ZX Spectrum Next Discussion

[DouglasReynholm]

The other issue you're mentioning has to do with the expansion bus. Xilinx discontinued 5v tolerant inputs on its fpgas starting with the spartan 6, which is the fpga used in the next. The fpga's interface to the expansion bus is really designed for fpgas with 5v tolerant inputs and really there should be level conversion done on the expansion bus interface instead. Xilinx was switched to from Altera because a good deal was had on their fpgas. I can't say if the Altera part was 5v tolerant but I am guessing it was and no change was made when the Xilinx part was substituted.

Altera and Xilinx haven't really had 5v tolerant io for years, the size of the io cell required being a limiting factor. I last worked for Altera in 2007, and it was a problem for a lot of customers then due to other legacy technology. Just off memory Altera Cyclone parts (the rough equivalent to Xilinx Spartan range) i don't think ever supported higher than 3v3. It's likely Xilinx just bought the business with a discount on the devices and maybe helped with the conversion work.

[Bizzley]

Who is the manufacturer of these keyboards, anyway?
I guess they have experience in keyboard production and these issues should be solved during testing process, instead of being delayed that much.

If a company with the experience that has been suggested can't get their prototypes or their machine-manufactured units to work reliably then that points to either a fundamental flaw inherent in the design or they don't know what they're doing. Obviously they do know because that's their business which just leaves the design. It's already been changed for aesthetic reasons according to the last Update (We experimented with the three layers and found the keys too ‘heavy’ to press) but with no indication whether those "heavy" keys did in fact do the job reliably or not.

If all the finger-crossing doesn't work and the latest experiments fail then I hope the back-up plan is that the Next team swallow their pride and go with a previous keyboard that works regardless of how it looks or how 'heavy' a key press is needed. If there has never been a working reliable keyboard at all then we truly are stucked!

[beanz]

I read somewhere else that the designer of the hardware ignored the concerns of people who pointed out a few flaws and recommended addressing them. Does anyone know if they were addressed? I don't know anything about hardware, so I don't understand the details. But it was something about the danger of having the FPGA directly connected to something, and bad implementation of HDMI.

Those problems were addressed when they were brought up I believe, addressed in the sense that they said the problem wasn't really there. I've been using the board now for 1 1/2yrs and it's been rock solid.

EDIT: Just read AAs post..note I've always run my board on VGA.

[beanz]

If all the finger-crossing doesn't work and the latest experiments fail then I hope the back-up plan is that the Next team swallow their pride and go with a previous keyboard that works regardless of how it looks or how 'heavy' a key press is needed. If there has never been a working reliable keyboard at all then we truly are stucked!

The previous keyboard might not work...the problem with the composite keys could have been on that one too...they just didn't get to that point in the testing phase as they'd already rejected it for being too heavy...which of course is bad testing practice.

[1024MAK]

Regarding TTL logic levels...

Well, at a VCC MIN of 4.75V, 74xxx, 74Sxxx and 74LSxxx types have VOH listed as a 3.4V typical.

Of course, the actual voltage is also determined by what other circuitry is connected to the same outputs. Pull-up resistors or other 74 logic gate inputs can increase the voltage towards VCC.

Mark

[1024MAK]

The problem has been and always will be that a purely mechanical membrane keyboard system is never good at making two independent switch contacts at the same time. Even during the days of Sinclair. Far too often on my ZX Spectrum+ keyboards, I get the wrong action unless I press the key squarely.

And although I don’t like the QL keyboard, it is slightly better than the keyboard on the ZX Spectrum+. The QL keyboard only being a normal two layer membrane meaning it is slightly easier to use.

The better solution would have been to do what Amstrad did with the +2A/+2B/+3/+3B and have a normal two layer membrane and perform the additional switching in a gate array (CPLD or FPGA) chip. But it’s too late for that now

Mark

[zxbruno]

The better solution would have been to do what Amstrad did with the +2A/+2B/+3/+3B and have a normal two layer membrane and perform the additional switching in a gate array (CPLD or FPGA) chip

So that's why they feel "right" and work so well. I love Sinclair computers, but always found the Amstrad keyboards better. Never had to replace a keyboard membrane on any of my +2s of +3s. They were always good for coding and gaming and I had, at some point, about a dozen of them. Never had a problem with any of the keys.

[Alcoholics Anonymous]

Does this mean that Q2 2019 remains as the deadline for delivery?

