Tales of the Black Cat — why the ZX Spectrum scene is not ahead of the rest of the planet

SibNews #0A

“Tales of the Black Cat”

Author: Black Cat

Why Aren’t We Ahead of the Rest of the Planet?

Chapter 1. Ahead of the Rest of the Planet

Imagine this hypothetical situation: Jörn Proda, the creator of SymbOS, comes to us and offers to port his system to our platform. What do you think would happen next? :) ... That’s right: chaos would break out, with everyone trying to settle the eternal question, “What is a Spectrum?” :) And what do you think it would end with? :) I reckon it’s predictable: Jörn Proda would take one look at the whole thing and say, “Well, you sort it out among yourselves; I’ll go ahead and port SymbOS to the ZX Next. :)”

It’s no secret that mass production of ZX Spectrum clones began earlier in the USSR than anywhere else, and as a result, we’re ahead of everyone in the development of Spectrum architecture. We were the first to create mass-produced, maximally simple single-field clones built from discrete components; the first to standardize memory expansion; the first to understand the fundamental Altwasserian principles underlying ZXBus and develop them into NemoBus; the first to create serially produced CPLD and FPGA clones; and the first in the world to create software-and-hardware Spectrum clones. But when people in the West talk about the Spectrum, our developments—if they’re mentioned at all—come last, as some kind of poorly understood exotic curiosity. They generally use Western designs, even though these are often significantly inferior to our own, and sometimes are frankly incompetent architecturally, as in the case of the ZX Next.

Of course, we could put this down to our not doing a good job of telling people in the West about our computers, but that wouldn’t be true. In the West, they know about many of our Spectrum clones, both common and less so. The problem comes when they need to choose a single architecture—that’s when their eyes start darting around at all the variety. After all, we don’t have a single architecture that could be called the ultimate expression of domestic Spectrum development. You might ask: how could it be that, despite being ahead in architectural development, we still haven’t managed to unite our different architectures? It turns out that developing hardware is a lot easier than reaching a consensus among people.

Chapter 2. NedoPC Against the Spectrum.

Article: “Tales of the Black Cat”

Fragment 2/10

In the late ’90s, Vyacheslav Skutin, aka Nemo, made attempts to organize a community and resolve the issue of standardizing Spectrum architectures. But the process was never completed, due to waning interest in the Spectrum and Nemo’s departure.

Then, literally a couple of years later, the NedoPC group, consisting of CHRV and Maks-Gor, announced that it would continue developing the Spectrum. Together with a few others who joined them, they started developing a KAY-1024 clone on a CPLD. But then, as luck would have it, they got the remains of the production run and all the documentation for the ATM-2 computer for free. Their love of a freebie abruptly changed NedoPC’s direction: development of the KAY-1024 clone was abandoned, and all efforts were redirected toward reviving ATM-2 production using the boards and components they’d got for free.

Whether due to CHRV’s lack of professionalism or his love of freebies, he turned a blind eye to the fact that, formally, the ATM-2 is not a Spectrum clone at all, but a completely unrelated CP/M computer with a Spectrum mode bolted onto it. So it could in no way be proclaimed by NedoPC as the “bright future of Spectrumism.” CHRV actively promoted himself as the Spectrum’s sole manufacturer and savior, and on the strength of this became moderator of the “Hardware” section on the newly created zx.pk.ru forum.

But in 2006, Black_Cat registered there as well. I hadn’t followed the development of the domestic Spectrum scene for a long time, and I was curious to study its current state. In the course of my research, I discovered irrefutable evidence that the ATM-2 is not a Spectrum clone. Naturally, I decided to share these facts on zx.pk.ru, which provoked an incredible backlash and campaign of harassment by NedoPC and its adepts.

Having studied the posts by NedoPC’s adepts, I concluded that its electorate—the group forming NedoPC’s main support—consisted overwhelmingly of technically illiterate sceners who had still been children in the ’90s and lacked the professional qualifications to form their own judgment on the subject under discussion.

It was clear that presenting any facts to an ignorant, frenzied mob egged on by NedoPC was pointless. So I made a successful attempt to create my own section on zx.pk.ru. Its main purpose was to educate technically illiterate sceners and thereby deprive NedoPC of its electoral base, which it was exploiting—and manipulating—thanks to their ignorance.

