ARM - Speccy processor of the 21st century
ZXNet echo conference «hardware.zx»
Hello hero
her> It’s not much simpler and I don’t think there’s a need to put it
her> mikruhi more than 144 paws
More specifically :-)
Hello caro
car> For example, this:
car> http://wwwcms.fh-pforzheim.de/inhal...te/motorola.htm
300 I
Not bad: And it’s not clear at what frequency it chirps.
Hello fan
fan> 300 I
fan> Not too bad: And it’s not clear at what frequency it sounds.
Judging by the processor, the frequency is up to 60 MHz.
Hello caro
car> Probably meant, for example, S3C4510B from SAMSUNG with an ARM TDMI7 core?
car> In terms of clock frequency, it is comparable to the eZ80 - up to 50 MHz.
car> And the cost is approximately the same - from 10 to 15 $.
car> I would even be glad if one of the adherents of automated workstations does it on it
car> system.
car>
Judging by the fact that not a single clone was received for more than $100 and probably
will not receive (?) wide distribution, then it’s not a matter of choice
processor. You can't make it cheaper than 100 either, because... high retail price,
both for components and for production. plat.
Who will dare to break such a vicious circle?
I don’t see much point in using the eZ80; if possible, use an automated workstation.
(Unless the z80 has fewer legs)
But this is a matter for everyone personally.
Inexpensive workstations:
S3C4510B (50MHz)
S3C3410X (40MHz)
have ARM7 core
price 10-13$
Hello Valen
Val> I don’t see much point in using eZ80
If the technical process is at the stage of board tracing, then it makes sense to discuss
or not - it's useless. Such issues are discussed at the concept development stage.
You can vote for the generation of a new section “Concepts and Concepts” here:
http://zx.pk.ru/showthread.php?p=56338#post56338
Hello, captain cobalt
Val> You can't do it cheaper than 100 either, because... high retail price
Val> both for components and for production. board.
Val> Who will dare to break such a vicious circle?
Among others, the most interesting way is
replace hard with software.
Each copy of the hard drive requires you to pay money for it. Compared to him,
The software is copied virtually free of charge (not taking into account the cost of its development).
This method is already in use. Examples:
General Sound - sound generation by a universal processor.
Multicolor - display of images via video memory with a capacity of 6912 bytes,
occupying more than 6912 bytes in closed form.
The close connection between soft and hard in such a solution is the main problem.
The iron workers say “who will write the software?” Programmers say: "Who will
soldering pieces of iron?" And they continue to do what is more expensive, but can be used here
and now.
Hello, captain cobalt
R4008 = 10-12$
EP1C3 = 15-17$
SRAM 4pcs 512x8 = 12-16$
+small things
In total, the basic price for the set is about $60, the average price is inflated, and it can be cheaper.
Hello Valen
Val> S3C4510B (50MHz)
Val> S3C3410X (40MHz)
Samsung has already cheated the AWS community with the sudden cessation of production of some
crystals :) so fuck it. and inexpensive ones nowadays are more likely to be Philips
Hello Ronin
So I thought, why doesn’t anyone offer Philips? it seems there are also 7 armies,
cheap..
although I don't really support the idea...
Hello, psb
They have a poor selection of controllers with an external bus :( and in principle they are tastier
Hello, captain cobalt
cap> 6. Z80 is better emulated in ARM than in FPGA.
Why?
cap> 7. FPGA, despite its capabilities, contradicts the general line
cap> (item 0). They have a closed architecture, and the toolkit
cap> development will not be available for the target platform.
What is meant by FPGA architecture and why is “it” so important? About
development tools: ISE or Quartus are more accessible than Keil or something similar
(I’m silent about GNU ARM, because GNU).
Hello mungo
mun> calculating the instruction execution time becomes difficult
The main idea in the presented concepts was the operation of the Z80 software model in
ARM with a peripheral software model in an FPGA or a Z80 software model in an FPGA
with a software model of peripherals in eZ80+FPGA. Program models without generation
real tires, as I understand it, were chosen precisely because of the difficulty of complying with
makeshift and to simplify coordination. Do you think that in the case of ARM, there will be
fundamental difficulties with harmonizing software models? After all, in such
option, the only strictly time-bound signals will be external
signals that can be synchronized in hardware.
Hello icebear
An interesting topic was raised here... :v2_clapp;
IN; At the moment my work is connected precisely with automated workplaces. Therefore some
there are thoughts too.
Firstly, the workstation is actually completely 32-bit, and byte addressing or
words (as in the z80) do not work quite simply. :v2_cry;
2.; The workstation is not suitable for self-production in principle. Case for 100
legs (this is the minimum required) with a pitch of 0.5 mm is in principle impossible to solder (according to
due to inexperience, we tried it right away). Of course there are other cases designed for
inserts into the block. Only 208 pad legs with the same pitch is completely unrealistic
solder. I won’t say anything about what it’s like to separate them...
3. Ready-made boards for development on an automated workstation cost from 80 bucks (nonsense). Normal
generally from 350.
4. The RISC command system is too complex. Firstly, there are few teams and they are needed
somehow forming the most complex team of traditional professionals. Secondly,
the syntax is quite complicated. Of course C rules here, but we can talk about optimization
successfully forget. Only what is built into a specific compiler.
5. Even some 7-series workstations have built-in tools to speed up execution
commands - as a result, calculating the instruction execution time becomes
a little complicated.
These are such sad thoughts... :v2_cry;
That's why; The conclusion is that no one will replace the z80. There are already plenty of emulators, but
There are no completely correct ones at all. And it’s not possible to do it in the army.The only thing you can try to use it for is to replace
all that scattering of buildings that stands everywhere right now...
Hello, Black_Cat
I completely agree with Comrade. mungo, because my work at the moment is related to
Workstation, or rather now it is LPC221x (Philips). Its internal bus can reach
60-80 MHz, and taking into account the fact that they want Z80 20 MHz, 40 MHz, etc., exactly
It will not be possible to emulate a Z80 at such speeds. Even if you take another workstation, then
it is unlikely to be an order of magnitude faster than the desired Z80 40 MHz, for an acceptable
price.
Here someone suggested that, supposedly, you can tear it out from the source code of the Unreal emulator
the part responsible for emulating Z80 commands - yes, of course it’s possible, you just have to
imagine that, despite the simplicity of the C lines, after compilation it
turns into dozens of machine commands per emulated instruction. Yes, yes,
this includes retrieving instruction code from memory and maintaining internal accounting
(counting clock cycles, cycles, etc.), and direct execution - and also
I would also like to have external hardware synchronization with the regular one
Spectrum equipment, i.e. The automated workplace must respond to external signals and itself
expose them to the tire. And these are still a couple of teams, and, most likely,
dozens of teams.
And let many automated workplace commands be executed in one clock cycle. As a result, it doesn't matter
This turns out to be several dozen clock cycles per Z80 command, which, I would like,
so that it is as close as possible to the original and has the ability to expand in
performance side.When calculating (albeit with some margin), it turns out that we need both
at least half a gigahertz percent.
By the way, what about video memory and memory in general? LPC221x, for example, cannot
work with RU5, give it SDRAM. This means either manually and working with memory
do (set the address, wait for the data), or..? What are you... it's better
use SDRAM, one thing per megabyte. But then working with
video memory will have to be assigned to the workstation. Unamplified video output too
AWP will have to do. And it will also have to handle access to ports
- not to all, but only to some - which have to do with memory. What if
Does the port relate to more than just memory? Beta Interface, for example (switching
ROM and port spaces)... what to do? How to separate external and internal?