They are still trying to get Q2. It's impossible to give firm deadlines as the project is at the mercy of time allocated by an external supplier. The project is just too small and must live off scraps given it. This keyboard thing should be solvable quickly but will it? The last membrane took 2-3 weeks to make and test not because it's hard but because the supplier was busy. Hopefully priorities have changed a bit now that it is known machine assembly will not solve the problem.

The longer the delay the less likely sms will be able to schedule time to assemble everything before the end of June. On the plus side, a lot of things are already done. The pcbs, eg, were made long ago. But as helpful as sms is, 6 weeks out is starting to look short IMO.

I don't deal with the companies so I can't say for sure what the scheduling deadlines and concerns are as of now.

The problem has been and always will be that a purely mechanical membrane keyboard system is never good at making two independent switch contacts at the same time. Even during the days of Sinclair. Far too often on my ZX Spectrum+ keyboards, I get the wrong action unless I press the key squarely.

It's not important to get simultaneous contact it just has to be close enough together in time. The fpga can delay keys to see if a shift arrives within a certain amount of time. As long as this difference in time is not too large, key responsiveness will still be good. But the kind of delays seen with the semi-circle contact was at the crazy end too often to be acceptable (10-20ms).

Well, at a VCC MIN of 4.75V, 74xxx, 74Sxxx and 74LSxxx types have VOH listed as a 3.4V typical.

The fpga can tolerate a steady 4V. Resistive pullups will drive it higher, though with current that will not exceed maximum (and I think this means long term degradation of the input rather than short term failure). There are other retro projects getting away without level conversion using fpgas from the same family. Close to home there is the uno which has a 5V ps2 keyboard interface. As far as I know, no one has reported their ps2 keyboard failing.

It's still better not to depend on what-ifs and maybes though.

[toot_toot]

It's not just the Right Cursor key that doesn't work reliably, it's quite a few composite key presses that are prone to failure as you can see from the picture and mentioned in the latest Update. They are going to try three different designs for the new contacts and hope that one of them does the job but they won't know if any of them work until they get the results back. That to me sounds more like desperation than having a definite and workable idea of how to fix the problem.

That is exactly what the problem is. It should be made clear that the keyboard is not designed by the next team - it is by a keyboard manufacturer which has plenty of experience building keyboards. They were sure the keyboard would perform well once it was machine assembled rather than hand assembled and they were wrong. Two semi-circle halves as contact means preference for one side or the other in striking leads to one contact being made before the other. If the difference in time is significant this can't be covered up in hardware without affecting the responsiveness of the unshifted key. And that's where it is.

The length of delay is not because there's anything being hidden; it's because the next is a low volume project that has bottom priority for suppliers. They deal with it after higher priority customers are dealt with. Nothing may happen for weeks. Now that they see the machine assembled keyboard does not work as expected, hopefully they have reason to spend some more urgent attention.

The keyboard was designed by the Next team, specifically Rick Dickinson (and with input from his partner). They’ve taken their design to a keyboard manufacturer and asked them to produce it, although the keyboard manufacturer may have had some input on the design.

I think that the whole membrane issue, and backwards compatibility of the board with the original Spectrum has clouded what, I believe, is the real issue behind this. They’re not trying to just make a Spectrum keyboard (and membrane), they’re trying to make a modern feel keyboard with the technical limitations of connecting it to a near 40 year old computer. That’s resulted in this unique three membrane approach and has, so far, resulted in every membrane design failing. We have to presume the keyboard manufacturer know what they’re doing, after all keyboard design and production isn’t new, it’s a tried and tested method.

But again, the crux of the problem is the dogged approach of having a modern keyboard “feel”. How many of the past updates have been around how “great” the keyboard feels? They even had one update saying the manufacturer stated it was the best keyboard design of 2018. That’s great, but the Next team forgot one thing. To test that it actually works.

As someone pointed out here, there was a post on Kickstarter asking if there was a “Plan b” and unfortunately I don’t think there is going to be one. They seem far too stubborn to give up on the modern keyboard feel, after all this problem could have been fixed months ago with the same two membrane design as the original Spectrum or even Spectrum+. But they started spending too much time on the individual keys, the mechanism behind them.

Honestly, with the updates from the past 9 months it feels like I’ve backed a new keyboard than a new Spectrum.

