ZX Yellow Spectrum Lines Computers
ZXNet echo conference «hardware.zx»
Hello, All
The idea to make a cheap alternative to Spinter was born at the turn of 1997 and 1998
years as a result of my discussions with Igor Maznitsa about development prospects
platform hardware. General principles of architecture were developed,
The technical specifications were developed and work was carried out on the schematic diagram. The procedural itself
We called the processor “Tornado” and received the future computer in 1999
year name Fenix-999. Then, after the well-known events in 1999, Igor
renounced his copyright on the computer and resumed work on
ZX Polyhedron architecture. I never gave up on my ideas and
intentions to resume work on the project if the opportunity arises. B
In the near future I intend to revive the project, strengthening it with an A26 chip :)
Hello andrews
and> In the near future I intend to revive the project, strengthening it
and> A26 chip :)
what is A26?
Hello andrews
It’s fun to read the good correspondence between two old friends :)
There’s only one thing I don’t understand - someone is trying to surprise the consumer in 2007 with hardware
mp3 decoder in the device and even relies on it? Hmm..
And it’s easier to attract thousands of new users with software rather than with a platform... Of course, if
only you are not Sony, Apple, etc..
In order not to be unfounded, remember the history of the Amiga computer.. Very
instructive.
But polyhedron is really interesting. Igor, it would be very interesting
see (at least under an emulator) how it will work live (I'm talking about
coloring old toys).
Hello Raydac
Ray> what is A26?
hardware mp3 decoder, details here
http://zx.pk.ru/showthread.php?t=3808
Hello Raydac
The main thing is to make an inexpensive gaming and multimedia computer for your home.
based on ZX Spectrum and attract thousands of new users.
Hello andrews
and> hardware mp3 decoder, details here
and> http://zx.pk.ru/showthread.php?t=3808
Clearly, then the main thing is to stock up on beds for hardware decoders
MP4-MP10, what would be the future so to speak...
Hello Cooper
And if the main goal is not to surprise, but simply to make the Spectrum convenient for home
multimedia computer? There will be someone to surprise without me, I would like a computer
do something normal for yourself first of all!
Hello andrews
Therefore, I declare this project initially not commercial, but copyright
belong to the authors who improved it. Schemes assumed
designate something like this
ZXY-yearday and monthmodifications-city-country, for example
ZXY-071301-SPB-RU, the circuit identifier is needed to avoid confusion with
firmware and for the convenience of drawing up documentation.
Now the main idea of architecture. This is an inside out emulator implemented in
gland. Have you read "The Head of Professor Dowell"? The head in this case is the processor
z80. Around it is an abstractly more powerful computing environment, to which
The processor is connected with all its pins.
Hello Cooper
Coo> But polyhedron is really interesting. Igor, that would be great
Coo> it will be interesting to see (at least under the emulator) how this will work
Coo> live (I'm talking about coloring old toys).
It’s a pity that time has passed, if the waves of computerization had not arrived here
residual principle, but as in the entire civilized world, it turns out that
ZX-Poly would have appeared in 1987/88, and look, maybe the pussy wouldn’t have caught on
:) .. ideas in the "parapunk" style (who doesn't know, this is the style of Japanese cartoons
about an alternative history where robots and computers are paired)
The emulsion is almost ready, you just need to finalize it, but the toy needs to be painted
the editor will scribble something, otherwise it’s somehow unaesthetic with his hands :)
Hello andrews
and> The head in this case is the z80 processor. More abstractly around him
and> powerful computing environment to which the processor is connected by everyone
and> with your pins.
In 96, I remember I had an idea (against the background of the rapid decline in price of 386x computers),
install them with a BIOS with a spec emulsion, like the spec seems to be fast... that's how it is
it came...
Hello Raydac
There is already a thread about ZX Poly, why spread it across the neighboring ones? As for the ZX
Yellow, such a brilliant idea clearly deserves a patent, but I’m giving it away for free
to all spektrumists. I’ll write documentation, and as the first hardware
implementation m.b. one of the starter kit boards from Alter, Xilinx and
perhaps even a fast multi-legged controller with ARM7/9 and DSP from TI. The main thing is
so that the USB-device or USB-host is miniSD and has 64-256 megabytes of memory.