;)
By the way, about the Spectrum ROM... oh! Maybe it's better to do it after all
really manually work with RU5 and ROM, and then you won’t have to use ports
divide...eh? But then - all signals must be like the original, set
this, that, wait there, wait here (memory delays or suddenly someone -BUSREQ
will it sag or press Reset?).
It turns out that synchronization with the electrical circuit can be more difficult than
all taken together additions to processor emulation in a regular emulator (output
sound, on the screen, working with disks, user interface), for example, under
PC.What am I talking about? It takes very little time to process each action. But
there are so many of these actions - and it’s not a fact that there are fewer than in a regular emulator,
- that willy-nilly you will think: what for?
You can probably make a 3.5, 7 MHz Z80 on a workstation, but then you
you'll hit the ceiling, and it's not a fact that everyone can afford one
crap :)
What do we want? We want expansion, if not in architecture, then in speed. So this is
It's unlikely to work.
Hello, ARTi
int mov_h_a(cpu){
cpu.h=cpu.a;
cpu.pc++;
cpu.t=cpu.t+4;
}
Easily in 4 ARM commands if written in assembler. And they talked about emulation in
context (emulate Z80 3.5MHz on ARM 56MHz) in the shortest Z80 command
gets 64 ARM commands
Hello, Black_Cat
Bla> If there are no more questions regarding ideology, then he can decide
Bla> specifically with the processor and FPGA?
Bla> - capabilities
Bla> - price/availability
I fell down again...
Hello mungo
mun> hero, and you add clock counting and screen updating, and you will get no
mun> one dozen commands...
What screen update????
Hello mungo
mun> Yeah, yeah...
mun> Have you tried running the simplest emulsion at 486-66 MHz? Doesn't fly
mun> really? But the emulation is incomplete. :v2_devil;
How much; Is it correct to compare 486 and workstation?
Hello mungo
mun> And plus they forgot again that you can never assemble THIS at home with a soldering iron...
“Oh, these storytellers” (c) PPP :)
Hello mungo
mun> What is this?
mun> And what - from memory the picture immediately miraculously appears on the screen
mun> hits? Who will process it, if, of course, we talk about complete
mun> emulation?
So, is it the Z80’s job to form a picture on the TV screen? Either I didn't understand that
exactly what you meant. By the way, it seems like we are talking about Z80 emulation on a workstation, but
not all Spectrum.
mun> And about storytellers...
mun> You're laughing in vain. At the dawn of the development of automated workstations, they tried to solder manually... Not
mun> it’s working out really well!
There are difficulties, but they are surmountable. Of course you need to have some skills, and how
It’s already been said that on a makeshift board you’ll cover up the tracks faster, but what?
I would categorically deny that you were in a hurry.
Hello hero
If there are no more questions regarding ideology, then he can decide specifically on the percentage
and FPGA?
- opportunities
- price/availability
Hello hero
her> To work, we write procedures for each command
her> We make a table in which the instruction code is an index for the selection
her> addresses of the procedure when fetching, we count the contents of R, the contents of PC
her> is corrected by the command emulation procedure (since the command can
her> be parameters). The only thing worth keeping track of when sampling
her> are interrupts. As a result, pure processor emulation does not represent
her> of complexity.
Yeah, yeah...
Have you tried running the simplest emulsion at 486-66 MHz? It doesn't fly, does it? A
emulation is incomplete. :v2_devil:
Hello hero
Oh, they made me laugh... :)
Let's take a look at PC emulation on large machines. There are a lot of emulsions, but none
one does not work 100 pounds! There's always something that works great in real life
and nothing or with errors on the virtual... :(
Therefore, there is no point in talking about emulation at all until the emulsion appears, on
in which everything goes absolutely well and which even holds the proportions of music, for example
right...
For example, I have not yet seen a single 8910 emulator that would be absolutely
reproduced all sounds correctly. In my real old man, I don’t hear quite
something that even the most advanced emuls show. :v2_cry;
hero; and you add clock counting and screen updating, and you’ll get more than one
ten teams...
Hello hero
Then the next question is: who will make the boards? who will develop the hardware?
:confused;
I; programmer, so I can make software, incl. emulation and BIOS.
Hello icebear
How is this what?
And what - from memory the picture immediately miraculously appears on the screen?
Who will process it, if of course we are talking about full emulation?
And about storytellers...
You shouldn't be laughing. At the dawn of mastering automated workstations, they tried to solder by hand... Not really
it works!
Hello icebear
Oh, gentlemen, we're tired! :mad;
What; did you attack me? :mad;
I; let’s say I don’t see much point in making a new device if it doesn’t
really replace the old one. This is my opinion.
If we're going to do something, it should be fully functional with a minimum of buildings. Why
attach an automated workplace to the standard scattering of traditional Speckie cases?
:confused;
I; and for recreating specs on new modern hardware. Damn even you
I need to draw the software for the armament itself! :v2_clapp;
But; I just wish the gentlemen of the iron industry would think about how
the design will be repeatable. I don't think everyone has micro soldering irons or
soldering stations for 300 bucks. I have no. :sleep;
That's why; I suggest cramming as much as possible into a small amount
buildings. Let's say there are two armaments - one is the core, the other is the hardware ports, etc. :v2_cheer;
I repeat.; This is just my opinion and there is no need to attack me with your fists.
Hello mungo
And plus, they forgot again that you could never assemble THIS kind of thing at home with a soldering iron... Well,
industrial iron will cost more...
So it’s a simple matter of buying a PDA and pouring emulsion on it!
Hello hero
Did you understand what you wrote? Did you read carefully what I was talking about? Where do you have
fetching a command from memory? Where is its decryption? Where is the update of register R and it
similar?
There is none of this, so it won’t work :)
P.S. Or is this some kind of cheating of posts? Then I don't play =)
Hello hero
To work, we write procedures for each command
We make a table in which the instruction code is an index for selecting an address
sampling procedures consider the contents of R, the contents of PC are corrected
command emulation procedure (since a command can have parameters).
The only thing worth monitoring during sampling is interruptions. In the end
pure processor emulation is not difficult.
Hello icebear
But there is no shortage of naivety.
Hello icebear
This is an emulation of the LD H,A command
And before shouting, I recommend that you read the source code US
Hello mungo
In fact, there is a lot of work there :)
Hello mungo
mun> And plus they forgot again that you can never assemble THIS at home with a soldering iron...
mun> Well, industrial iron will cost a lot...
mun> So it will be easy to buy a PDA and pour emulsion on it!
Although it’s not entirely easy to put 100 paws in makeshift conditions, it’s also possible if
I have experience holding a poling iron, it’s not even easy, the only thing is that it’s a fee
should be normal
Hello mungo
mun> Have you tried running the simplest emulsion at 486-66 MHz? Doesn't fly
mun> really? But the emulation is incomplete.
We are talking only about the processor, and the screen, etc., is implemented in hardware, including
including different memory models. Otherwise, two different extremes are compared and
expect some intelligible results from them. Video MPG4 on Windows too
It runs fine on any 600++ hemp thread, but from under the board it can easily run on a Celeron
300A
Hello mungo
We are talking about a spec whose native processor will be ARM and so that it can be used
to call it a spec, it should be able to run everything that is made for the spec, and as
the emergence of software for the new percentage, it was possible to gradually switch to it. I'm about
that Z80 emulation would be a functional part of the BIOS (or OS)
Hello, captain cobalt
Regarding emulation of z80 in pls...