Thus my “Concepts” section was created—the only section in the history of zx.pk.ru named after its author. I became its moderator, not because CityAceE wanted it, but rather in spite of her wishes, by the will of the forum users who elected me and whose interests I promised to represent.

Article: “Tales of the Black Cat”

Fragment 3/10.

Naturally, NedoPC was not pleased, and its followers repeatedly raised the question of removing the column and banning its author. But under the forum rules, that was impossible. So in April 2007, CHRV, along with a couple of other moderators, attempted to take over zx.pk.ru in practice, demanding that CityAceE give him the right to ban people he disapproved of and delete information he didn’t like.

Under pressure from CHRV, CityAceE’s nerves gave out, and he announced that the forum was closing, giving users a week to download its contents. But when the seven days were up, it turned out CityAceE didn’t have the guts. Instead of annihilating the forum, he reformed the moderator team, annihilating all moderators who had not sworn personal loyalty to him—which, naturally, included me :). New moderators were recruited on zx.pk.ru, whose chief qualification was vassal-like loyalty to their suzerain and readiness to ban anyone he disapproved of. CHRV’s dream of a “new order” came true: zx.pk.ru was transformed from a more or less democratic “forum of friends,” where users’ opinions prevailed, into a totalitarian “forum of the administration’s friends,” where you got banned for expressing an opinion different from the administration’s.

Miraculously, CHRV himself went from being CityAceE’s ouster to his most servilely devoted supporter, keeping his moderator title—unlike those who had taken part in the palace coup alongside him. Now there’s a substance that really doesn’t sink :). Conan was given an unlimited banhammer license. Meanwhile, CHRV tried to cash in on his new title, demanding that “Concepts” be deleted and Black_Cat banned. But Conan ignored him and made it clear he should keep quiet and be glad he hadn’t been banned himself. CHRV realized that in the new order he was number six, and wouldn’t get any of the perks he’d launched the coup for. So he quietly quit the moderator team, leaving his partner to clean up the Augean stables that had piled up over the years CHRV had been strutting around with his moderator badge in the “Hardware” section.

Meanwhile, the Pentagon-1024 v.2.2, developed by KoE and released to the public, practically put an end to ATM-2 sales, and the later release of Speccy 2007 finally buried all of NedoPC’s efforts. The “bright future of the Spectrum scene”—nobody needed ATM-2 anymore. NedoPC could do nothing about the popularity of Pentagon-1024 v.2.2, so they tried to cut KoE off from the Spectrum community and bring him over to NedoPC. KoE left zx.pk.ru for dlcorp, but preferred to remain independent. I continued the work I’d started in “Concepts,” but, so as not to depend on the banhammer on zx.pk.ru, continued it on my own…

“Tales of the Black Cat”

Fragment 4/10

On the zx.clan.su forum, Mick starts developing ZXM-Phoenix, and I get actively involved in the project, changing its entire ideology along the way. ZXM-Phoenix becomes the first step toward an architecture that unites all the domestic branches of Spectrum development. The goal of developing ZXM-Phoenix is to combine the closely related architectures of the Scorpion-256 and KAY-1024 computers into a hybrid SKAY architecture. Along the way, the computer gets an improved expansion bus. ZXM-Phoenix becomes the first computer with a NemoBus v.1.1m bus.

ZXM-Phoenix sales break all records. In the first six months after its release, more boards were sold than NedoPC sold ATM-2s in total. Given that these are computers in the same class, the difference is striking. At NedoPC, they began to understand that if Spectrum architecture continued to develop according to the unifying ideology I had proposed, nobody would need the ATM at all.

So CHRV starts a campaign on zx.pk.ru under the slogan “Do whatever you want, don’t listen to anyone,” urging people not to cooperate with Black_Cat and not to follow the Spectrum standards for the bus that I support. Instead, he promises NedoPC’s backing to everyone who takes this path. CHRV needed chaos in Spectrum development, so that a unified architecture supported by the entire Spectrum community would never emerge. After all, only if the scene remains splintered into a heap of extended architectures, for which hardly anyone writes software, does the ATM—with its meagre software, written by two and a half people—have a chance to compete with the Spectrum, for which the whole community writes software. NedoPC’s goal is for the Spectrum community to write software for the ATM instead of the Spectrum.