Hello andrews
What is fundamentally new in this idea is that the head (ZX80) is mounted on a powerful
architecture and when the user program needs to execute the z80 code this
the processor, fully controlled by the main part of the system, performs all its
real-time machine cycles. The progress of any program is under
100% control.
Hello andrews
and> What is fundamentally new in this idea is that the head (ZX80) is mounted
and> on a powerful architecture and when to the user program
Sprinter did this, but it didn’t make him any happier, since everything is right there
realized that it is more profitable to make your processor and firmware in FPGA, and the Z80 as
"wedding general" just says "start/stop"
Hello, Vladimir Kladov
Vla> in general, since I dealt with this, I’ll say right away (and this
Vla> discussed): not everything can be repainted like this
It’s better to discuss these things in another thread.. I never talked about 100%
adaptability of games, but as can be seen in the example of ATW and other games that are on Spec256
ported, their percentage is not so small and perhaps the most important thing is that they exist..
Spec256 was probably abandoned because there was nothing with it other than coloring games
It’s simply physically impossible to do.. narrow idea
Hello Lethargeek
Let> So there is EmuZWin, which does the same thing, you just need to write
Let> converter of saved files from 8 planes to 4.
Let>
The EmuZWin architecture is designed specifically for emulsion, it does not suit me, since the work
with snapshots only on emulsion and is designed, you need a converter that saves in
something that can be loaded on "real"
Hello Lethargeek
in general, since I dealt with this, I’ll say right away (and this was discussed): far
Not everything can be repainted this way. There are toys with sprites
are not stored entirely, and not for all possible rotation options. There are
Problems with xor operation in the screen area. And there are cases when sprites
stored in packaged form and unpacked as needed, for example,
upon entering the "room". Such games even have special processing inside
The processor cannot be remade. I think the original Spec256 project was stopped
precisely for the reason that it was discovered that without reworking the code it would still be
Only a very small number of toys can be painted. No, it's not the right platform if
you need VGA, not that one...
Hello Raydac
Yes, please, about “the mountain gave birth to a mouse” in another thread.
Hello Raydac
Nothing like that! No one has done THIS yet... you just didn’t get it!
Hello Raydac
Ray> yes, in order to color the toy, you will need to scribble some kind of editor
Ray> otherwise it’s somehow unaesthetic with your hands
So there is EmuZWin, which does the same thing, you just need to write a converter
saved files from 8 planes to 4.
and> What is fundamentally new in this idea is that the head (ZX80) is mounted
and> to a powerful architecture
For some reason I immediately imagined the microcephalic muscleman from the film “Mario Brothers”. :)
Hello andrews
Firmware in FPGA z80 is also completely different, so that it is what I propose
you need to flash another 32-bit processor into the FPGA along with the z80, but why waste it?
valves for z80? This is both labor-intensive and completely unnecessary.
Hello Raydac
Ray> The EmuZWin architecture is designed specifically for emulsion, it doesn’t suit me, so
Ray> how to work with snapshots only on emulsion and is designed, you need a converter
Ray> which saves to what can be loaded on "real"
What other architecture? The storage format for unpacked graphics is the same everywhere. And with
packed, etc., you still need to get into the code in any case...
Ray> Spec256 was probably abandoned because it was used in addition to coloring games
Ray> it’s simply physically impossible to do anything.. narrow idea
What, were there any other goals? Emulate four independent processors
it’s unclear why? But it’s interesting, if it appears on ZX-Poly, someone will
anything to do "besides coloring games"? Even more traditional clones
non-special software is not particularly supported...
Hello Lethargeek
Why do people look for new solutions? Because the old ones have basically exhausted themselves!
A processor that cannot multiply two even eight-bit numbers in hardware in 1 μs by
what is he good for?