Yes, to get the fastest z80 it is quite possible,
but here seems to be a different approach...
That is, I already wrote this, it seems, a 21st century speck -
constructor arm+plis, similar to z80+ULA.
And emulation of the 20th century spec is for preservation
traditions... In that vein.
Actually a question. Who has already done/is going to do what?
how they chose, etc. Otherwise your head will swell :)
Older models look beautiful...
Atmel seems like a dma at least has enough...
And if we take into account the development time, say a year, then the younger
may not exist anymore :)
Hello ASDT
cap> this must be done until the moment when such software (for fast
cap> Z80) will grow and become an obstacle to such a step
Each new clone with an expanded non-standard mode for ZX claims to
that further development will follow exactly this non-standard path. The first was
Sprinter, then Turbo-XX, now Caro is making a car with a non-ZX mode on
eZ80-50MHz!!, well, then maybe there won’t be a ZX mode on ARM (if everything doesn’t end
just talking). The sprinter has already withered away, for Turbo-2+ it is now intensively
promote writing software for its advanced modes, adapt software for
eZ80-50MHz is simpler than for ARM, which means its software content will grow
faster. So what next? Splitting the platform into a bunch of small branches? So it will be
the same as with the Sprinter; there is no one to support it, and who will support it
each budding branch? Maybe it’s worth thinking ahead and deciding such
conceptual questions for this section: for what tasks, who will
user, what is the concept of development of a specific branch and the entire platform in
in general? Or did the experience of previous developments teach you nothing?
Hello, Black_Cat
YES, at least 30 varieties at the moment are all doomed to exist in piecemeal form
quantities, the most important thing is not what comes out, but the creative process
Hello, Black_Cat
The fact of the matter is that an ARM-based computer will not be a development of the Spectrum.
You obviously skipped the comparison of HDD and FDD speeds...
Put a modern propeller at 7200 rpm, you will understand what I mean ;) Other than that
equal, a modern propeller’s response speed to commands will be almost
instant.
And to make a “client-server”, you need an OS... And there are so few programmers left
:(
Hello jdigreze
jdi> an ARM-based computer will not be a development of the Spectrum
So, the advanced modes in Turbo-2+ and Sprinter are not a development of the Spectrum, and
in the Caro car in eZ80 mode - these are all separate cars in one bottle with the ZX.
In this series, ARM looks the best as the basis of a non-ZX computer.
jdi> You clearly skipped the comparison of HDD and FDD speed...
It was somewhere on the forum - they timed the time of downloading a file - so the speed
the difference was not very different, I don’t remember exactly, but 2 times (since it is not the HDD that is determined, but
Z80), but the access time is different, it will be shorter for the screw
significantly than the flop - the speed of rotation and positioning are not comparable.
Regarding the axis/non-axis - options are possible, up to a primitive file
manager in ROM that launches the selected program to load from the PC via TAPE input
(for machines without a RAM disk for example), or make a hardware TR-DOS emulator
controller, for example, on the same 51 single-chip as the PC emulates the keyboard, but
working instead of FDD, via COM/USB/Ethernet (which is easier, you can use
fast) from a PC - almost all TR-DOS capabilities will remain, but in terms of speed
with USB/Ethernet there won’t even be a difference. Or the same thing but with a flash card - but
this is worse, because in this case, it will not be the PC that will work with FAT, but the single chip.We get - in the first case the OS is not needed, in the second - it is already there, etc. etc.
Hello, captain cobalt
cap> This is far from the same as decrypting ports and emulating
cap> device hanging on them.
cap> Through the same place dynamic can potentially be performed
cap> memory allocation.
Well, my question was specifically about intercepting calls to familiar ports
available devices. Well, hero answered quite comprehensively.
cap> The support of old spektrumists is more important. We need to smooth out the transitions
cap> Z80 <-> ARM in both directions.
That is Are you going to transfer old Spectrum players from Z80 to ARM over time?
Then it definitely won’t be a Spectrum. And will the old spektrumists go to
this?
Hello hero
her> And who will finish the war completely, I am almost 100% that the majority
her> software that uses undocumented processor capabilities does not work
her> will be
Do you think these problems will not arise with an emulator based on an AWS? Need to bring it up
it will be both here and there, another thing is that the crust is less alien than the automated workplace.
Hello hero
her> And after a while when the pentagons are finally scorpions, etc.
her> will fall apart, why should I use emulators??
On the crusts. He sits down immediately. They (the crusts) will never fall apart. And no one
no need to retrain.
Hello caro
car> And note that all these modes require working with
car> Z80 command system native to the Spectrum
The Z80 command system is a real asset, but only in ZX modes. In nonZX
mode, the main goal should be, as it seems to me, to use the opportunities
another software platform (OS, software), which is simply not a platform for the Z80.
This is the only point of such a Siamese twin on ARM.
Hello caro
This is not exactly closer to the spirit of the Amiga, the ZX emulator is the same as the MS-DOS emulator
for Windows (that is, an integral part of the operating system)
Hello caro
What is there to describe, the computer turns on and boots from its Beavis and when you try
launching FDI or TRD images starts the spec emulator which is part of
firmware and if you connect it to the 3.5 MHz interrupt, then the emulator
Can easily work with any thread with a multi-tasking axis (the same tnkernel)
Hello caro
x86 sockets are difficult to solder :)
But ARM has such an option for further development ARM7 60 MIPS -> ARM9
200MIPS. A eZ80 averages about 40 8-bit MIPS, while ARM7 has 60
32-bit MIPS, that is, in reality ARM will be 2 times faster than eZ80. Next, eZ80 50MHZ
can send a maximum of 25 meters per second, and if you take some ARM7 nit with 16
EBI and put it on the same frequency, it can already make 50 transfers.
Hello icebear
And who will finish the crust to mind, I am almost 100% that the majority of software
using undocumented processor capabilities will not work
Hello icebear
And after a while, when the Pentagons are finally Scorpios, etc.
will fall apart, what should I use on emulators??
Hello, Black_Cat
Bla> ...In this series, ARM as the basis of a non-ZX computer looks better
Bla> all....
I didn’t want to get involved, but let me disagree:
I. If we are talking about a realistically possible development,
then this is most likely ARM from PHILIPS from the series
LPC21xx, or something equivalent from others
manufacturers. And this is no more than 60 MIPS for
RISC commands. I doubt it's cooler than
50 MIPS short CISC commands of the eZ80 processor.
II. A computer based on eZ80+FPGA has the ability
work in 4 modes:
1 - ZX Spectrum mode
(Z80 with a frequency of 3.5 MHz is implemented in FPGA,
eZ80 with FPGA emulates ZX Spectrum peripherals)
2 - Turbo ZX Spectrum mode
(the same thing, on the Z80 it works at maximum
possible frequency - judging by the CPC emulator on
TREX C1 board, this is about 20-25 MHz).
3 - eZX Spectrum mode, with control ability
eZ80 processor speed up to 50 MHz.
(FPGA implements all computer peripherals,
modified ZX BASIC firmware is used
and TR DOS, which takes into account changes in peripheral addresses.
Programs that use standard entry points
BASIC and TRDOS work as if nothing had happened.
The performance of BASIC increases by 2 orders of magnitude.)
