Dickie troubles forever ;)

20 Apr 99 you wrote here:

MK> for everything below 16550a I would not recommend more than 9600

Well, I have something TMS. And in the dock for momed it says that 16550.
It works at 19200, and at 34800 and at 57600. I haven’t tried above. :)
So here's the question: which one is best to register in the Wild config?
What recommendations will there be?
("turbo" is always on, modem - USR14400)

RM>> IMHO we need to look for the reason. What if he buys himself a new modem for about 50 bucks, with a similar problem? Then he goes to the modem and the PC to buy.

MK> and what about the moneyback?

What do you mean, return it back? And you know how it happens, I was looking for something cooler and cheaper, a random seller, or I traveled far to buy something. There in St. Petersburg the problems
periphery are unfamiliar...

All the best.
Sincerely Roman.

Fri Apr 23 1999 10:15 you wrote to Michael Kondratyev:

RM> Well, I have something from TMS. And in the dock on momed it says that 16550.
RM> It works on 19200, 34800 and 57600. I haven’t tried anything higher. :)
RM> So the question is: which one is better to register in the Wild config?

It won’t work any higher ;) but in principle you can put any one, with 16550 fifo it is
not so critical.

RM> What do you mean, return it back? And you know how it happens, I was looking for something cooler and cheaper,
RM> a random seller, or I traveled far to buy something. There in St. Petersburg
RM> the problems of the periphery are unfamiliar...

well then the reasonable principle would be to prefer the one about which
there was information about a successful docking with zx

Bye, Michael.

MK> it won’t work any higher ;)

What about 115200? :)

MK> but in principle you can put any one, with 16550 fifo it
MK> is no longer so critical.

By the way, I would recommend putting 38400. So cps is a little higher.

WBR, Oleg.

OG> What about 115200? :)

MK is sharing some strange information with everyone. :) For 450 the limit is 19200.
For 550A - 115200. Misha, apparently, is guided by the experience gained when working
with MD. :) There, indeed, Windows does not look at the UART type, but only at
the maximum speed of the modem... For 14.4 with 550A - this is 57600, for 33.6 with 550A -
115200 and for 2400 with 450 - 9600... Insanity...

Tuesday April 27 1999 21:51, Michael Kondratyev ─── Roman Milukov:

RM>> Well, I have something from TMS. And in the dock for momed it says that 16550.
RM>> It works at 19200, and at 34800 and at 57600. I haven’t tried anything higher. :)
RM>> So here’s the question: which one is better to register in the Wild config?
MK> it won’t work higher ;) but in principle you can put any one, with 16550 fifo

Thu Apr 29 1999 07:23 you wrote to Michael Kondratyev:

EM> By the way, in 16650 fifo is implemented with an error.

what am I talking about?

=========== Cut and save ===========
ld b, 0x0B
ld a, 0x03
out (c), a
dec b
ld a, 0x01
out (c), a
ld a, 0x87
out (c), a
in a, (c)
ld h, a
and 0x30
jr nz, @@no_uart
dec l
ld a, h
and 0xC0
cp 0xC0
jr z, @@fifook
xor a
ld (_fifo_value), a
ld l, a
out (c), a
inc a
ld (_fifo_tx_level), a
@@fifook:
========== Cut and save ===========

Bye, Michael.

Thu Apr 29 1999 17:06 you wrote to Oleg Grigoriev:

AN> Something MK shares some strange information with everyone. :) For 450 the limit is
AN> 19200. For 550A - 115200. Misha, apparently, is guided by the experience gained
AN> when working with MD. :) There, indeed, Windows does not look at the UART type, but
AN> only at the maximum speed of the modem... For 14.4 with 550A - this is 57600, for
AN> 33.6 with 550A - 115200 and for 2400 with 450 - 9600... Insanity...

you know, maybe I have some experience I am guided, but I certainly don’t consider Windows as something worthy of even the slightest attention. and
the 57600 limit was due to nothing more than 16-bit

Bye, Michael.

How are you, Michael?