But there is a wealth of ready-made software for it, the development of which at an average price
programmer, taking into account invoices in Russia, more than $1,500 per month. This, not
the scheme chewed up and down is the main circumstance of our interest. But
Exactly the same free software is available for ARM7, for example. So pour into
The “new blood” platform can be “merged between a hedgehog and a snake” in one way. It's possible
done in many ways. Which one will be the most effective will show
only time. I also understand very well the people who remain on the classic Spectrum,
As they say, “a bird is better in the hands.”
Hello andrews
and> I focus on microcircuits actually available in St. Petersburg... Then
and> I'll order one of the more advanced people here
and> programmable video controller. This is my program of action.
I hope you, in addition to your program of actions, as well as the list of microcircuits, also
post something that would allow the public to understand what you are doing
you're going to :) and at the same time you'll inform everyone about the progress of "work on
BAM"
Hello andrews
But what's the point of asking questions there and needing to process them?
and your processor will spin in a stupid loop simulating which $3 CPLD thread
And then slower
Hello andrews
ARM7 will not work
ARM9 is stupid
Hello hero
At 66 MHz, RISC with a clock per command will not be able to expose it to the buses through
port/poll 4MHZ CISC ports with the fastest command (not just any one! note
itself, but accessing asynchronous external memory and ports through its pins) in
4-6 or more bars?!
Hello hero
I hope ARM7 will still have time for processing, like, address the desired cell
internal memory, load the next instruction, etc.
What about spinning idle, but when the program is running on the z80, then of course
ARM should only simulate instruction memory, RAM, ports, interrupts, etc.
But it can do a lot of other things with OS support, debugging and hacking
programs, etc. etc.
You can write new programs taking into account the fact that there is ARM, noting in the source code,
the most slow procedures are like slow and then the compiler will no longer compile the code
for z80, and for ARM. You can also rewrite old sources if they
saved, or by first cracking the code with the help of this Centaur :)
get the sources and remake them already.
Hello Raydac
I'll post the results when they actually arrive...
I have already explained what it will be:
ARM7+Z80...to connect the latter, I came up with the idea of turning it on simply as in
debugger...i.e. every pin z80, excluding clock, power and
some others are connected to the I/O port of a microcontroller with ARM7...performance
the latter should ensure that the z80 operates in real time so that "it doesn't
guessed" that his external signals were "simulated" and "controlled"
All this could be done on corduroy in order to “please the orthodox,” but why,
if there are inexpensive and high-performance microcontrollers that can
do the same thing as logic?
You also need to put some kind of calculator into the logic, but here it is ready. Besides
here is static memory
256 kbytes.
What's missing is a video controller, and a lot of small things, you'll still need them
put away in a pleat.
I will draw up other details in the form of a document, but I haven’t gotten around to it yet. Besides
I would like to start with the actual hardware layout. Maybe I'm in something
I miscalculated, why then cast a “shadow on the fence”.
Hello andrews
and> What is missing is a video controller, and a lot of little things, too
and> still have to be put away in pleated fabric.
Regarding the video controller - pay attention to the SED13xx and S1D lines from
Seiko-Epson. They should be available in Russia, cost about $10, have
the simplest accelerator. There are versions with their own video memory on board. Here
for example
http://www.epson-electronics.de/cgi-bin/panamafe/panama/demand/catalog/browseCa
talog.do?colorStyle=green&tabId=1&categoryOid=-10083&BV_SessionID=@@@@070143933
6.1169034926@@@@&BV_EngineID=ccccaddjiimgeejcflgcefldhgjdhkg.0
Hello hero
Okay, but at home.
This will not affect the idea itself, either there will be a 100 MHz ARM, or several will have to be
lower the operating frequency for the z80...but not to 1MHz
Hello andrews
Write an example for at least 128K memory emulation and applying RESET according to the specified
address
Hello andrews
All the same, the idea is fundamentally wrong to make a CPLD from a processor
a little more expensive than one ARM, the FPGA is large enough to implement
hardware debugger functionality, etc. Yes, and the Z80 will fit there if it wants to
Hello hero
her> Anyway, the idea is fundamentally wrong to make a CPLD processor out of it
her>
There are no wrong ideas... there are harmful or unacceptable ones :)
But knowing Andrey, I recommend everyone not to criticize his ideas, but rather ask more often
"Where? WHERE?!" .. let him at least do something .. otherwise he’ll say later that they ruined the idea
Hello Raydac
No, I’m not just trying to write something real-time on ARM for the spec
As a result, the performance was not enough, although the device is less real-time than
suggested here.