4 - eZ80 based computer with advanced
possibilities.
(Here everything will depend on the possibilities and ideasdeveloper and available hardware resources).
And note that all these modes involve working with
Spectrum's native Z80 command system :)
I’m thinking about possible modes of operation with ARM processors
adherents of the SUBJ idea will be able to describe it :)
Hello hero
her> This is not at all closer to the spirit of the Amiga, the ZX emulator is the same as an emulator
her> MS-DOS for Windows (that is, an integral part of the operating system)
What then is the meaning of the SUBJ phrase?
It would be more correct to say then: “x86 is the Speccy processor of the 21st century.”
Hello hero
her> Why describe it? The computer turns on and boots from its Beavis and
her> when trying to launch FDI or TRD images, the spec emulator starts
her> which is part of the firmware and if you hang it on
her> 3.5 MHz interrupt then the emulator can easily work with which thread
her> multi-task axis (same tnkernel)
This is a description of how an IBM PC works with the ZX Spectrum emulator :)
Hello hero
her> But ARM has this option for further development ARM7 60 MIPS ->
her> ARM9 200MIPS.
It's practically unrealistic.
her> A eZ80 has an average of about 40 8-bit MIPS, at the same time for ARM7
her> 60 32-bit MIPS, that is, in reality ARM will be 2 times faster than eZ80.
This is if the data bus is 32-bit.
I have a suspicion that real development can count on a maximum of
16-bit bus.
So you will lose this coefficient 2 :)
And in general, where did you see ARM programmers?
Those who appeared and tried to express their point of view were quickly
placed against the wall :).
It seems like I shouldn't have gotten into this again..... :)
It seemed to me that we were trying to talk about the Spectrum, and not about a new platform
on ARM.
Hello hero
her> x86 sockets are difficult to solder :)
Alternatively you can use:
Microcontroller R8610 from RDC Semiconductor.
The microcontroller is built on a 32-bit RISC architecture and has
Compatible with x-86 platforms Windows, Linux, and other 32-bit RTOS. From
built-in peripherals, the microcontroller has a first-level cache of 16 KB in size,
32-
bit interface PCI rev. 2.1, operating at 33 MHz, SDRAM/ROM
controller
memory, Fast Ethernet 10/100 controller and USB 2.0 Host.
Structurally, the microcontroller
implemented in a 216-pin LQFP package.
The cost is about $15.
For more information, see here:
http://www.prochip.ru/support/rdc_r8610/
File: R8610.zip http://zx.pk.ru/attachment.php?attachmentid=3899
Hello, captain cobalt
"ARM7 60 MIPS -> ARM9 200MIPS."
"It's almost unreal." - unproven...
"It would be more accurate to say: 'x86 is the Speccy processor of the 21st century.'"
More likely no than yes... this is the direction for “big” computers,
even if they are small :)
"It seemed to me that we were trying to talk about the Spectrum, and not about a new platform
on ARM." Uh...
In theory, it may well be suitable... But will it take place as a spec. 21st century...
That's the question.
And one more thought... :)
Spec. 21st century should allow video and audio processing
Hello, captain cobalt
"Buy any Philips KIT, it will be much cheaper and more effective."
We are not looking for easy ways :) And atmel is not bad either... And others...
Hello, captain cobalt
Yes, I forgot... Studying and experimenting with armature is also a good topic,
on its own... :)
Hello, Black_Cat
Bla> The optimality of Z80 emulation on ARM or FPGA is approximately the same
Bla> - this has already been discussed, except that the FPGA is a little faster for the same money
Bla> percent can be done.
Did you pull this out of thin air? _Roughly speaking_, when working at an automated workstation you need to know C,
When working with FPGAs, you need to know something else besides C. By "work" we mean
writing Z80 emulation code.
Bla> They work on PCs in emulators and don’t complain about x86. Work in the system
Bla> RISC commands will only be needed in nonZX mode, and this is for the Spectrum
Bla> has no relation and will not affect the ZX scene in any way.
How is this possible? Let me guess: either “some kind” will work in ZX mode
Z80, or in nonZX mode does the workstation suddenly cease to be a RISC processor?
Hello, Black_Cat
Bla> Creating an emulator in ARM/FPGA is a one-time procedure
Bla> attitude to the creation of the clone itself, to the ZX software that is on it
Bla> ZX mode will be executed, this has nothing to do with it.
It seems to me that we were both talking about abstract topics :) What was meant by
"Is the optimality of Z80 emulation on ARM or FPGA or approximately the same"? Speech
Are you talking about writing an emulator for an automated computer or writing a crust on an FPGA?
Bla> I clarified the phrase in order to eliminate any misunderstandings.
It's still unclear. You're starting from full emulation of the Z80 on the workstation, right? If so,
then the Z80 is physically absent. And if it is absent and everything works on the workstation, then
where will its (ARM) RISC commands go in any of the operating modes of the entire system?
Hello hero
her> Let me remind you that a Z80 emulator already exists, and if not the best
her> is definitely one of the best, and also well tested and
her> labor costs are minimal.
Are we talking about an automated workplace or a crust?
Hello icebear
ice> We are talking about writing an emulator for an automated computer or writing a crust for
ice> FPGA?
About both. It was meant that the complexity of both spellings is practically
is the same given the complexity of creating a clone as a whole. Those. their share in total
labor costs for creating clones are almost equal.
ice> where will his (ARM) RISC teams go
The teams are not going anywhere, it was meant that ZX programmers are in ZX mode, c
RISC teams will not deal, they will still work in the system
Z80 commands, which will be executed by the Z80 emulator.
Hello caro
Well, at least 32bit OS *NIX and software for it are better and more accessible than it
8bit clone meant.
Hello caro
car> optimality in terms of Spectrum emulation.
The optimality of Z80 emulation on ARM or FPGA is approximately the same - this is already
It was discussed, except that in FPGA for the same money you can do it a little faster.
car> do they want to have a system with two
car> with completely different processors.
They work on PCs in emulators and do not complain about x86. A programmer can work with
the RISC command system will only be needed in nonZX mode, and this is for the Spectrum
has no relation and will not affect the ZX scene in any way.
car> it is the non-ZX mode that will allow implementation on such a device
car> a full-fledged system with its own Axis, be it CP/M or UZIX
I completely agree. The same is the case with ARM, but only in theory - better,
simpler and more accessible.
Hello caro
Something tells me that 32bit *NIX and 8bit UZIX are two big differences
:)
Hello icebear
ice> when working on an automated workstation you need to know C, when working with FPGAs you need in addition to C
ice> something else to know
Creating an emulator in ARM/FPGA is a one-time procedure related to
creating the clone itself, to the ZX software, which will be executed on it in ZX mode
it has nothing to do with it.
ice> How is this possible?
I clarified the phrase to eliminate any confusion.
Hello, Black_Cat
Let me remind you that a Z80 emulator already exists, and if not the best, then
definitely one of the best, also well tested and labor intensive
minimal.
Hello, icebear
Об US :)
Hello ASDT
ASD> "ARM7 60 MIPS -> ARM9 200MIPS."
ASD> "It's practically impossible." - unproven...
I think the one who developed microprocessor technology
with clock frequencies greater than 100 MHz, I understand.
Let me remind you that we are not talking about industrial development.