Saturday May 08 1999 23:34, Michael Kondratyev wrote to Alex Noman:

MK> and the limit of 57600 was due to nothing more than 16-bit

16-bit what? UART? So they are all 16-bit, except for 8250, which
have long been forgotten... And the divider is programmed independently of 57600 or
115200 in two bytes, whatever you say...

MK> Bye, Michael.

With respect, Alex Noman.

MK> what am I talking about?
MK> =========== Cut and save ===========
[.....]
MK> ld (_fifo_tx_level), a
MK> @@fifook:
MK> =========== Cut and save ===========
And I also determine the type of uart, and write it beautifully in the window. :) And
fifo support is enabled only if 16550A.

AN> 16-bit what? UART? So they are all 16-bit, except for 8250, about
AN> which have long been forgotten... And the divider is programmed independently of
AN> 57600 or 115200 in two bytes, whatever one may say...

thoughts are simpler. 16-bit speed string parameter in .cfg

Bye, Michael.

Wed May 12 1999 07:48 you wrote to Michael Kondratyev:

EM> And I also determine the type of wart, and write it beautifully in the window. :) And
EM> fifo support is enabled only if 16550A.

as you can see, I do the same, only without windows and inscriptions. we don't
strive for cheap effects ;)

Bye, Michael.

Once on Monday May 17, 1999 at 16:49:29 Alex Noman wrote to Dmitry
Sharikov on the topic 'Dickie troubles forever ;)':

AN>>> No, well, 1800 when transmitting _unpackable_ data by modem on
AN>>> 14.4 - that's for sure steering wheel.
DS>> I download arcmail.
DS>> You are not confusing anything, as far as I know, if the data
DS>> is not packed, then the cps is larger..
AN> Read the letters more carefully. ;) I specially emphasized the key
AN> word. Precisely _unpackable_, not _unpacked_. ;)

And I thought: what did you mean? ;)
But this does not mean that cps is always 1800...

With best wishes, Dmitry.

DS> And I thought: what did you mean? ;)
DS> But this does not mean that cps is always 1800...

People, enlighten me, how is cps calculated in the soaper? Ints mean they don't mess around? I haven't seen anything wild. :(

DS> With best wishes, Dmitry.

With respect, Alex Noman.

21 May 99 18:01, Alex Noman wrote to Dmitry Sharikov:

AN> Folks, enlighten me, how is the CPU calculated in the soaper? In
AN> sense, they don’t end up? I haven't seen anything wild. :(

file size 8
CPS = ──────────────────────── * ───────
file transfer time 10

It’s probably better to ask the RTC - in the Kontratiev scheme, the ints
disappear.

P.S.: There are 10 bits in one byte! I’m serious!

Kirill Frolov. [ZX]

AN>> Ints mean they don’t end up? I haven't seen anything wild. :(
KF> file size 8
KF> CPS = ──────────────────────── * ───────
KF> file transfer time 10
KF> It’s probably better to ask the RTC - in the Contratiev scheme
KF> the ints are lost.

This is all clear, but for the Kondratiev scheme such a formula is unacceptable. :(
Probably you need to conduct some experiments on data transmission at different
speeds and make a table of correction coefficients for the obtained value
tsps. :)

KF> Kirill Frolov.

With respect, Alex Noman.

Sunday May 16, 1999 16:52, Roman Milukov wrote to Evgeny Milun:

RM> I wonder how it will work for me if the dock says that
RM> 16550 is installed; This above procedure finds that
RM> fifook; and in the modem it’s generally not clear what it’s worth. :)

That’s not the point, but the fact that the bourgeoisie are not idiots either. As soon as the 16550 glitch was discovered, they were no longer used. That. The probability of meeting such a mikruha
is not great. That’s why they _very_ often write in the docks (especially the bourgeois ones, I’ve read so many of them...) that 16550 is installed, meaning 16550A, 16550AN and their
analogues. You have an analogue along the way. ;)

RM> Sincerely, Roman.

Regards, Alex Noman.