Hello Raydac
yeah it's not interesting :)
And at 3.5 it will be difficult without debugging about 7 MHz is unrealistic
Hello hero
her> No, I just tried to write something on ARM real-time for the spec
her>
her> as a result, the performance was not enough although the device is less real-time
her> than what is proposed here.
He also wrote how he was going to get around this underwater and obvious stone.. Z80 with
clock frequency of 500 kHz will save the father of Russian democracy
Hello andrews
and> Why do you use the same methods that you don’t like in relation to
and> to your project from other people, are you using it here with might and main? You
and> do you think this is fair competition?
I’ll tell you a secret, I don’t consider competition at all... as for the best or
not the best of times, then I rather persuade people to be gentler with you, so that new
the beginnings did not end up in the same place as the old ones, but they accumulated as if abandoned
trucks behind the "perky tits" bar (from the movie "From Dusk Till Dawn").
Just in the last two months you have started several full-scale projects
(judging by the entries on this forum) and if they start criticizing you, then you
you will get bogged down in correspondence and answers and we will retire without seeing yours
creations..
Hello andrews
and> I myself perfectly understand the restrictions imposed, including
and> performance...
and> At the same time, for example, I don’t understand why the critics of my approach
and> believe that this is just a replacement for FPGAs, overlooking many others
and> positive aspects (presence of internal static memory 256K,
and> powerful 32r. calculator, sequence ports, etc.)
Because it’s cheaper and more convenient to do this on an FPGA. In your assigned task
It turns out that you are using the workstation “for other purposes”. Naturally your right, but
Bus monitoring and event delegation from/to Z80 can be easily done with a simple FSM,
which just screams about FPGA :) Moreover, why do you need 256K in this case? You
Are you somehow going to manipulate the data from the bus before delivering it to the Z80?
Hello icebear
All engineers present here are experienced (I work in the field of Embedded System
since 1983, started on i8080 Intellec MDS Series-II) understand that not in a day,
You can’t master new technologies for yourself in a week. I have about six months ago
there were some experiments on ARM7 Philips LPC, I bought a book, installed IDE c
GCC Keil, allowed examples), initial experience. Now downloaded IAR Embedded
Workbench for ARM version for quick start with 32K code limit. Why
every day we need to exaggerate the topic that “knowing Andrei”...Igor knew me not in
the best times for me and continues to extrapolate into the future. Igor,
This is ultimately bad, I declare publicly. I'm not getting into your project, and I wish
I wish you every success in its implementation. Why are you using the same methods?
who are disliked in relation to your project by other people,
making full use of it here? Do you think this is fair competition?
Hello hero
her> I just tried to write something on ARM real-time for the spec
as they say, this is “something” for the studio, plz... regarding interesting or not interesting
It’s generally pointless to discuss...some are not interested, and some are interested...
about what to do Igor, start with yourself (about hardware)
Hello andrews
and> as they say this is “something” in the studio, plz...
Look again at the topic "ARM - Speccy processor of the 21st century." There's a couple there
convincing arguments both for and against. In general, one can calculate
How much do you have, for example, for one Z80 machine cycle at a certain frequency?
workstation cycles are leaving and try to fit into this gap using the code so that
The workstation managed to process the state of the Z80.
Hello icebear
ice> In general, you can calculate how much you have, for example, for one machine
ice> the Z80 cycle at a certain frequency takes ARM clock cycles and try it
ice> code to fit into this gap so that the workstation has time to process
ice> Z80 state.
I myself perfectly understand the restrictions imposed, including
productivity...
At the same time, for example, it is not clear to me why critics of my approach believe that
that this is just a replacement for FPGAs, overlooking many other positive
moments (presence of internal static memory 256K, powerful 32 rubles computer,
afterbirth ports, etc.)