Hello ASDT
ASD> Yes, I forgot... Studying and experimenting with armed forces is also good
ASD> topic,
ASD> by itself... :)
Buy any Philips KIT, it will be much cheaper and more effective.
http://www.mt-system.ru/price.tree.mt?id=20460
Hello, Black_Cat
Bla> ... Only in this application there is a sense of such a Siamese twin on
Bla> ARM.
Again 25.
You should at least ask Spectrum programmers if they want to have a system with two
completely different processors.
And as has already been noted here
http://zx.pk.ru/showpost.php?p=60399&postcount=209,
you simply ignore the technical issues raised
implementation of such a system and its optimality in terms of Spectrum emulation.
And regarding the axis - it is the non-ZX mode that will allow implementation on such a device
a full-fledged system with its own Axis, be it CP/M or UZIX.
Hello, Black_Cat
Bla> Well, at least 32bit OS *NIX and software for it are better
Bla> more accessible than its 8bit clone was meant.
All this is already in C, so the difference in platform does not matter.
Hello, Black_Cat
Bla> The same is the case with ARM, but only in theory - better, simpler and
Bla> is more accessible.
As ASDT says in such cases - unproven :)
Hello, captain cobalt
I propose to discuss false logic based on false data...
Hello caro
car> 25 again.
car> You should at least ask Spectrum programmers if they want to have
car> system with two
car> with completely different processors.
I agree that there is no need to repeat the mistakes of the Amiga with its transition to PPC from m68k.
The fact is that the Oska has developed, and the computer itself (like the Mac in its time)
became PC-like, losing its zest. It makes sense to save the work
specovskie in terms of software. At the same time, people cannot be prohibited from using
computer only as spec. This is also a misconception. So it turns out that the stick is about
two ends.
IMHO, the ability to directly use arm should be left. R8610 is no longer available
ideological reasons ;)
Hello, captain cobalt
> And according to the laws of logic (there is such a whole science), if the basis
> a false premise (initial data) is placed in the logical construction, then
> regardless of the logical analysis system itself, the output is we
> we get a false result.
According to the laws of logic, anything can follow from a lie. ;)
Hello, Black_Cat
I will speak only for Turbo 2+:
Bla> Why are advanced modes being introduced at all?
Bla>
Bla> Analysis of extended modes for compliance with membership
Bla> ZX platform.
Bla> If we consider the existing, under construction and only discussed clones
Bla> from the point of view of compliance of their extended modes with
Bla> ZX platform, we get the following:
Bla> 1) In Turbo-2+, when working in CP/M, ZX is not used at all
Bla> architecture.
CP/M is CP/M. The regime has nothing to do with it. And in general, what is meant by
"advanced mode"? New graphics? New architecture or both? I will
imply the latter. If so, what does CP/M have to do with it? Yes, this operating system
requires disabling the ROM, that's it. But the concepts themselves are incompatible: one thing
features of the computer's technical data, and another - the OS that uses them.
In addition, in Turbo 2+ there is no division into ZX-non-ZX architecture. There is architecture
flexible and programmable, including memory page layout, screen type
vacuum, keyboard operating mode and masking mode for a particular type of port.
T.N. ZX mode is a special case of architecture programming. Therefore, it is possiblerunning the program in CP/M, enable the ZX architecture (or part of it), or in TR-DOS
Having launched the program, continue to treat the architecture and all modes as you like
whatever.
> 1) In Turbo-2+, the main meaning of its extended modes is work in iS-DOS
> and CP/M incl. using HDD and CD;
The assessment is completely wrong. As I said, there is no separate non-ZX in T2+
mode. It’s just that the capabilities themselves on the computer are so expanded and so flexible,
what does the ZX architecture include as a special case (by the way, even on a regular
ZX can change the architecture to a limited extent - through port #7FFD, of course. Try it
turn on page 5 at the top, and then launch the 48K software in this form. And when he
is more likely to freeze during loading, think again
why do you need an extra mode?). Further, the meaning of new opportunities cannot be reduced to
using HDD/CD, since they are not needed at all for this (except for the
IDE ports, of course). You can access these devices directly from the architecture
BASIC 48/128 (on ASMA, of course).
And the point of advanced modes is to use new features
(screw, graphics/palette) and in more flexible work with memory. For example, you need
move a block of codes (well, at the maximum - 16K) from page 4 to page 7. What
will you do it? (let’s assume that the program that executes all this is locatedsomewhere thread at #6000)
On the "standard ZX":
1) Turn on page 4 via OUT at address #C000.
2) LDIR (or some other method, depending on need) 16 KB to the bottom
memory. For example at #8000.
3) Turn on page 7 via OUT at address #C000.
4) LDIR (or some other method, depending on need) 16 KB back to
#C000.
5) If necessary, restore the previous page via OUT.
On Turbo 2+:
1) Turn on page 4 via OUT at address #C000.
2) Turn on page 7 via OUT at address #8000 (if necessary, you can use both #4000 and
#0000).
3) LDIR (or some other method, depending on the need) 16 KB from #C000 to
#8000
4) If necessary, restore the previous memory configuration using OUT.
Just like that - cheap and cheerful. And most importantly, calculate the time savings.
Of course, if the goal is full compatibility, you will have to use
old methods. But if the program is written specifically for ATM (the same OS
TASiS, for example) and it uses its specific modes (yes, at least
advanced graphics), then why not take advantage of the new features,
flexibly turning them on and off as you go?
> the goal is the convenience of disk operations on a fast computer (on Turbo-2+
> for example, HDD is not much faster than FDD) and in addition to this
> some Who told you such nonsense about HDD? The screw works like on others
computers, an order of magnitude faster than the flop (there are no serious WAITs on the IDE controller
exhibits) and is only slightly inferior to the RAM disk. And judge for yourself, it would have happened then
perhaps real-time video from a CD (not a screw, but since you’re talking about slowness,
this means you are not blaming the device, but the controller) with data of 15 frames per
second (EGA high vacuum screen) and sound via 8-bit COVOX with frequency
17.5KHz (to be fair, I’ll say that all this is in turbo mode).
In OS TASiS, a 640Kb TRD image is uploaded from a screw to memory in 3.5 seconds. THIS
SLOW? In OS CP/M, however, it is really much slower - in 30-40
seconds But what does the screw have to do with it? He reads information no slower there either.
It’s just that the OS itself, with all its archaic advantages, is a drag: considering
at normal speed one block, she will spend “half an hour” looking for another, etc.. Well,
how she rummages through a directory in search of a file if the directory is large.... Without
comments. In this case, I repeat, reading the block/sector itself takes place with
normal speed for propellers. Moreover, one cannot say that the screw does not work
faster than FDD. Because you should have seen how fast it works in CP/M
actually FDD! You can't even compare it to the flop in TR-DOS/iS-DOS/TASiS. Although,
Of course, it's still faster than tape. And this makes me happy.In short, there is no need to blame the screw/CD in vain. IDE controller on Turbo 2+ as hardware
the device works no worse than other IDE controllers (and in some places, maybe even
better. For example, in comparison with SMUC, where each time a byte is transferred from/to
the screw must go into TR-DOS). And it’s all about the software and crooked hands - like software
screwed it up, so it will work quickly. And to blame iron on this basis,
it’s like blaming the feet if the dancer himself is bad.