AN>> which have long been forgotten... And the divider is programmed
AN>> regardless of 57600 or 115200 in two bytes, whatever one may say...
MK> thoughts are simpler. 16-bit speed line parameter in .cfg

You finish the sentence right away, then you won’t have to think at all... :)

EM>> :) And fifo support is enabled only if 16550A.
MK> as you can see, I do the same thing, only without windows and
MK> inscriptions. we don't strive for cheap effects ;)

People here even ask to do a test, and you're talking about cheap effects... ;)

Why convert to a number? :-/

After all, there is a fixed number of speeds - consider the number with
speed as a text string and match it with the value written
in the UART.

Kirill Frolov. [ZX]

AN> soaper? Ints mean they don't mess around? I haven’t seen anything wild.
AN> :(

I don’t know, I can say that cps starts to appear after 1 minute
downloading the file. ;)

Do you need an algorithm?

With best wishes, Dmitry.

KF> Why convert to a number? :-/

I thought about this too... :)

KF> After all, there is a fixed number of speeds - consider the number
KF> with speed as a text string and match it with the value
KF> written to the UART.

You can simply enter some constants or letter designations, such as A -
115200, B - 57600, etc. But on the other hand, there are modems that support non-standard boxes. I came across one like this. So MK is flexible in this regard. Really, who needs all this (I’m talking about non-standard)? :))

EM>> And I also determine the type of uart, and write it beautifully in the window. :) And
EM>> fifo support is enabled only if 16550A.

RM> I wonder how it will work for me if the dock says that 16550
RM> is installed;
If it really is 550 (and this can be determined by registers - 550 from 550A
are different somehow bit), then my terminal will not enable fifook.

RM> this above procedure finds that fifook; and in the
RM> modem it’s generally not clear what it’s worth. :)
I don’t even know. It is written everywhere that Fifo 550 was implemented with errors, but
it is not written anywhere what they are.

Sunday May 16 1999 16:59, Roman Milukov ─── Evgeny Milun:

RM>>> it was time to test the modem. Or a separate program. :)
EM>> Test it completely? I don’t even know for the presence/absence of what exactly
EM>> it can be tested. :)))

RM> Yes, you can come up with a lot! At least the same disappearances of symbols,
Hmmm... Ah, not a bad idea...

RM> all sorts of errors
RM>, because it’s usually all corrected, and outwardly you can’t see what’s going on there
RM>.
RM> Do you remember, there was a whole FAQ here, about the fact that you need to find some
RM> basic term and set up an interface circuit with computer?..
Nope, I don’t remember that. Must watch.

AN> This is all clear, but for the Kondratiev scheme such a formula is unacceptable.
AN> :( Probably you need to carry out some experiments on data transmission
AN> at different speeds and make a table of correction coefficients
AN> for the resulting CPS value. :)

Do you even understand what you wrote?

CPS = Charactesr Per Second !!!

This is mine! formula and reflects. One Character has 10 bits - probably 1 start bit and one more. All programs think this way - the ratio of the number of bits transmitted to the transmission time.

If it is not true for the Kondratieff scheme, then time in the Kondratiev
scheme does not flow the same way as in the rest of the universe...

Kirill Frolov. [ZX]

How are you, Dmitry?

Sunday May 23, 1999 16:36, Dmitry Sharikov wrote to Alex Noman:

DS> I don’t know, I can say that cps begins to appear after 1 minute
DS> of downloading the file. ;)
DS> do you need an algorithm?

Well, if it differs from what Kirill gave, then go ahead. ;))

RM>> что 16550 установлена;

Мало ли, что в доке написано. Главное - это то, что есть на самом деле.

---
Chip Detection

It is rather easy to detect what kind of UART is installed:

An 8250 does not have a scratch register
A 16450 does not have a FIFO
A 16550 has bad FIFOs, indicated by bit 7 of IIR
A 16550A has good FIFOs, indicated by bit 7 and bit 6 of IIR

You can use the following algorithm to detect the UART type. BASE is
the base address of the serial port. (usually 3F8 for COM1, 2F8 for COM2,
etc)

IIR = BASE+2 = interrupt identification register (read only)
FCR = BASE+2 = FIFO control register (write only)
SCR = BASE+7 = scratch register (read and write)
Bits are numbered from 0 to 7, 7 is high bit