Hello andrews
That is, for 66MHz and 3.5MHZ, the fastest ARM instruction is executed at 37.7
faster...it's like 486DX66
Hello andrews
As I understand it, everything is determined by a fairly simple thing, and in particular
the number of executed ARM commands per Z80 clock cycle (3.5 MHz). Can anyone
tell how much ARM7 is currently being developed (maximum performance
stone) can give performance per clock cycle of the Z80? otherwise I don’t know about ARM
protsov..
Hello andrews
and> with the size of standard ZX Spectrum video memory this should not take
and> a lot of time, if there is a CPDP, even more so
You already wrote here and for ARM9 at 200MIPS it won’t be boring
ice> you can only make software on FIQ
ARM7 in FIQ includes current for 24 core clock cycles (1.5 clock cycles of the Z80 processor) and also
transfer, that is, per second, a current of 8 million cycles of the current core goes to the input
interrupts
And don’t forget that with ARM, each transition is 3 penalty cycles, and the code for that
what you're describing doesn't smell like linearity
Hello icebear
How many instructions would the z80 have to select FROM MEMORY, programmatically executing
multiplication. ALL this time it is now resting as ARM7 does it
fast enough to not miss frame updates. But a note
Essentially, I agree, the video controller should not be another active one
system device with such system memory. ARM7 should dump into it
internal video memory for the next frame quickly and completely.
Hello icebear
ice> You are somehow going to manipulate the data from the bus before delivering it to
ice> Z80?
Of course! A simple example: rotations on the plane of objects before displaying
video controller. Who will make them faster: z80 without multiplication or arm7? That's what
and the thing is that since they have a common memory, and besides, the z80 only works
when ARM7 allows it (simply due to the asynchronous operation of the z80 with
external memory), then there is simply no problem of dividing this shared memory.
Hello icebear
ice> How many Z80 instructions will you skip in this case?
if the screen has not been updated, then there is no need to forward it... in the case of a slow z80, everything
would the screen have to be turned off during the transformation? what prevents you from enrolling
same here, but to extinguish it for a much shorter time...
>What does it mean to implant?
the z80 does not have instruction code for hardware multiply, it is required in your
in my case, enter it, not in mine...since during compilation in this place it will simply be
code for arm7, and when executed, since the entire command flow goes through arm, it
will know that this command is for him and issue it through its parallel port
no need, no need to go through gcc...you just need a cunning linker and
linker of object modules, so you have to write it to get
the resulting mixed-code.
Hello icebear
ice> Why extinguish?
Well, skip frames... it’s the same here, but it’ll slow down less anyway
ice> All that remains is to teach the entire gcc to additionally understand Z80 assembler
why is this?
you skip the source code first with the -arm7 key,
then with the -z80 switch
And the rest is conditional compilation, or you can automate it by passing through
a code analyzer that knows which operations are slow on the z80.
I'm telling you, you need a cunning linker.
ice> Since when has the PIC executed commands in one clock cycle?
well, in two...except for the transition commands and something else...it was implied that
ARM commands vary greatly in execution time
Hello icebear
ice> Because modern FPGAs already have multipliers that work
ice> inside at very high speeds.
that's not enough! such a multiplier must also be implanted into the kit
z80-core commands and supported by the compiler. GCC iterate?
Hello icebear
"Normal data processing 1s;
Processing data with the number of shifts specified in register 1s+1i;
Data processing with writing to the PC 2s+1n register;
Processing data with the number of shifts specified in the register and writing to the register
PC 2s+1N+1i;"
Quoted from "ARM7 Instruction Set Overview"
Hello andrews
From 256K - 128K (at least) this is regular ZX Spectrum 128K memory!
Hello andrews
Here, perhaps, is one of the options for mixed code stream;
code; ARM7...non-existent command code ARM7, code Z80...RST ?, code ARM7... etc.
If the z80 had processing of non-existent commands, or if it did, I don’t remember.
Hello andrews
In general, comparing performance with the z80 is a separate problem.
Hello andrews
The “trick” of the linker is precisely in organizing a mixed code flow
instructions...for non-existent commands, ARM7 will have an "exception" and then
a piece of code that supports execution on the z80 will take effect. It's foggy because
I haven’t quite figured out how to do this yet, there are many options here.