[djnzx48]

Honestly, with the updates from the past 9 months it feels like I’ve backed a new keyboard than a new Spectrum.

Maybe they should rename it the ZX Spectrum Text.

[hikoki]

The problem has been and always will be that a purely mechanical membrane keyboard system is never good at making two independent switch contacts at the same time. Even during the days of Sinclair. Far too often on my ZX Spectrum+ keyboards, I get the wrong action unless I press the key squarely.

Would there be a mechanical way to detect when a hit is wrong? so that a tad of energy is released to perform the keystroke again. I don't know, a coil spring that only stores an releases energy at a certain angle. Maybe a magnet kind of thing, I don't say they build expensive magnetic keyboards, just something to aid a square press.

[hikoki]

Hall effect switches seem to be cheaper nowadays
https://drop.com/buy/xmit-hall-effect-m ... d#overview
who knows if the chinese manufacturer would be interested to support the Next batch to advertise this type of switch.

[toot_toot]

Honestly, with the updates from the past 9 months it feels like I’ve backed a new keyboard than a new Spectrum.

Maybe they should rename it the ZX Spectrum Text.

very good, sir!

I went through some of the previous updates and highlighted some of the keyboard specific updates. These ones are from two separate updates in October 2018

The most important feedback here is that the key format feels incredible good -- in fact, even the keyboard manufacturer, a specialist in keyboards with years of experience and dozens of models under their belt -- praised how good the Next keys feel to the touch, and how accurate the keyboard becomes with the key shape compared to a typical, modern keyboard.

Enter and Space, now on a sexy angle

They are looking rather nice already.

And this one

We’re happy to announce the membrane was produced as planned, and… It is performing as expected, within the 45gr-55gr of pressure required to result in a responsive and elegant keyboard for the Spectrum Next. What an improvement from the dreaded +200gr imposed by the third layer a few weeks ago!

(I made a mistake in my earlier post, they're going for a two membrane design instead of the three membrane in the original Spectrum)

That membrane update was from July 2018 - nearly 12 months ago and it looks like we're not any further ahead, it's taken 10 months to identify that the membrane design doesn't actually work, on a project that's now 15 months behind schedule.

Looking at the original Kickstarter Campaign page

Rick also took great care to create a keyboard that’s more responsive and features higher bandwidth input than the original, using a butterfly mechanism to power a tactile response to any fingertip touching its keys. No more stuck keys while you type!

That one line from the Kickstarter Campaign appears to have been the biggest part of the delay. Yet other features that were touted at the start have been dropped or quietly forgotten - ULAPlus, 14MHz boost mode, SID chip support, Rex Next, the new Dizzy game, Nodes of Yesod. The team clearly have made a decision to drop some features, yet have stuck doggedly to the keyboard design.

Lets go back to June 2018

With the Battle For The Case pretty much won (famous last words), we move onto the final challenge in production: the three-layered keyboard matrix. In order to keep the keyboard's and Next full compatibility with the original Speccy, the keyboard matrix needs to be composed of three layers like the Spectrum Plus and 128 in order to support composite keys such as BREAK, DELETE and the ARROWS. As a result, the amount of pressure required for registering a keypress is too high -- around 200g instead of the ideal range of 45-55g to get it as responsive as Rick Dickinson envisioned it.

But, why is it we really need a modern "feel" keyboard on the Spectrum Next?I suppose it's great if you're going to be typing a lot, but is the Next really going to be used for that? Even text adventures are going to be limited with the amount of text you're typing in! If you're developing anything for the next, surely you're going to be using a PC to develop? Why is it dreaded to have +200gr which I presume is the same as the Spectrum+ or 128k Toastrack? Yes, the original Spectrum keyboard wasn't ideal, but at least it worked and there's plenty of after market membranes available that would have done the job just fine.

Now the project is hinged on the keyboard manufacturer testing out 3 membrane designs, the team are not even doing the testing now to save time, which might reduce risk in hitting the promised delivery date of Q2, but significantly increases risk to reliability and functionality. If none of those 3 designs work, then looking back at the update in June 2018:

In order to solve this problem, the keyboard experts are trying a myriad of solutions:

Increasing the hole size in the middle layer
Adding extra material on the activation point on the membrane
Using thinner PET layer
Changing the rubber dome to another type
Increasing the rubber dome size
And lastly, abandoning the middle layer and breaking each activation point from a circle to 2 semi circles, forming two contact points

Does that mean that no other option worked apart from the two membrane option... which now doesn't work. But, why has it taken nearly 12 months to identify that it doesn't work? The project was already delayed by 6 months when that decision was made.