Read and save the SCR
Store a test value into SCR (hex 5A is good)
Read SCR and compare to test value
If not equal, there is no scratch register, so the chip is an 8250

Store another test value into SCR (hex A5 is good)
Read SCR and compare to test value
If not equal, there is no scratch register, so the chip is an 8250

Restore the saved value from the SCR
Read and save the IIR (saves current possible FIFO status)
Store 1 into FCR (enables possible FIFOs)
Read IIR
If saved IIR value had bit 7 clear, store 1 into FCR (FIFOs were off)
If IIR had bit 6 set, the chip is a 16550A
If IIR had bit 7 set, the chip is a 16550
Otherwise, the chip is a 16450

---

AN> People here even ask you to do a test, and you're talking about cheap
AN> effects... ;)

What's so funny?! I got a new Momed the day before yesterday. In the morning.
And what was I supposed to do, sit until the evening, wait for the node to start working, try
to connect? (I shoved that piece of Mishin into the BASIC file when he threw it here.) And then I came, started it - the inscription “FIFO OK!”, and I calmed down until the evening. :)

All the best.
Sincerely, Roman.

AN> Well, if it differs from what Kirill gave, then go ahead.
AN> ;))

I won’t give it... :))
Because it is no different...

With best wishes, Dmitry.

AN>> is unacceptable. :( Probably you need to carry out some experiments
AN>> on data transmission at different speeds and make a table
AN>> of the correction factors for the resulting CPS value. :)

KF> Did you even understand what you wrote?

Well, it would be strange if you wrote and didn’t understand.

KF> CPS = Charactesr Per Second !!!

It can’t be!!! :-() I measured everything in liters...

KF> This is my formula and reflects it. One Character has 10 bits - probably 1
KF> is the starting one and there is one more. This is what all programs think - the ratio
KF> of the number of bits transmitted to the transmission time.

KF> If it is not true for the Kondratieff scheme, then time in the
KF> Kondratieff scheme does not flow the same way as in the rest of the universe...

That is so. Time slows down slightly... This is the only drawback of Kondratiev's scheme... It's easy to do. The watch will help you...

KF> Kirill Frolov.

With respect, Alex Noman.

ps/ Do you really believe in 1800 bytes? ;(

EM> I don’t even know. It is written everywhere that Fifo 550 was implemented with
EM> errors, but nowhere is it written what they are.

Fri May 14, 1999 at 16:40:16, Alex Noman => Dmitry Sharikov:

AN> No, well, 1800 when transmitting _non-packable_ data by modem on 14.4 is definitely
AN> steering wheel.

In the subject, this is definitely not realistic. For many reasons.

WBR, Oleg.

Eeeeeee... You're already slipping on your butt. It is important there how many bits are in a byte.
For the DSP, it is only important for how long it took for such and such amount of useful
information to arrive.

KF> This is what all programs think - the ratio of the number of transmitted bits to
KF> transmission time.
Not a bit, a byte.

KF> If it is not true for the Kondratiev scheme, then time in the Kodndratieff
KF> scheme does not flow the same way as in the rest of the universe...
This is where you didn’t understand him... He meant that in this scheme, when calculating time using the standard method (IM2), it is necessary to take into account the fact that modem
interrupts have a higher priority, which means the clock will be slightly behind. This is where correction factors are needed.

DS> after 1 minute
DS> downloading the file. ;)

Now I have a problem. Or did it disappear?

There used to be a USR modem. No matter how I distorted it, cps began to appear after
some time (maybe a minute, I didn’t time it). In general it worked well.
Now I installed Acorp. Good too. At first, with default settings, everything was exactly the same. Then I turned the modem on, configured it as I wanted, and
suddenly cps began to appear after receiving the first 2-kilobyte block
(small) and increased before our eyes. And, it seems, the estimated time for receiving a file
began to be shown in seconds (previously it was in minutes).
Question: what affected it? what adjustment? maybe it should be returned as it was?
I changed a lot of things, but look:
there was &Q5-modem negotiates an error corrected link
&R1-CTS always active
set &Q0-select derect asynchronous mode.
&R0-CTS tracks RTS (asinc)
Did I do it right? And how to put it:
&S0-DSR alweys active or
&S1 - this is not in the dock, but the command responds OK.