Hello andrews
and> Here, perhaps, is one of the options for mixed code stream;
and>;
and> ARM7 code...non-existent ARM7 command code, Z80 code...RST ?, code
and> ARM7... etc.
That is in my opinion, the software interrupts of the workstation will be used (exclusive
situation). It won't happen quickly.
and> If the z80 had processing of non-existent commands, or if there is, what do I
and> I don’t remember.
The Z180 has it, for example, but it’s somehow complicated. It seems like there was something more in the eZ80
advanced in the Z80 line.
Hello andrews
and> Of 256K - 128K (at least) this is regular ZX Spectrum 128K memory!
Oh! Those. half of the ARM code has already been chopped off :) Now think about what
since all the Spectrum memory is sitting in your workstation, then it turns out that the workstation needs to somehow
generate a video signal, or when interrupted, break away from the main work and
output the current displayed line data to an external video controller. That's all
this is in the middle of processing the Z80 machine cycle.
Hello andrews
and> Of course! A simple example: rotations on the plane of objects before output
and> to the video controller. Who will make them faster: z80 without multiplication or
and> arm7?
FPGA. Because modern FPGAs already have multipliers that work inside
at very high speeds.
Hello andrews
and> How many instructions would the z80 have to select FROM MEMORY, programmatic
and> performing multiplication. ALL this time he is now resting because ARM7
and> does this fast enough so as not to miss frame updates.
and> But the remark is important, I agree, there should not be a video controller
and> another active system device with such system memory.
and> ARM7 should dump internal video memory into it for the next one
and> frame quickly and completely.
Yeah, that is. a frame buffer also appears, which must be filled according to some
event (for example, the deactivation of a blanking signal, aka the beginning of an active region).
Now take the screen size - 6912 bytes you need to transfer from the internal
memory to external. R40008 does not have a hardware DMA, you can only make a software one
FIQ, i.e. at best, you will need 6912 automated machine cycles for transfer
(all FIQ handler preparations are omitted here). How many Z80 instructions do you have?
will you miss it? Or will the Z80 hang on /WAIT? Then this is a slow clone
There will be one that also has an automated workstation on board. Plus you're going to do
manipulation of memory contents, rotation, etc. , as I understand it, there wasI mean video memory? Will you make it? But in FPGA this is all really possible
parallelize so that one will not interfere with the other. And squeeze out the speed
optimization is twice as high as that of the proposed workstation.
Hello andrews
and> and they are unified (CPU cores) so divide MHZ and multiply by 2...but
and> is a meaningless comparison because... ARM7 is not a PIC, it does not have all the commands
and> are executed in 1 clock period
Since when did the PIC execute commands in one clock cycle?
Hello andrews
and> if the screen has not been updated, then there is no need to forward...
That’s why the event (interruption, if that’s more convenient). In this case you will have to do the same
and> in the case of a slow z80, the screen still remains during the transformation
and> would have to be extinguished? what prevents you from doing the same here, but extinguishing it
and> for a significantly shorter time...
Why extinguish? What do you mean by extinguish?
and> >What does it mean to implant?
and> the z80 does not have instruction code for hardware multiplication, it is required
and> in your case enter, in mine not...since when compiling in this place
and> there will simply be code for arm7, and when executed, since the entire thread
and> commands goes through arm, he will know that this command is for him and
and> there is no need to issue it through the parallel port, there is no need to go through gcc
and> will be required...you just need a clever linker and object linker
and> modules, so you will have to write it to get the result
and> mixed-code.
Yeah! All that remains is to teach the entire gcc to additionally understand Z80 assembler +
teach to distinguish it from ARM code. Is it worth the price?
Hello andrews
and> well, skip frames... so here it’s the same, only less anyway
and> will slow down
Demomakers and sympathizers will tear it up :)
and> why is this?
and> you skip the source code first with the -arm7 key,
and> then with the -z80 switch
and> And the rest is conditional compilation, but can be automated
and> passing through a code analyzer that knows what operations the z80 has
and> brake.
and> As I say, a cunning linker is needed.
