard and screen, so you could seamlessly switch between development environment on the Pi and the running code on the NEXT, possibly even with the NEXT display appearing in a window when the Pi is "active". In the real world, something that slightly approaches that would be nice.
Backer 754

User avatar
sol_hsa
Posts: 82
Joined: Fri Jun 02, 2017 10:10 am

Re: Musical potential of the NEXT.

Postby sol_hsa » Wed Jun 20, 2018 3:33 pm

Pi as ice debugger does sound like fun, but probably quite unlikely at this point.

User avatar
SevenFFF
Posts: 189
Joined: Mon Jun 05, 2017 5:30 pm
Location: USA

Re: Musical potential of the NEXT.

Postby SevenFFF » Wed Jun 20, 2018 4:01 pm

Garry has started work on the multiface debugger now, btw, which is great news!
Robin Verhagen-Guest
SevenFFF / Threetwosevensixseven / colonel32
Spectron 2084blog

PiyoTaro
Posts: 130
Joined: Thu Jun 01, 2017 11:13 am

Re: Musical potential of the NEXT.

Postby PiyoTaro » Thu Jun 21, 2018 5:21 pm

Debugger application? Is it that LINUX application of "Raspberry Pi Zero" can access the memory bus of Z80 or can it access the SRAM address of the FPGA?
The application I thought is emulation of "Z80DMA/Z8410" which canceled 28 MHz mode. Since "tile graphics screen" was not implemented in SpectrumNext, it is supposed to use Z80DMA for rewriting the screen of the game application. By the way, because "Z80DMA" did not have a product with a high clock model, I think that it did not have to be concerned with reproducing real hardware.

Let's put the discussion back to "sound", from the "Raspberry Pi Zero" on the coprocessor board, is it possible to output sound from the HDMI terminal or EAR output terminal on the SpectrumNext main unit through that GPIO port?
It seems that additional implementation of the FPGA is necessary.

PiyoTaro
Posts: 130
Joined: Thu Jun 01, 2017 11:13 am

Re: Musical potential of the NEXT.

Postby PiyoTaro » Thu Jun 21, 2018 5:25 pm

Another question is, in order to reproduce "YM2203" which is the sound chip of "Turbosound FM" on the FPGA, does it need more than 50000 logic elements?


2018/06/28, Additional notes.
In the "ZX-UNO" forum, there was a thread of FM sound "YM3812 (OPL2)" in February 2017 in the topics of the PCXT core.
The remaining capacity of the FPGA is "Son 9720 F / F libres, el 85%". 'jotego' said that 9000 instances can be safely reproduced ("Con + 9000 FF libres cabrá seguro. Es una pena que de haya más BRAM libre para implementar las dos ROM que lleva el chip en ellas.") .
SpectrumNext was adopted FPGA SLX16, PCXT can be reproduced on the same design board with smaller FPGA SLX9 + 2MB extended SRAM size hardware.
Last edited by PiyoTaro on Thu Jun 28, 2018 7:46 am, edited 1 time in total.

User avatar
Timbucus
Posts: 211
Joined: Tue May 30, 2017 7:43 pm

Re: Musical potential of the NEXT.

Postby Timbucus » Thu Jun 21, 2018 7:11 pm

There is simply not enough documentation yet on the final PI config but, I know currently there is no way that the PI display output or sound output (sound is only via the PI HDMI without extra hardware on the PI0 /PI0W) can be merged with the output from the FPGA as it sits on the main bus - not within the FPGA. It can watch the BUS looking for things to do with the current configuration. At the moment it is useful as a second display to visualize the ULA screen - separately if the display layers are stacked correctly as a form of Twin Screen Spectrum. The PI has always been billed as a co-processor / accelerator which will be its purpose if and when people develop some libraries to use on it - perhaps FP math for 3D? This data would still have to come back to the Z80 for rendering if part of the main output. That return path for data is not yet documented or finalised as I understand it.

As regards the FM sound on the Next I think that is an error on the original Kickstarter listing - It has two AY chips that make it compatible with the Turbosound (twin AY not FM) for the Spectrum and then another AY that allows even more to be done with a total of 3. Other Audio has be via the Covox (SpecDRUM) compatible port using either the CPU or the Copper to feed sample data.

The DMA chip is already implemented and is mostly compatible with other implementations of the same chip with some tweaks.
I'm Infinite Imaginations when not in work... PAWS for thought.

PiyoTaro
Posts: 130
Joined: Thu Jun 01, 2017 11:13 am

Re: Musical potential of the NEXT.

Postby PiyoTaro » Thu Jun 28, 2018 8:11 am

The topic of "Debugger" SevenFFF says was in other threads.
link (Thread of "Organisation to get other FPGA core working..." on "Other Cores")
There was a post in December not to understand how FPGA and Rasberry Pi are connected because board specifications are not disclosed.

Yesterday there was an announcement to release the specification of "version 1.0", but I hope that not only the program source of the FPGA but also the specification of such a board will be released.
PiyoTaro wrote:
Thu Jun 21, 2018 5:21 pm
Debugger application? Is it that LINUX application of "Raspberry Pi Zero" can access the memory bus of Z80 or can it access the SRAM address of the FPGA?
The application I thought is emulation of "Z80DMA/Z8410" which /---/
Let's put the discussion back to "sound", from the "Raspberry Pi Zero" on the coprocessor board, is it possible to output sound from the HDMI terminal or EAR output terminal on the SpectrumNext main unit through that GPIO port?
It seems that additional implementation of the FPGA is necessary.
SevenFFF wrote:
Wed Jun 20, 2018 4:01 pm
Garry has started work on the multiface debugger now, btw, which is great news!