P.S. The protocol is hydra. Interrupts are implemented, I use dosx mmki_rtc.

RM>> characters,
EM> Hmmm... Ah, not a bad idea...

And the modem may not be configured the way your terma/myler wants. All sorts of
DSR, CTS/RTS, etc., and the user can be a teapot. It’s good when there is detailed
documentation, but look at the Wild ones; and someone has a unique modem and without docking...

AN>> this is definitely a steering wheel.
OG> In the subject this is definitely not real. For many reasons.

This is exactly what I wrote about... This is not real, not only in the subject. It's easy to count
you have to be able to...

OG> WBR, Oleg.

With respect, Alex Noman.

27 May 99 17:21, Alex Noman wrote to Kirill Frolov:

AN> That’s right. Time slows down a little... This is the only
AN> drawback of Kondratiev's scheme... It can be done easily. The watch will help you...

Or 580ВH59.

AN> ps/ Do you really believe in 1800 bytes? ;(

And in Russian in white?

Kirill Frolov. [ZX]

KF>> in the Kondratiev scheme flows differently than in the rest of the
KF>> universe...
EM> That’s where you didn’t understand him... He meant that in this scheme
EM> when calculating time using the standard method (IM2), one must take into account the fact that in
EM> modem
EM> interruptions have a higher priority, which means the clock will lag a little.
EM> This is where correction factors are needed.

Useless - the Sevan alarm clock will be more useful.

Kirill Frolov. [ZX]

I wonder if I don’t need to believe my eyes?
1800 rarely happened, but I see 1700 quite often... ;)

With best wishes, Dmitry.

Friday May 28 1999 17:20, Roman Milukov wrote All:

RM> Now I installed Acorp.

33.6?

RM> it as I wanted, and suddenly cps began to appear after receiving
RM> the first 2-kilobyte block (small) and growing before
RM>'s eyes. And, it seems, the estimated time for receiving a file is now shown in
RM> seconds (previously it was in minutes). Question: what influenced? what
RM> adjustment? Maybe we should return it to the way it was?

Try setting the factory settings as an experiment and see. Maybe nothing will change. :)

RM> I changed a lot of things, but
RM> look: it was &Q5-modem negotiates an error corrected link
RM> &R1-CTS always active

In principle, in many source codes that I saw, CTS was completely ignored... In
the idea, when sending you need to wait for CTS. But this was all in theory. In practice
to obtain greater efficiency of the handler, many things are neglected.
Moreover, it was always active. :) I don’t know how in the subject.

RM> set &Q0-select derect asynchronous mode.
RM> &R0-CTS tracks RTS (asinc)
RM> Did I do it right? And how to put it: &S0-DSR alweys active or
RM> &S1 - this is not in the dock, but the command responds OK.

Well, load the c-terminal and look at all the signals. And CTS, and DSR, and DTR, etc.

RM> P.S. The protocol is hydra.Interrupts are implemented, I use dosx mmki_rtc.

I think that the modem settings, although they influence something, but not to such an extent that the calculation of time in programs can be changed (minutes to seconds). ;)
Although, I’ll say again that since I temporarily gave the spec to MMSM, I haven’t seen dosx at all
yet... :( Therefore, I can’t say anything definite.

RM> Now I have a problem. Or has it disappeared?

RM> There used to be a USR modem. No matter how I distorted it, cps began
RM> to appear after
RM> some time (maybe a minute, I didn’t time it). In general
RM> worked well.
RM> Now I installed Acorp. Good too. At first, with default
RM> settings, everything
RM> was exactly the same. Then I twisted the modem, configured it the way
RM> I wanted, and

[skip]

RM> I changed a lot of things, but look:
RM> there was &Q5-modem negotiates an error corrected link
RM> &R1-CTS always active

Mon 31 May 1999 at 00:45:56, Dmitry Sharikov => Oleg Grigoriev:

DS> Interesting, I don’t need to believe my eyes?
DS> 1800 rarely happened, but I see 1700 quite often... ;)
~~~~
Ha arcmail?

WBR, Oleg.

Thu May 27, 1999 you wrote here:

EM> This is where you didn’t understand him... He meant that in this scheme, when calculating time using the standard method (IM2), one must take into account the fact that modem
EM> interrupts have a higher priority, which means the clock will be a little slow. This is where correction factors are needed.

Correction factors will not help much. The error will very much depend on the quality of the connection. If you want real cps, set a real clock.

EM>> I don’t even know. It is written everywhere that Fifo 550 was implemented with
EM>> errors, but nowhere is it written what they are.

KF> And in Russian?

Modem on 14.4 + circuit, eating ints gives an unreal DSP of 1800 bytes... Do the math.
Or conduct an experiment on your PC... It seems to add up without glitches...

DS>> Interesting, I don’t need to believe my eyes?
DS>> 1800 rarely happened, but I see 1700 quite often... ;)

AN> Well, load the s-terminal and look at all the signals. And CTS, and DSR, and DTR, etc.
Come on? What version is this? :) In both 0.1 and 1.0, only the state of the
DSR line (or DTR?) is visible.

AN> Try setting the factory settings as an experiment and see. Maybe
AN> nothing will change. :)

But at the beginning I tried it with almost factory settings and everything was fine,
as with the old modem.
And here’s what I think: the time units have not changed, it’s just that with cps=28, which
initially popped up, it (the value) is huge. :)
Only I don’t understand where it comes from in the subject, when cps is not yet known?

RM>> look: it was &Q5-modem negotiates an error corrected link
RM>> &R1-CTS always active

AN> In principle, in many of the source codes that I saw, CTS was completely ignored...
By
...
AN> Moreover, it was always active. :) I don’t know how in the subject.

I’m now interested in exactly “how in the subject?”.

Yes, it seems... ;)
Don’t forget, we both have electronic telephone exchanges...
You won’t get such a DSP from Melon (800-900), I was dragging my feet at first... ;))

With best wishes, Dmitry.

AN>> DTR, etc.
EM> Come on? What version is this? :) In both 0.1 and 1.0 you can only see
EM> the state of the DSR line (or DTR?).

Screw you... I confused it with my program. There I show everything that is possible. ;)

EM> Good luck! Evgeny.

With respect, Alex Noman.

KF> How do they differ?
“One bit is a unit of information that can be transmitted by two
equally probable symbols” (C) theory of information transmission. "A byte is usually one octet, i.e. 8 bits." (C) she is the same. :)))
And the difference is that, as you say, 10 bits are transmitted over the wire for one character in one octet. But, a sign is equal to an octet, and therefore
it is necessary to count the information in octets.

AN> Modem on 14.4 + circuit, eating ints gives an unreal DSP of 1800 bytes...
AN> Do the math. Or conduct an experiment on your PC... It seems to add up without
AN> glitches...
As far as I remember in the modem, the following formula for calculating the
ideal CPU was once used:
CPS=0.95*SpeedConnect/8

FK> No, that's not right! What you need is &Q5! Or &Q9, if you want
FK> to connect only to v42bis.

Exactly. If you first set Q0 and then turn on force LAPM mode, then it itself
changes to Q5.
And I don’t have Q9.

DS> Yes, it seems... ;)

So, it's a shitty archmail. :) Or it’s considered crooked.

DS> You won’t get such DSP from Melon (800-900), I was dragging my feet at first... ;))

Where? :)

WBR, Oleg.