I have nothing against the ATM as a standalone platform, but deliberately misleading technically ignorant sceners by claiming that the ATM is a development of the Spectrum is despicable, in my view… though not in CHRV’s. That is precisely why the base configuration for the NedoPC “ZX Evolution” dev board, developed in 2011, is called “PentEvo.” With this name, NedoPC tried to mislead users, presenting the configuration as an evolutionary development of the Pentagon, when in fact it was the same ATM-2 with a Pentagon-1024 KoE bolted on in place of the ZX Spectrum 128.

As soon as the ZXEvo dev board appeared, someone proposed making it possible to load a second configuration with a Spectrum architecture. But NedoPC didn’t like that prospect: they had created the PentEvo configuration as the only one, with the goal from the outset of herding all users into their ATM concentration camp. So ATM loyalists on zx.pk.ru immediately began obstructing the proposal. NedoPC adherents tried to take advantage of the novelty for users

Tales of the Black Cat

Fragment 5/10.

…elements such as FPGA, in order to mislead them by equating the devboard with the config, and calling the devboard everywhere by the name of the base config, “PentEvo.” So, to thwart their plans, I had to make an effort to explain to the sceners that the ZXEvo devboard is merely a container for the config, which is itself the computer architecture that gets loaded. In this way, I separated the concepts of config and devboard for users.

Some time later, I was finally banned from zx.pk.ru, and NedoPC immediately tried to take advantage of it. They figured that with their main opponent gone, they could at last see their plan through and foist an ATM in a new PentEvo wrapper on the Spectrum community. To this end, they sent AloneCoder as their emissary, outfitting him from Mordor’s stores with a horse, a saber, a burka, and an endless supply of ammunition, as well as a ZX Evolution rev.A board, and dispatched him to zx.pk.ru :).

Unfortunately for NedoPC, years of work educating the sceners technically—aimed at taking away NedoPC’s electoral base, which it had previously parasitized on account of their technical illiteracy, and manipulated thanks to that same illiteracy—had not been in vain for the community. Even in my absence, AloneCoder was told to get lost, along with his ATM :). After that, NedoPC barricaded itself in its own forum and made no further attempts to conquer the world :).

By the way, I created a test account on that forum and CHRV banned me immediately. At last, CHRV had found a place where he could make his dream come true: banning and deleting :). And you might think that’s the end of the story, but a little later the TSL forum branched off from the NedoPC forum. I created a test account there too, and was banned almost immediately, in a particularly perverted fashion :). Since I had never crossed paths with TSL anywhere before that, it’s safe to say that CHRV stayed true to himself, zombifying and poisoning everyone who comes into contact with NedoPC. So if you’re planning to visit this wonderful place, stock up on condoms in advance :).

Chapter 3. The Heavenly Chimera

So, having clarified the historical aspects of the difficulties involved in achieving consensus between people, let’s move on to the conceptual side of creating a single architecture that could be described as the quintessence of domestic Spectrum development. The following stages of unification are proposed:

  1. SKAY = Scorpion-256 + KAY-1024.

Column: “Tales of the Black Cat”

Fragment 6/10

The process was set in motion in ZXM-Phoenix, which combined the hardware-related architectures of the Scorpion-256 and KAY-1024 computers into the hybrid SKAY architecture. SKAY is not intended to support Scorpion-specific features such as turbo switching via rd#1FFD/#7FFD, reading the bits of #1FFD/#7FFD through AY, reading VG93 interrupts via Kempston, the Scorpion’s RS232 and printer ports, the wait states of the Yellow Scorpion, or the Green Scorpion’s 316 lines. To complete the unification of the architectures, it would be desirable to modify the Scorpion ProfROM by replacing the ScorpionIDE drivers with NemoIDE, ScorpionRTC with GluckRTC, and also modifying the drivers for the excluded functionality.

2. SKAYP = SKAY + Pentagon-128

The first steps in this direction were taken by KoE with his Pentagon-1024 v.1.x. KoE arrived at the unification concept on his own, but the conceptual implementation turned out somewhat raw. Mick also experimented in this direction in ZXM-Phoenix rev.06, but without much success.

The SKAYP architecture assumes SKAY hardware and Pentagon timings. Switching INT/ according to the 1991 and 1993 Pentagon standards is supported. Support for Pentagon-512 is planned, combining control via D4#1FFD and D6#7FFD, but without the ability to address memory above 128 KB with the out(#fd),a command. Computers with 4 MB of RAM may support the Pentagon-1024 v.2.2 memory mode, but without support for #EFF7.