> the possibility of modern network communications (albeit in terms of performance
> you can already use a UNIX clone for the Z80, which is in all respects
> should be better than CP/M, even in the version with such cut-down
> capabilities);
That's what you're all about CP/M (I say, again, only within the framework
discussions about Turbo 2+)? CP/M is old and is only supported solely due to
good software and toys written for it (the same “Prince”). And in T2+ in
iS-DOS clone - OS TASiS is used as the main one, just widely
using all the flexibility of the computer architecture, and not being offended by speed.
> Conclusions:
> 1) From consideration of the main purpose of the extended modes it is clear
> that there are attempts to eliminate the existing ZX platforms> disadvantages by combining two hardware platforms. At the same time
> borrowing of missing capabilities due to introduction is used
> additional hardware architecture and use on it
> software of other software platforms.
The conclusion is fundamentally incorrect in the particular case of Turbo 2+ that I am considering. No
no crossing of two different platforms. And there is only an increase in capabilities
and modes that the programmer himself is free to either use or disable and
be content with the “Spartan” environment “128k+TR-DOS+AY”. Moreover, in one and
the same program in its various sections.
> 2) That Siamese twin that is in with ZX is not
> is a development of the ZX platform, because does not belong to it).
The conclusion is not just incorrect, but the introduced term “Siamese” itself is initially false
twin." The new modes in Turbo 2+ are exactly the same “Siamese”
twins" as usual COVOX in relation to the standard architecture, For
nothing will change for the programmer. Only a new port will be added (in cases of problems with
We will deliberately omit port decryption. This is a question of crooked developments, not
the very principle of adding capabilities), using which we will get newcapabilities in extracting sound, and if we don’t use it, we won’t lose anything.
In other words, if we figuratively imagine the principle of increasing capabilities in
T2+ compared to the ZX-128, then this can be done by comparing a tenth grader with
himself in the past, when he was, say, a third grader.
In the third grade he could perform 4 operations of arithmetic and write statements with
dictations of the difficulty level “mom washed the frame” (exaggerating). And when he was in 10th grade
everything can be the same (only an order of magnitude faster and better), but in
In addition, he also knows trigonometry, and can tell you the principle of construction in a heartbeat.
periodic tables, won the regional literary Olympiad, wrote an essay
on the occasion of the 300th anniversary of Pushkin, and even, having graduated from music school, he knows how to play the
piano and guitar.
And your idea of a “Siamese twin” is:
In the third grade he could (see above)... And in the 10th grade he had new opportunities
were added, but he can use it depending on which leg he got up from
(what mode I was in) or what I ate for breakfast:
In one case, he does almost everything he could do in the third
class, but only that.
In another case, he plays musical instruments very well and can draw a graph.
functions of a quadratic equation, read all of Pushkin’s poetry by heart, but he
The ability to write and count/multiply is completely cut off.
I'm dying...In general, to summarize, I will say that this analysis is not without logic, which is quite
slim. But its only drawback, and a deadly one, is
unreliability of information about the platform(s) - that is, the analysis is based on
unverified information that turns out to be false. And according to the laws of logic (there is such
whole science), if the logical construction is based on a false premise
(initial data), then, regardless of the logical analysis system itself, on
we will get a false result. Which is exactly what happened. So, your analysis, when
all his harmony is of no use, along with the conclusions. Throw them away
trash heap :v2_finge; :);
> 3) For the development of the ZX platform in most cases (i.e. for
> ordinary role user) at a minimum, the presence of an interface is sufficient
> data exchange and capabilities provided by client software in the system
> client-server with PC. In this case, the PC is used as a file server, and ZX
> as a diskless station.
But a standard interface would definitely not hurt the Spectrum (by the way, in T2+
There is a standard COM port. Just until you get around to supporting him
programmatically).
Hello, captain cobalt
cap> According to the laws of logic, anything can follow from a lie. ;)
Anything can follow from a lie if the logical chain is broken.
And human lies are, according to the law of logic, either conscious
the use of false premises (do not confuse the term “lie” with “falsity”) for the purpose of
bring someone in the end (with the correct logical chain) with a false one
conclusion. Or, again, a deliberate distortion of the logical chain, with the aim of even
with correct initial data, get a false conclusion, or with correct data
get a false conclusion (for example, for someone to take on a difficult task that
actually impossible, and spent a lot of time and effort on it).
Since there is no “criminal intent” here to fool one’s brains, I am not talking about lies, but
about the falsity of premises and conclusions. :)
Hello, captain cobalt
When it comes to conclusions and logical chains, we are talking about implication.
Implication is not the same as equivalence.
The implication is briefly described as follows: “from truth only truth can follow, and
Anything from a lie."
This means that you can take a false premise and construct a correct logical
chain, and get the true output.
Right?
Hello, The Exploited
Well, this is a theoretically possible move: smile:, like we just transmit/receive and
nothing more. And even without whites :smile:. And for the flop, so from the max that
he knows how to give on his own, he thought, obviously the controller was not mentioned
:smile; Those.; purely idealized case. But because idealization was used
for both devices, the ratio should be relatively real :smile;
Yours; 437/60=7.28
Hello, captain cobalt
To Maxagor:
You gave a somewhat chaotic answer to Chamberlain :smile; ,; but I'll try
It’s time to systematize it.
1) Turbo-2+ is mentioned here because-to the extent, only because
In general, the question of equipping it with a new CPU was discussed (I really don’t believe in this,
unless you convert the new computer into an FPGA). Because the clone already exists in the form
ready-made boards, and nothing can be changed in it, then the discussion is purely
statemental character.
2) Use of the term in the meaning ZX/nonZX in relation to Turbo-2+
is conditional and is explained in detail here:
http://zx.pk.ru/showpost.php?p=61049&postcount=113
3) By HDD vs FDD;
Max; transfer i/o Z80-7MHz non wait =1.75Mb/s => Max transfer HDD
Max transfer FDD =0.5/0.25 Mb/s (two modes).
=>Max transfer HDD/FDD=3.5/7 times (for 2 modes).
If you get a different ratio, then this does not mean that the HDD
Suddenly it started working faster, but FDD is working even slower for you. So
Thus, in relation to HDD & FDD there is no need to talk about performance at all,
rather about their lesser or greater inhibition. Regarding eZ80 - even without DMA
it can give ideally 50Mb/s.
Hello, Maksagor
> the phrase about the slowness of the propeller in the last post was constructed like this
> in a way that clearly indicated that the propeller is slowing down specifically on ATM
No, ATM was not meant, it was just the old one in consideration.
> Maybe you took the potential speed of the flop as a device in general
Yes, and the results were considered to be maximally possible in theory.
> For example, OS TASiS was created last year
It says -OS TASiS was a development, not created.
> with the chronology of the appearance of certain hardware and software mules
> you are so confused there
No, it’s not confusing, I checked your site, but the order in which everything is
is stated - determined not by chronology (there was no such task), but by logic
analysis.
Hello, Black_Cat
Bla> 3) By HDD vs FDD;
Bla>; Max transfer i/o Z80-7MHz non wait =1.75Mb/s => Max transfer HDD
Bla> Max transfer FDD =0.5/0.25 Mb/s (two modes).
good to make fun of. Z80 in turbo mode with hdd will take a maximum of 437kb/s
(7mhz/16), and from a flop, even with HD floppy disks, about 60kb/s (regular ones - 30kb/s).
Hello, Black_Cat
Bla> To Maxagor:
Bla> You gave a somewhat chaotic answer to Chamberlain :smile; ,; but
Bla> I’ll try to systematize it as much as possible.
Come on...