Someone pointed out on Kickstarter that since the board only option was launched, there seemed to be a real lack of drive in getting the cased Next out on time. Unfortunately I have to agree with that.

[beanz]

14MHz boost mode, Rex Next, the new Dizzy game, Nodes of Yesod.

Those haven't been dropped. The 14mhz is fully functional and has been, you can choose 3.5 / 7 / 14. I think you mean 28mhz was dropped.

The others are stretch goals and from what I understand are still in development. There have been updates on Rex Next progress. The other two I don't think I've seen updates on but that doesn't mean they've been dropped.

The games are also being developed by 3rd party contributors, not the next team so progress on those..or lack thereof doesn't have any effect on the progress of the Next itself.

[Alcoholics Anonymous]

That one line from the Kickstarter Campaign appears to have been the biggest part of the delay. Yet other features that were touted at the start have been dropped or quietly forgotten - ULAPlus, 14MHz boost mode, SID chip support, Rex Next, the new Dizzy game, Nodes of Yesod. The team clearly have made a decision to drop some features, yet have stuck doggedly to the keyboard design.

????

ULAplus was replaced with ULANext because of disagreements that I am not privy to. ULANext fits better with how the palette is handled in the rest of the Next's display layers. I would not choose to use ULAplus except for compatibility.

14MHz is there - where did you hear it isn't? 14MHz was never in the original kickstarter - only 7MHz was. This was a bonus that came later.

SID was never promised. It was experimented with and a demo shown. People cried it was C64, a prominent AY musician gave up his pledge over its possible inclusion, the prospect of not being allowed to call it a SID was raised. Par for the course in the Spectrum scene.

Rex Next, the new dizzy game, the nodes of yesod are still being developed. In the meantime there are at least 30 other games being made.

But, why is it we really need a modern "feel" keyboard on the Spectrum Next?I suppose it's great if you're going to be typing a lot, but is the Next really going to be used for that? Even text adventures are going to be limited with the amount of text you're typing in! If you're developing anything for the next, surely you're going to be using a PC to develop?

This is 2019. People use keyboards in their daily lives and have expectations for them, especially since they are the primary interface to the machine. The quality of the keyboard can impact on the perceived quality of the whole.

Will the keyboard be used for development? Maybe since there is an optional pi attached that can allow cross dev tools to be used on the machine. I know there are those who want to try it.

The keyboard itself seems to have turned out well. It is the last step - the electrical contact on composite keys that needs to be made satisfactory.

Someone pointed out on Kickstarter that since the board only option was launched, there seemed to be a real lack of drive in getting the cased Next out on time. Unfortunately I have to agree with that.

Come on now.

[Pegaz]

SID was never promised. It was experimented with and a demo shown. People cried it was C64, a prominent AY musician gave up his pledge over its possible inclusion, the prospect of not being allowed to call it a SID was raised.

You are kidding?
Full SID support, can only be a great bonus feature and cant hurt anyone.
Prominent immaturity and stupidity, I would say...

[Alcoholics Anonymous]

You are kidding?
Full SID support, can only be a great bonus feature and cant hurt anyone.
Prominent immaturity and stupidity, I would say...

Nostalgia and passion is a weird & personal thing.

[beanz]

Remember this is supposed to be open source when everythings done and dusted...meaning people can write their own cores. If you want sid and other features I'm sure at some point cores will be released for it.

[toot_toot]

That one line from the Kickstarter Campaign appears to have been the biggest part of the delay. Yet other features that were touted at the start have been dropped or quietly forgotten - ULAPlus, 14MHz boost mode, SID chip support, Rex Next, the new Dizzy game, Nodes of Yesod. The team clearly have made a decision to drop some features, yet have stuck doggedly to the keyboard design.

????

ULAplus was replaced with ULANext because of disagreements that I am not privy to. ULANext fits better with how the palette is handled in the rest of the Next's display layers. I would not choose to use ULAplus except for compatibility.


14MHz is there - where did you hear it isn't? 14MHz was never in the original kickstarter - only 7MHz was. This was a bonus that came later.

SID was never promised. It was experimented with and a demo shown. People cried it was C64, a prominent AY musician gave up his pledge over its possible inclusion, the prospect of not being allowed to call it a SID was raised. Par for the course in the Spectrum scene.