3. SKAY2P = SKAYP + Profi+

Support for #DFFD control is planned, combining the equivalent functionality of the Profi and SKAYP memory managers. In CP/M mode, support for Profi peripheral ports and color is not planned. A monochrome 512×192 video mode is planned for CP/M, with scrolling by one third of the screen to display a 512×256 area. Support for the Profi raster generator is not planned in CP/M mode. The CP/M peripheral drivers are to be modified to support SKAYP peripherals.

4. Хiмеrа = SKAY/SKAYP/SKAY2P + АКМ

(hardware-switched multitasking)

An extended hardware architecture supporting up to 256 hardware-isolated virtual machines in 4 MB of RAM. Each machine can use from 16 KB to 2 MB of RAM and operate in hardware-switched multitasking mode.

Chapter 4. TSU vs ZX Next?

Tales of the Black Cat

Fragment 7/10.

In the domestic Spectrum scene, there have been repeated attempts to create additional graphics cards with tile-and-sprite accelerators. These range from experiments connecting a Dendy or a Yamaha v9990, to the TSU, which is an integral part of the TS configuration for ZXEvo. Something similar is also included in the ZX Next configuration.

Of course, tile-and-sprite accelerators like these cannot formally be considered part of the development of the ZX Spectrum architecture. But as optional external peripherals, or integrated peripherals, they can certainly serve as additions to the computer’s architecture. So there is no point in fighting such peripherals as if they had nothing to do with the Spectrum’s development: they are not, in and of themselves, the root cause that prompts programmers to move to other platforms. What makes programmers move is reaching their programming ceiling on a given platform, and wanting to transfer that ceiling, without any effort, to a platform where nobody has done it yet. So there’s no need to fear that all programmers will abandon the Spectrum for the Dendy just because it has a sprite accelerator. :)

Can we say that the TSU and the ZX Next tile-and-sprite accelerator are competitors? Architecturally, they’re not even worth comparing. Suffice it to say that the TSU was originally developed as the foundation for the entire TS configuration, and thus represents a compilation of the best architectural solutions that could possibly be crammed into the rather feeble ZXEvo FPGA. By contrast, the ZX Next tile-and-sprite accelerator was bolted onto an existing ZX Uno-based configuration as an afterthought—that is, “whatever worked, worked.”

As a result, unlike the clean architecture of the TS configuration, the ZX Next tile-and-sprite accelerator’s architecture is a patchwork, bristling with hacks and supports. For example, every element of the TSU can run at a maximum frequency of 28 MHz, while the processor runs at 14 MHz (albeit with wait states). But on the ZX Next, as soon as you display the second layer (Layer2), you have to switch from 14 MHz to 7 MHz. The same goes for all the other memory, including the second Spectrum screen: all of this memory in the ZX Next can run only at 7 MHz.

There’s no point even talking about the announced 28 MHz on the ZX Next; at best, a bare 48K Spectrum could run at that frequency. But why? It would be physically impossible to play old games at that speed. :) So, from an architectural point of view, the ZX Next can be summed up in three words: “horror—pointless—

“Tales of the Black Cat”

Fragment 8/10

… and merciless! :) All that remains is to feel sorry for those programming for this horror, which was released in a run of three thousand copies. There’s no point in discussing the ZX Next architecture any further; it’s hard to find an architecture in a heap of shit and sticks.

Chapter 5. Prospects for the TS config

The TS config is not a monolithic platform, as Alone Coder tried to portray it by contrasting it with the ATM. The TS config can be divided, roughly, into functional units: the Pentagon itself, Shadow DMA, CASHE, the interrupt controller, and the TSU. Remove the TSU, and what remains is a computer with cache, DMA, and an interrupt controller—a perfectly full-fledged ZX Spectrum clone by today’s standards. So claiming that the TS config isn’t a Spectrum is just as wrong as considering it a standalone platform. The TS config is neither one nor the other; it’s simply an integrated, expanded set of peripherals for a Pentagon clone, extending the platform’s capabilities. The fact that it’s all packed into a single FPGA doesn’t matter, since one configuration can consist of many devices. All of the TS config’s functional units except the Pentagon bring something architecturally new, because they had never been used on the Spectrum in this form before (with the exception of the interrupt controller).

Shadow DMA deserves special mention. Unlike a traditional DMA controller, Shadow DMA doesn’t stop the CPU and take control away from it; instead, it accesses memory only during cycles left free after CPU and TSU accesses.