AN> Come on... I confused it with my program. There I show everything that is possible. ;)
Judging by the date and time, you realized this 40-50 minutes after we
left you? You're quick, quick. :). You have to kick him, the lazy fellow. Kick him, otherwise he, you see, has gotten tired of painting the frames. :)))

Wed 2 Jun 1999 at 21:55:47, Alex Noman => Oleg Grigoriev:

OG>> On the arcmail?

AN> But what does the arcmail have to do with it? Is common sense just gone? ;)

What if you transmit a meter of zeros? :)

WBR, Oleg.

Tuesday June 01 1999 10:29, Roman Milukov wrote to Evgeny Milun:

RM> Correction factors will not help much. Very much

It's better than nothing...

RM> the error will depend on the quality of the connection. If you want real cps
RM>, set a real clock.

It's all clear...

EM>>> I don’t even know. It is written everywhere that Fifo 550 was implemented with
EM>>> errors, but nowhere is it written what they are.

RM> Yeah.

That's good. This means that Pover definitely has glitches. :)

RM> have not changed, it’s just that with cps=28, which initially popped up, it
RM> (raising) is huge. :) But I don’t understand where it comes from in the subject
RM> when cps is not yet known?

I think that if the cps is not displayed, this does not mean that it is not known.
;) Errors in the calculations. Apparently, the program has some restrictions on the range of
output/calculated values. Although, this is all guesswork. I didn't see it.

And you ask your boss what kind of CPU he has when you have 1800. ;)

With respect, Alex Noman.

ps/ 1600 is the limit.

Ideal-maximum? ;)

AN> They consist in the fact that characters are lost. :))

Catch up. As far as I remember from the documentation, in 16550 the maximum buffer size is 16 bytes, 16550A and analogues are 14 bytes. Although, all smaller sizes, such as 8, etc. - coincide. You can, of course, make the assumption that the 16550 will work perfectly well with a buffer of less than 16 bytes, but the assumption cannot be verified for the reason that you won’t find the 16550 during the day. :)

With respect, Alex Noman.

AN>> missing? ;)
OG> What if you transmit a meter of zeros? :)

You skipped everything. We are talking about unpacked data. Data that cannot be
packed. Otherwise, the meaning of the very concept of TsPS is lost. ;)

Friday June 04 1999 16:39, Evgeny Milun wrote to Alex Noman:

EM> Judging by the date and time, you realized this 40-50 minutes after
EM> we left you? You're quick, quick. :). I've already forgotten what's what... Nobody
wants to write the interface part and scripts (full-fledged)? :)

EM> 2ALL: This is who the modem tester should be entrusted with writing - he has a mania
EM> to show everything that is possible. You have to kick him, the lazy fellow. Kick,
EM> otherwise he, you see, got stuck painting the frames. :)))

It’s a shame the moderator upvoted you, lamer. :) There is soap for such messages... F:)

Good luck! Evgeny.

■ The night is shorter than the day... The day will kill me... The world of illusions burns down in it. (C) Aria.

AN> And you ask your boss what kind of CPU he has when you have
AN> 1800. ;)

I won’t ask ;)
He doesn’t sit near the computer all the time.

AN> ps/ 1600 is the limit.

Why?

With best wishes, Dmitry.

AN> You skipped everything. We are talking about unpacked data. Data that cannot
AN> be packed. Otherwise, the meaning of the very concept of TsPS is lost. ;)

We were talking about unpackable data before the advent of the phrase “common sense”.
For this phrase claimed the impossibility of the existence of cps
more en in principle. Which is generally not true. And on archmail - yes.

WBR, Oleg.