Rex Next, the new dizzy game, the nodes of yesod are still being developed. In the meantime there are at least 30 other games being made.

This is 2019. People use keyboards in their daily lives and have expectations for them, especially since they are the primary interface to the machine. The quality of the keyboard can impact on the perceived quality of the whole.

Personally, I think ULANext is not the same as ULAPlus as it's just essentially a completely new Video Mode for the Next making Next only games look better, more unique and ultimately less like a Spectrum. While I understand ULAPlus isn't to everyone's tastes, at least the games still look like Spectrum games, albeit slightly more enhanced graphically.

https://www.youtube.com/watch?v=OlP76nWAtDg

Personally again, I would have preferred the ULAPlus, but if it's been dropped then that's fine.

Yes, 14MHz is still in there - sorry I forgot it was 28 MHz that was dropped. However 28MHz boost was announced as an official update in June 2017 here: https://www.specnext.com/new-turbo-modes-14-and-28mhz/

(with a retracted update from January 2018 saying 28MHz wouldn't be available).

SID might have been one of those features that was hinted at, but the FM sound was definitely listed in the Campaign (in fact it's still there), but that too seems to have been quietly dropped.

There have been no updates on the status of Rex Next, Dizzy or Nodes of Yesod in months - all three of those games were promised as part of Stretch goals (which were met)



But my point is that many features were promised but have been dropped. I'm OK with that as they've made the decision that it would either take up too much time or resources to implement them.

But what they haven't dropped is the Modern Feel Keyboard and while that's is nice to have, let's be honest it's caused significant headaches and delays. They've doggedly stuck with the keyboard design, as with my original post making the recent Next updates feel like I've backed a keyboard, not a new Spectrum.

Has the old 35+ year old keyboard design of the original Spectrum stopped people playing them? No! I still play my original Spectrum - in fact you could say the keyboard feel adds a bit of genuine experience!

[toot_toot]

Remember this is supposed to be open source when everythings done and dusted...meaning people can write their own cores. If you want sid and other features I'm sure at some point cores will be released for it.

Sorry to be "moaning" again, but yes you're right. This was an Open Source project, the Campaign stated

Most importantly, once launched, the Spectrum Next will be fully open source, including its firmware and schematics, meaning the community will be able to take it further into the future by improving both its software and hardware!

At the time people were excited at the thought of the same board design being used for other projects, like a Commodore 64 Next, MSX Next etc, after all the FPGA is reportedly capable of handling most 8-bit computers (I believe 16-bit computers were a bit too demanding).

The board only version of the Next was released in November 2017, but reportedly the team haven't released the source code or schematics in fear of clones being made before the cased Next is released. That's fine, but now that they're running 18 months behind schedule, that's 18 months lost for other teams using the same, open source, design for other machines or even improvements by the community to the Next itself.

But what worries me is that the team are now talking about a second Kickstarter campaign once the cased Nexts are released, does that mean Open Source will also be quietly dropped as any clones could impact the success of the second Kickstarter campaign?

[beanz]

Yes I was wondering myself about the Open Source situation now they have pretty much announced another run of not only the cased next but the board only option too.

I can see it being a bit of "shoot yourself in the foot" to release the docs and then try to sell more of them. Still, I doubt anyone has the resources and/or inclination to create the case...the board only though should be pretty much something anyone could easily replicate.

[Alcoholics Anonymous]

Personally, I think ULANext is not the same as ULAPlus as it's just essentially a completely new Video Mode for the Next making Next only games look better, more unique and ultimately less like a Spectrum.

I think you have a misunderstanding of what ULAnext is - it serves the exact same purpose as ULAPlus. Both change the palette of the existing spectrum colours but do nothing to the colour resolution. You can colour existing games using ULAnext just like you can with ULAplus.

There are two things you can do in the next.

One is you can use the spectrum like normal and you can change the palette entries for ink, bright ink, paper, bright paper to different colours (each colour is 9-bit). So you could change ink 2 (red) to something else, eg, brown. Then all red ink pixels will be brown. Bright red ink is different - you could change that independently to orange, eg. This is straight up colour substitution.