Developing an amateur architecture like the TS config—which doesn’t copy a donor architecture for the sake of its software, but instead assumes that its own software will be developed—is a rather debatable undertaking, since it takes considerable effort to promote and popularize it, something the TS config clearly doesn’t excel at. The TS config still doesn’t even have a name, let alone any proper documentation. With this approach, despite the config being implemented quite professionally, the development is usually destined for nothing but languishing on the fringes of amateur exotica.

But thanks to VBI, who accidentally hit upon a way to change the situation with developing software for the TS config without even realizing it, things could change dramatically. What exactly did VBI discover? In terms of its capabilities, the TSU surpasses many platforms that were mass-produced and have a very large library of games written for them. Yet the hardware functionality of those platforms is quite standard, and the TSU can fully match it. While working on software localization, VBI arrived at a way, by trial and error, to speed up porting other people’s programs to the TS config. Usually, other people’s…

“Tales of the Black Cat”

Fragment 9/10

The program was taken apart, and then effectively rebuilt from scratch for the new platform. This is an extremely labor-intensive and unproductive process that takes a very long time and does not allow for any radical increase in the platform’s popularity. That is precisely why hardware projects of this kind have so far brought their developers no success.

But the combination of two coincidences—the optimal set of hardware features implemented in the TS configuration, and a new approach accidentally stumbled upon by VBI—created the effect of an evolutionary leap, a dialectical transition from quantity to quality.

The essence of VBI’s discovery is that the donor program does not need to be completely taken apart and then rebuilt from scratch. It is enough to add a software shim between it and the TSU, to reconcile the interaction interfaces of the donor program and the recipient computer’s tile-and-sprite accelerator.

Usually, this emulation method causes a significant slowdown when the program runs on the recipient computer, due to the software shim. But concentrating hardware capabilities such as turbo mode, Shadow DMA, CASHE, and an interrupt controller in the TS configuration makes it possible to gain enough performance to compensate for the overhead of the software shim, allowing foreign software to run without slowing down.

Meanwhile, the increase in porting speed comes from being able to work with large, unmodified blocks of software. These do not need to be taken apart and reworked: they work with the recipient computer’s hardware through the software shim that reconciles the interfaces. The “thickness” of this shim varies depending on how architecturally different the donor’s and recipient’s tile-and-sprite accelerators are. The more alike they are, the “thinner” the software shim, and the lower the cost of reconciling the interfaces.

You can see the result of this accelerated development in the Sonic port to the TS configuration by VBI.

Given that development of the TS configuration for ZXEvo has effectively been halted by the FPGA’s capacity, a specialized dev board for this configuration with a larger FPGA could make it possible to further improve the architecture—specifically, by simplifying the reconciliation of the donor and recipient interaction interfaces through additional architectural capabilities designed precisely for this kind of use.

Also, in that case, it would become possible to use the Хiмеra unifying architecture as the base platform for the TS configuration. This would make a significant contribution to consolidating the domestic Spectrum community and intensifying software development.

“Tales of the blAck Cat”

Fragment 10/10

...software for a unified architecture [s]and into poultry farming![/s] :)

P.S. Chapter 6. IT has appeared!

Shocking, earth-shattering news! Just as SibNews was getting ready to go to print, IT appeared for the TS-config that had lived for five years without a name! And don’t think I mean an evil clown (although that’s an interesting thought too :) )—no, I mean IT, i.e. the name :).

Now, instead of the vague and perplexing “TS-config,” we can finally give the config something more like a proper name: “ZX Enhanced”! Well, fancy that! What flair, what a flight of imagination! “ZX Enhanced”! :) I immediately shortened it to “ZXE” for myself, and I recommend you do the same. I was especially pleased, because try typing “TS-config” 100500 times in an article, and you’ll quickly feel a strong desire for revolutionary change in the direction of fewer letters :).

Who should we thank for this surprise? Ivan Pirog, VBI, TSL, and Robus have been thinking about a name for these five years! :)

But that’s not all the shocking news! The synchronicity is astonishing: I’d only just written in an article that hadn’t even been published yet that the TS-config had been nameless for five years, when VBI immediately announced the name on Telegram! :) This isn’t the first time I’ve shown extraordinary abilities, though. :) And you can find out more about it in the next issue of SibNews—that is, the August issue—in the “Tales of the blAck Cat” section!

See you in the next issue! Unless Nibiru arrives first. :)

Share your thoughts about the article