? Some kind of bullshit. After all the compilations, you get the firmware
binary, which you feed to the automated computer. What kind of mixed code are we talking about? How are you
at runtime you will tell the automated computer “but now this is not your code, but your ward’s,
therefore, give it simply as it is and do not fulfill it"?
and> well, in two...except for jump commands and something else
and> there...it was implied that ARM commands are very different
and> by execution time
For two, not for four? What do you mean by a large difference in time?
execution? Just at the workstation, all commands can be considered single-cycle.
Hello andrews
and> this is not enough! such a multiplier still needs to be implanted in
and> z80-core command set and supported by the compiler. GCC iterate?
What does it mean to implant? Expand the Z80 command system? Then the Z80 speed
as such will fall. Why such perversion? It will be faster on FPGA by the way
anyway. And again, GCC for new commands you will have to “sort out” and so on
yes.
Hello Raydac
Ray> ARM7 (maximum performance stone) can produce
Ray> performance per clock cycle of the Z80?
and they are unified (CPU cores) so divide MHZ and multiply by 2... but this
a meaningless comparison, because ARM7 is not a PIC, not all commands are executed in it
1 bar
Hello Raydac
Ray> Just in the last two months you have started several full-scale ones
Ray> projects (judging by the entries on this forum) and if they start you
Ray> criticize, then you will get bogged down in correspondence and answers and we will leave
Ray> to retire without seeing your creations..
if the criticism is not constructive, I will simply ignore it...what about
don’t worry about pensions, it seems I don’t give any guarantees to anyone and in the tender for
I don’t participate in the supply of computers to adherents... all this exoticism that you’ve me into
managed to get involved in 97-99. no longer has me...I have become wise, Igorek!
Hello andrews
and> internal video memory for the next frame quickly and completely.
with the size of standard ZX Spectrum video memory this shouldn't take much
time, if there is a CPDP, even more so
Hello andrews
and> By the way, the sources of the ARM7 emulator are nowhere on the internet?
There is a package called Proteus, you should really like it. There
there is a simulator called VSim, read it here
http://www.labcenter.co.uk/index_uk.htm
Hello andrews
By the way, are the sources of the ARM7 emulator anywhere on the Internet?
Hello icebear
ice> Demomakers and sympathizers will tear apart
but gamers will approve :)
Hello andrews
and> Process data with the number of shifts specified in the register and with the record
and> to register PC 2s+1N+1i;"
Everything except N can be 0, depends on the operands
Hello andrews
and> By the way, the sources of the ARM7 emulator are nowhere on the internet?
Sourceforge.net
Hello andrews
and> no iron
In general, I would now like to initiate a discussion of the switching scheme here... so that
there was a substantive conversation, download the file doc1354.pdf from www.atmel.com or
some other one, where there is an "excavation" of the AT91R40008 chip
Hello hero
Found in the sources:
Simlt-ARM, Generator ARM, The Amsterdam Compiler Kit.
But while the ZX Yellow Lines Spectrum emulator is still far away, I was more interested
ready tools for
development. I wanted to use IAR Embedded Workbench for ARM, but specialists with
Telesisov was persuaded to move to Keil, fortunately I played around with him at LPC,
I compiled it yesterday at night looking at the blinker with a bit of the port... no hardware... simulator
blinks :)
Hello andrews
and> discussion of the connection diagram
this is necessary for programming...I would like to discuss
1) interconnections z80 - AT91R40008 (it is clear that the data bus can be turned on
directly, with all other connections options are possible);
2) selection of external devices for EBI (can be up to 8 external.
devices, since the chip provides up to 8 Chip Selects, including external
memory, video controller, A26...WHAT else?) and Programmable Page size
Size 1,4,16,64 MBytes.
Hello andrews
and> it is capable, but you can’t fit it into 8-16 megabytes of memory... and 130 euros
and> too...memory will have to be collected from garbage dumps for 600 rubles. maximum
and> How many will 640x480 eat even 256 flowers?
scanner ~23 MB/sec +
blitter 70 MB/sec =
~95 MB/sec