> 1) Turbo-2+ is mentioned here because-to the extent, only because
> the fact that the issue of equipping it with a new CPU was generally discussed (really
> I don’t believe in this, unless I convert the new computer to an FPGA). Because
> the clone already exists in the form of ready-made boards, and there is nothing in it anymore
> change, then the discussion is of a purely statemental nature.
Why? Turbo-3 will be made from scratch in the near future...
> 2) Use of the term in the meaning ZX/nonZX
> in relation to Turbo-2+ is conditional and is explained in detail
> here: http://zx.pk.ru/showpost.php?p=61049&postcount=113
Oh, I don’t have time to comment on what’s written there now - I’ll try tomorrow. But
Here I will briefly say that with the chronology of the appearance of certain
the hardware-software mullets you have there are so much mixed up that all the conclusions about
home|business-computer and about professional or non-professional decisions
Software developments float on shaky ground. For example, OS TASiS was created inlast year by Yuri Korsunin with my humble participation - so selfless
same ZX-enthusiasts like AlCo, CPU, Triumph and everyone present here. So how
the phrase “After it became clear that the niche business
it was not possible to get computers, further development of the computer was associated with
Spectrum OS iS-DOS and its development TASiS"? If anyone hoped
to use the subject for business are its inventors - the company MicroART,
which created the computer, wrote the first ROM, and introduced CP/M there. I just did
she did all this in 1991-1993. And I stopped using this computer altogether for about ten years now.
engage. So what does this have to do with business plans from 10-15 years ago? And was reborn
Turbo-2+ just two years ago, when Chunin Roman, the same ZX enthusiast as
everything here, at my own peril and risk, I revived the production of circuit boards (again, under my
modest participation). Could Roman in our times seriously think about
business application of the machine, and after failure, suddenly reorient to TASiS?
And how could this axis be developed “professionally” by one person? Any software
on ATM is written with exactly the same motivation as everything on Speccy: out of desire
create, make further work on the computer more convenient, show the limit
possibilities (especially since they are wider on ATM), etc.. Because Turbo 2+ is
and there is Spectrum. And the use of new modes makes it not a Spectrummore than 256KB memory usage on Scorpio compared to
Pentagon-128. If this is not so, then it appears that the progress of architecture
while maintaining top-down compatibility, it should be prohibited, and only added
switchable mules type GS. But I categorically disagree with this.
> 3) By HDD vs FDD;
>; Max transfer i/o Z80-7MHz non wait =1.75Mb/s => Max transfer HDD
> Max transfer FDD =0.5/0.25 Mb/s (two modes).
> =>Max transfer HDD/FDD=3.5/7 times (for 2 modes).
Stop, stop! It's like your disk drive reads 256-512Kb per second (even with
accounting for brakes due to crooked wood)? Something is wrong. Let's see how
It is known that with all the tricks in non-turbo mode, transfer the entire screen 6912 bytes
It will not be possible to transfer from one place to another in one interruption - there will not be enough clock cycles. Well,
Let's reduce, specially rounding (to make calculation easier) the maximum transfer
per frame - 5Kb. In total, you can transfer 5x50=250Kb in a second. Well, let's say
280Kb. In turbo - 560Kb or a little more than 0.5Mb. And at the same time, the RAM_disk also works on
order is faster than flop, and even faster than screw. And this takes into account the fact that when
When reading from a flop/screw, fast INI/INIR commands are used, as opposed to
RAM disk, where more “costly” LD/LDI/LDIR/PUSH/POP are used.Maybe you took the potential flop speed as a device in general, by
passport, outside the characteristics of the computer? Then it won't do. We must take into account
the real speed of the device specifically for a specific type of computer, and there
compare them to each other...
> If you get a different ratio, then this means the wrong thing
> that the HDD suddenly started working faster, and that FDD is still working for you
> is slower.
It works in TR-DOS exactly the same as for everyone else. TASiS is even a little faster (due to
driver specially written by Yura Korsunin).
> Thus, in relation to HDD & FDD, talk about performance in general
> not necessary, rather about their lesser or greater inhibition.
In fact, even the dubious FDD/HDD performance ratio of 1/7 is already
It’s worth talking about the screw as being significantly faster than the flop.
Well, I simply DO NOT BELIEVE about the 1/3.5 ratio. Measurement results with a stopwatch
in your hands to the studio! And one more thing - the phrase about the slowness of the propeller was in the last post
constructed in such a way that it clearly indicated that the propeller brakes precisely at
ATM. Please clarify again. And if this is so, then the justification for some special
slowness compared, for example, with the Nemo controller, again in the studio.
Hello, Black_Cat
Bla> So, even having written me down as an enemy of ATM: smile; -; his problems from
Bla> this will not be solved.
Wow, no one is classifying you as an enemy. I believe that in the description
The positioning of this clone today has been misjudged. Yes and in
I also don’t agree with the division (even if not hardware) into ZX/non-ZX. And about the problems,
which you are hinting at - if they exist, they are completely different. A
this is what at the beginning of the review I laid out the basic concepts on the shelves - very
right. Okay, I'll go. We'll leave the argument for next time.
Hello hero
People, is the process going? Is the cat doing something???
Hello, captain cobalt
"People, is the process going on?"
What? :)
Hello skyther
sky> Is the cat doing something???
Cats (and cats) do things pretty quickly. Slow down for people :)
Hello skyther
sky> People, is the process going? Is the cat doing something???
What do you think? Won 20 signed that this is the future, so cards in hand -
make a fairy tale come true! ;)
Hello skyther
sky> People, is the process going? Is the cat doing something???
What do you think? Won 20 signed that this is the future, so cards in hand -
make a fairy tale come true! ;-)
Hello, ewgeny7
I just got the impression that the subject was forgotten :(
Hello, Scorpion(lv)
Xilinx has released the most capacious FPGA type FPGAs
01/05/2007 [13:48], Alexander Budik
Xilinx has begun shipping new high-performance programmable
logic integrated circuits (FPGA) LX330 (330 thousand logical cells) FPGA type
(Field Programmable Gate Arrays). The company also announces the release of the LX220 (220
thousand logic cells), thus completing the market launch of the Virtex-5 series
LX. Since May 2006, the LX50, LX85 and LX110 have been commercially produced.
LX330 is manufactured according to 65 nm standards, which allows increasing the number of gates
65% compared to the old 90nm series. Of course, performance is strong
depends on the specific project, but theoretically using new FPGAs
System frequencies up to 550 MHz can be achieved. LX330 have large logic
capacity and can accommodate up to 330 thousand logical cells (which is the best indicator
today), the maximum number of available inputs/outputs
is 1200, the built-in block RAM memory reaches 10 Mbit (each block
16 Kbit each), while 3.4 Mbit of distributed RAM memory is provided.
The LX330 costs $2,395 for 1,000 units.
If only I understood anything about this: v2_smoke;
3dnevs.ru; copied, I have no second thoughts. Maybe someone can explain!