PiyoTaro
Posts: 130
Joined: Thu Jun 01, 2017 11:13 am

Re: Musical potential of the NEXT.

Postby PiyoTaro » Sat Jul 21, 2018 8:21 am

PiyoTaro wrote:
Sun Jul 08, 2018 4:59 pm
PiyoTaro wrote:
Sun Jun 10, 2018 5:42 pm
Previously, I posted that I found a post by a person developing YM2203 or YM2151 in the forum of ZX-UNO.
Additional
There was his interview article on the game information site.:
RetroManiac May 5, 2017
"Creo que acabaré diseñando una placa FPGA orientada a sistemas arcade". Hablamos con Jose Tejada, autor de cores de chips Yamaha para FPGA http://www.retromaniac.es/2017/05/creo- ... placa.html
Last week I found a post that he helped develop "Turbosound FM" using YM2203 for the ZXSpectrum core for FPGA computer "MiST".
Also, in his post there is a description "While still working on a new generic core FM for all Yamaha fit the ZX-UNO", and it was also difficult to implement YM2203 in "ZX-UNO"(FPGA SLX 9) It was.
I found a recent post by 'jotego'.
He seems to have made a conclusion about "FM sound core that fits the ZX-UNO" with a thread of "Ghosts'n Goblins" of ZX-UNO Forum.
His JT51 (YM2151) FPGA core costs 1200 slices.
So he changed his mindset and said that by implementing PicoBlaze 8bit CPU (26 slices) in the FPGA, it was possible to implement two YM2203 at the cost of "312 slices and at least six blocks of RAM".

I also said that "Turbosound FM" is to emulate with the coprocessor board "Raspberry Pi Zero", but I think that implementation on FPGA is good as "standard sound device" with an official compatible machine.

*Additional notes
I found explanations about his "NPFM(New Processor FM)" sound core. This thread was a topic on "PicoBlaze".
"Re: KCPSM6 PicoBlaze: ¡Soft cores al poder!" http://www.zxuno.com/forum/viewtopic.ph ... =30#p23671

/* original
Mensajepor jotego » 07 May 2018, 08:02
He estado pensando este fin de semana sobre cómo meter un sintetizador FM en la FPGA del ZX-UNO ocupando muy poco sitio. Es un tema muy técnico pero muy interesante también para entender por qué en los 80 y 90 era habitual que hubiera un chip de sonido aparte y todos compráramos tarjetas de sonido para el ordenador. Creo que escribiré un artículo en retrónica explicándolo. Pero mientras os dejo unos números:

El JT51 (YM2151) ocupa alrededor de 1200 slices de una Spartan 6. La Spartan 6 del ZX-UNO tiene 1430 slices así que viene tan justo que a mí el sintentizador me falla al intentar meterlo. El G&G usa dos YM2203, que combinados es algo similar a un YM2151 (menos canales FM pero tiene canales PSG). Así que en principio no cabe haciéndolo de forma puramente hardware con segmentación a 1MHz.

Pero hay un procesador minúsculo hecho por Xilinx: el PicoBlaze. Esta es una CPU minimalista de 8 bits que ocupa sólo 26 slices en una Spartan 6. Esto quiere decir que podemos meter muchos PicoBlazes en una FPGA, pero muchos. Además puedes llegar a correrlos a 125MHz según la documentación. ¿Da una CPU de 8 bits a 125MHz para hacer síntesis FM? Pues no, para hacer 3 canales no. Para hacer 1 tampoco lo tengo muy claro... Pero estoy viendo si es posible dividir el problema entre varias CPUs. Si con 2 CPUs pudiera hacer un canal, un YM2203 necesitaría unas 6 CPUs (o sea unas 156 slices más 3 bloques de RAM -hay 32 en total-). Y dos YM2203 -o sea el G&G- el doble (312 slices y al menos 6 bloques de RAM). ¿Cabría esto en el ZX-UNO y además el Z80 y además el 6809 y además el vídeo del G&G y además el mínimo de interfaz SD? Pues no lo sé aun. Pero al menos bajamos ya un factor 4 en cuanto a espacio del sintentizador.

PiyoTaro
Posts: 130
Joined: Thu Jun 01, 2017 11:13 am

Re: Musical potential of the NEXT.

Postby PiyoTaro » Sat Jul 21, 2018 8:31 am

When I posted, his Github had FMSound firmware "JTx" present.
As I follow the remarks of that forum, it seems that the implementation test of the CPU on the "ZX-UNO" board is not done.
I will ask for a person who implements "PicoBlaze 8bit CPU" in the ZXSpectrum core and completes "TurbosoundFM".
Firmware is published in his Github (link is listed at the end of the previous "Interview article").
PiyoTaro wrote:
Sun Jul 08, 2018 4:59 pm
PiyoTaro wrote:
Sun Jun 10, 2018 5:42 pm
Previously, I posted that I found a post by a person developing YM2203 or YM2151 in the forum of ZX-UNO.
Additional There was his interview article on the game information site.:
RetroManiac May 5, 2017
"Creo que acabaré diseñando una placa FPGA orientada a sistemas arcade". Hablamos con Jose Tejada, autor de cores de chips Yamaha para FPGA http://www.retromaniac.es/2017/05/creo- ... placa.html
His JT51 (YM2151) FPGA core costs 1200 slices.
So he changed his mindset and said that by implementing PicoBlaze 8bit CPU (26 slices) in the FPGA, it was possible to implement two YM2203 at the cost of "312 slices and at least six blocks of RAM".
This thread was a topic on "PicoBlaze". "Re: KCPSM6 PicoBlaze: ¡Soft cores al poder!" http://www.zxuno.com/forum/viewtopic.ph ... =30#p23671


Who is online

Users browsing this forum: No registered users and 4 guests