ULAnext is a step further. It changes the meaning of an attribute byte by letting you specify how many of the rightmost bits in the attribute byte will be considered the ink colour. The rest are taken as the paper colour. ULAplus does the same thing but the rightmost 3 bits are always ink and the next 3 bits are always paper. This would be equivalent to a ULAnext format byte of 0x07. Where they depart is ULAplus uses the bright and flash bits to select different palette sets whereas ULAnext only uses the flash and paper bits to extend the paper colour selection (or ink selection if the format byte has more bits set).

Yes, 14MHz is still in there - sorry I forgot it was 28 MHz that was dropped. However 28MHz boost was announced as an official update in June 2017 here: https://www.specnext.com/new-turbo-modes-14-and-28mhz/
(with a retracted update from January 2018 saying 28MHz wouldn't be available).

It crashes the machine. It's not really useful if it's going to do that

Everyone is committed to providing the best machine possible but there's also an eye on what was promised in the kickstarter as that is what people pledged money for. In the kickstarter, 7MHz is what was promised so getting 14MHz is icing.

FM sound was definitely listed in the Campaign (in fact it's still there), but that too seems to have been quietly dropped.

Yes FM has been dropped because it takes up too much space. You'd have to choose other things to remove to make space for it. The SID was being considered as the FM option as was this:

https://www.youtube.com/watch?v=mJLFV96Tt7c

Another option being considered is enhancing the AY. But what can be done depends on priorities and available fpga space. The core is not done yet - it's still being developed.

There have been no updates on the status of Rex Next, Dizzy or Nodes of Yesod in months - all three of those games were promised as part of Stretch goals (which were met)

The oliver twins were taking the new dizzy game design + screenshots on tour to many retro events last year. The game will be for both the 128k and the Next with the Next seeing some enhancements (but not too many from what I have seen - multiple AY and clashless sprites are what I noticed. Perhaps they would change the palette too as that's something easy to do). I think the dev is fairly far along.

Rex Next and Nodes are being done by the original programmers. I haven't seen much from them but it is still going. This is dependent on their spare time.

There's a wip video for the 128k version of wonderful dizzy on the specnext games site:
https://www.spectrumnextgames.uk/#Games

But my point is that many features were promised but have been dropped. I'm OK with that as they've made the decision that it would either take up too much time or resources to implement them.

I don't think very much has been dropped.

Processor: Z80 3.5Mhz and 7Mhz modes
Memory: 512Kb RAM (expandable to 1.5Mb internally and 2.5Mb externally)
Video: Hardware sprites, 256 colours mode, Timex 8x1 mode etc.
Video Output: RGB, VGA, HDMI
Storage: SD Card slot, with DivMMC-compatible protocol
Audio: 3x AY-3-8912 audio chips with stereo output + FM sound
Joystick: DB9 compatible with Cursor, Kempston and Interface 2 protocols (selectable)
PS/2 port: Mouse with Kempston mode emulation and an external keyboard
Special: Multiface functionality for memory access, savegames, cheats etc.
Tape support: Mic and Ear ports for tape loading and saving
Expansion: Original external bus expansion port and accelerator expansion port
Accelerator board (optional): GPU / 1Ghz CPU / 512Mb RAM
Network (optional): Wi Fi module
Extras: Real Time Clock (optional), internal speaker (optional)

Everything in this list is in except the FM sound.
The expansion bus and accelerator are still seeing some work.

Stretch goals:

(done) there must be two
(done) box of memories
(active) nodes of yesod
(active) internet toolbox
(active) we play more
(done) holy manual
(done) never forget
(active) make me dizzy
(active) rex next
(done) expand me easy
(done) stronger zx next

Let's not forget that there have been additions way beyond these lists in both the software and hardware. As just one example, neither Sinclair nor Amstrad ever contributed as much to the operating system as what is in NextZXOS.

[Firefox]

...
Memory: 512Kb RAM (expandable to 1.5Mb internally and 2.5Mb externally)
...

Did they really mean 512 kilobits of RAM, or did they perhaps mean 512 kilobytes?

[Seven.FFF]

Kilobytes lol.

So arguably the original goal of 512 Kilobits and the stretch goal of 1 Megabits have been vastly exceeded! Even more so if you consider Kb and Mb to denote 1,000 and 1,000,000 respectively, like nobody sane did ever unless they were referring to a drive storage context.

[Sokurah]

Did they really mean 512 kilobits of RAM, or did they perhaps mean 512 kilobytes?

Please tell me that wasn't a serious question?