Hello, CHRV
Well, I’m not thinking about automated workstations, I’m thinking about assembling a real z80 on such
programmable cells and then with common efforts you can modify the kernel
if it is of course modifiable damn it: v2_conf3; Not; I don't chop what I'm carrying
I repeat, the programmable processor is the same at the beginning when loading it is Z80 and then
each program can rearrange commands as it needs, it can reduce clock cycles by
execution of something or the complexity of calculations with minimal clock cycles: v2_confu;
wrapped; however, it obviously won’t work out :v2_cheer; :v2_conf2:;
Hello, CHRV
I won’t argue with your opinion, it’s not correct, maybe I’ll add a little to mine
if such a device would be equipped with a bios that would explain a little how to
To work with this matrix in one word, you need to solve it if anyone does it
There are a lot of problems and it’s possible to work out some kind of standard and some other links.
all in one bottle CPU rom frames and Spectrum peripherals on one such matrix
Hello, Scorpion(lv)
Sco> Well, I’m not thinking about automated workstations, I’m thinking about assembling a real z80 on
Sco> such programmable cells and then with common efforts you can beat
Sco> modify the kernel if it is of course modifiable damn it: v2_conf3; not;
Sco> I’m not cutting what I’m carrying I’ll repeat the programmable processor at the beginning
Sco> when loading it is 380 and then each program can rebuild the commands
Sco> as she needs it can be a decrease in clock cycles for the execution of something or
Sco> complexity of calculations with minimum clock cycles :v2_confu; wrapped;
Sco> however, it obviously won’t work out :v2_cheer; :v2_conf2:;
This is not good news, since we will move away from the roots. Of course from the point of view of free
creativity is interesting, but optimal kernels can be calculated using
the corresponding mathematical apparatus - and this is already boring.
It's like a demo on a PC - it's just not interesting, because results can be achieved
knowing only the technique, and not using clever programming or some other
mathematical delights.
Secondly, there is no standard, that is, we will lose compatibility (one will make a spec
on one matrix, another on another matrix - and each matrix is loaded according to
to your own). Personally, I am for a RIGID, immutable, but COMPLEMENTARY architecture. Those.programs must be able to use a certain inherent architecture - program
released in 2005 should be executed on a computer in 2010, but the opposite is not true
necessarily. ;)
Hello, Scorpion(lv)
Sco> An interesting topic, but I still don’t understand what everyone agreed on. my
Sco> look like this, you need to replace the z80 with the plm so that you take it out and put it in
Sco> I forgot, or maybe I can actually order a new processor somewhere Next to me
Sco> I liked the topic of when information is loaded into the matrix. For starters, you can
Sco> just assemble an analogue of the z80 and then upgrade the commands
Sco> each program can add new commands. programmable
Is Sco> really a processor? If in reality I’m the first to buy this device!
This is a dead end path. Just like an automated workplace, this is no longer a Spectrum - it is an automated workplace. I'm really upset about
8 bits - and this can be written in capital letters!
The workstation is amazing in its simplicity, I’ve been programming them for many years
(most bank terminals are based on them and the new generation
plastic cards), but it is very far from the ideology of the Z80 and 8-bit systems in general.
It is better to consider the directions Rabbit Semiconductor and Zilog ez380.
Hello, Scorpion(lv)
Let me explain (Scorpion): this is exactly what I’ve been interested in lately.
Your percentage What a Spectrum. A couple more years, and FPGAs will be possible with Celeron
shove it... The whole point is that my IMO: it's time to throw the x86 into the landfill. era
shared return stack and data addressable as normal memory, protected with
granularity of 4 or 64 KB - to the dustbin of history. And if not to the landfill, then
Soon we’ll have to close the Internet, throw away our mobile phones and personal computers and go live
somewhere in the village and heat it with coal. Produced with my own hands. Because holes
in software are not taken from the crooked hands of programmers. And not from stupid compilers.
Anyone can make a mistake. As they say, those who do nothing make no mistakes.
The problem (the source of the holes) is not the errors. She is in a completely outdated architecture
current computers. And this applies not only to x86, but also to Motorola, and Spark, and
poppies, and whatever. Except maybe Elbrus, but Elbrus in FPGA is for now
it won't work :)
Hello, Scorpion(lv)
Sco> I won’t argue with your opinion, it’s not correct, I’ll add a little
Sco> your own device could be equipped with a bios that would
Sco> explained a little how to work with this matrix in one word
Sco> if anyone will solve many problems and work out something like this
Sco> standard and other links, can you do everything in one bottle
Sco> CPU rum frames and Spectrum peripherals on one such matrix
Alas, how to work with the matrix is proprietary information and if you are in
If you put such information in your bios, “intellectual rights defenders” will come to you...
It’s another matter if you do open-source matrix emulation on the matrix - but that’s already
Kashchenko :).
You can do everything in one bottle and some do, see the topic about ev. board
Altera.
Hello, Vladimir Kladov
I look at everything that goes on in the fire PS moralism is getting stronger and soon to
to run the operating system you will have to change the apartment since the new computer will not fit
with its mega tera flops, well, if you look at the future, what are we going to?
so that after all the modifications the spec does not cost more than 100, as it was before, I think
this price ceiling should not go higher
Hello, Scorpion(lv)
Yes, but I don't like the noise of the fans. Every time I turn off the computer
in the evening, I understand what my ears are missing: silence. Yesterday I wanted a movie from DVD
look directly at the screw without copying it. Such a howl arose, as if a computer
I wanted to take off like an airplane. And then he screamed for another hour until he cooled down.
I had to open the balcony to speed up the process. Just 3.2 GHz.
Hello, Scorpion(lv)
Sco> The cost of LX330 is $2395 for 1000 pieces.
$2395 for 1 piece in a batch of 1000 pieces will be correct. What is possible
It shouldn't cost more than $30 to use on the Spectrum.
Hello andrews
Sorry for not finishing my word a little. I wanted to say that the new clone should not
cost more and about the peripherals, like a new percentage or a new graph mode, about 30%
from a new clone. Well, for now the theory remains, I haven’t seen it in reality
development of these has begun.
Hello, Scorpion(lv)
One of the big advantages of workstations is the ability to run Linux, and
along with it a large number of programs.
Everything you need for the Internet would be immediately available.
AWP can be installed as the 2nd percent.
Hello Valen
Val> AWP can be installed as the 2nd percent.
this is interesting, especially since there is (almost) such a tandem arm7+z80 in
GBA...but is there a diagram? But regarding the Spectrum, we need to figure out how to do better than them
match
Hello andrews
Anyone interested in pocket consoles with ARM
http://gp2x.com/
Hello andrews
andrews, please explain your calculations. It’s not clear how at a price of 2395 per piece
will be 30. Or is it still the price of the batch. It seems that I have still seen prices of both 10 and
20, and 200 for 1 piece. For these new 2K - I admit that there are not many of them yet. But that's all
bites just the same. If for 1.
Hello Raydac
Ray> For everyone who is interested in pocket consoles with ARM
Ray> http://gp2x.com/
What is this, to catch up with the PSP? How much does this device cost and where is it available?
Hello icebear
http://www.play-asia.com
But the characteristics there are not very good. By the way, this miracle came out before the PSP, at least
at least the first generation for sure.
Hello, Vladimir Kladov
Vla> andrews, please explain your calculations. It’s not clear how at a price of 2395 per
Vla> 1 piece will be 30. Or is it still the price of the batch. It seems so far
Vla> I have seen prices of 10, 20, and 200 for 1 piece. For these new 2K -
Vla> I admit, there are not many of them yet. But it still bites. If for 1.
yes, exactly for 1, because it’s new in the top with a lot of multi-legged body in
industrial version may well cost that much for all major
manufacturers of programmable logic... it is generally always more expensive than ready-made chips
comparable capabilities due to the redundancy of its architecture... that’s according to
essentially a constructor for prototyping in hardware