p.s. It's funny, before this I had never encountered a person who would make every effort to misunderstand me. :) Maybe
KF, but I'm already used to it. :)

DS> Why ?

He did. Ppedel -- the scope of the popta's bow.

Kirill Frolov. [ZX]

EM>> As far as I remember in ru.modem, the following formula was once used
EM>> for calculating the ideal CPU: CPS=0.95*Connect Speed/8

Good luck! Evgeny.

■ The night is shorter than the day... The day will kill me... The world of illusions burns down in it. (C) Aria.

KF> No limit. The limit is the speed of the port link.
For the DSP ????!!!!!
Then I’ll explain it again. Slower:
CPS - an indicator characterizing the speed of the LOGICAL transmission channel
(in simplified terms of the DTE-DTE channel).
Link speed & connection speed - indicators characterizing the speed
PHYSICAL transmission channel (moreover, the link sets the channel speed DCE-DTE,
connection speed - DCE-DCE).

ïÄÎÁÖÄÙ × ðÏÎÅÄÅÌØÎÉË 7 éÀÎÑ 1999 × 16:15:55 Alex Noman ÎÁÐÉÓÁÌ(Á) to Dmitry
Sharikov ÎÁ ÔÅÍÕ 'Dickie troubles forever ;)':

DS>>>> 1800 ÒÅÄËÏ ÂÙ×ÁÌÏ, ÎÏ 1700 ÄÏ×ÏÌØÎÏ ÞÁÓÔÏ ÎÁÂÌÀÄÁÀ...
DS>>>> ;)
OG>>> HÁ ÁÒËÍÅÊÌÅ?
DS>> äÁ, ×ÒÏÄÅ ÂÙ... ;)
AN> á ÔÙ Õ ÂÏÓÓÁ Ó×ÏÅÇÏ ÓÐÒÏÓÉ, ËÁËÏÊ Õ ÎÅÇÏ ÃÐÓ, ËÏÇÄÁ Õ ÔÅÂÑ
AN> 1800. ;)

÷ÏÔ, ÓÍÏÔÒÉ ÉÚ ÓÔÁÔÉÓÔÉËÉ ÂÏÓÓÁ ×ÙÄÒÁÌ. ðÏÓÍÏÔÒÉ ÎÁ max.

==================== file: cps .C =========================

Summary link statistics
(08/06 00:08:42 - 08/06 17:00:01)

‚€€€€€€€€€€€€€ Password: '*' - Present, ' ' - Absent, '?' -
Error
‚€€€€€€€€ Outgoing calls
‚€€ Sessions
‚ Last Session Result: 'ž' - Success, ' ' -
Aborted
¥       ¶              ¶           ¶                  ¶             ¨
¡P O S L Address Days/Time Bytes Bytes CPS ¡
¡ Online received sent Min. Max. Avg.¡
±       ¼              ¼           ¼                  ¼   ¼    ¼    µ
¡* 0 4 ž2:5030/675.7 0:06:30 334,050 83,652 344 1987 1176¡

==================== end of cps .C ========================

AN> ps/ 1600 c ÇÁËÏÍ - ÐÒÅÄÅÌ.

ëÁË ×ÉÄÉÛØ ÎÅ ÐÒÅÄÅÌ... ;)
õ ÎÅÇÏ ÄÕÍÁÀ ÎÉÞÅÇÏ ÎÅ ×ÒÅÔ...

ÚÙ. ÷Ù×ÏÄ: "äÉËÉÊ" - ÒÕÌÅÚ.

With best wishes, Dmitry. [ZX]

AN>> cannot be packed. Otherwise, the meaning of the very concept of TsPS is lost. ;)
OG> We were talking about unpackable data before the phrase “common
OG> sense” appeared. For this phrase claimed the impossibility
OG> of the existence of a cps greater en in principle. Which is generally
OG> not true. But on archmail - yes.

It just so happens that the DSP is a rather conditional characteristic. It depends on the
line, software, hardware, settings, etc. etc. In short, it will simply be an empty
sound if you do not adhere to some framework when discussing “what is your
CPS??? :-()”.

OG> p.s. It's funny, before this I had never encountered a person
OG> who made every effort to understand me
OG> wrong. :)

I understand you... Meter of zeros... It happens...

With respect, Alex Noman.

Do the math. Even the formula that EM gave shows this. Although there is an ideal there,
it seems like it. ;)

DS> With best wishes, Dmitry.

With respect, Alex Noman.

Wednesday June 09 1999 00:17, Kirill Frolov sent Dmitry Sharikov:

11 Jun 99 23:13, Dmitry Sharikov wrote to Alex Noman:

DS> As you can see, the limit is not... ;)
DS> I think he’s not lying...

DS> Speak. Conclusion: "Wild" - rulez.

Have you seen what kind of CPS there are at the beginning of file transfer? While the modem and fossil buffers are being filled 4 times more! And then my FHMail is incorrect
CPS calculates, shows inflated results (1.5 times exactly) - look
only for long files.

Kirill Frolov. [ZX]

EM> That’s why they came up with logs. :))

Yesterday in the statistics, in general, I saw the maximum cps = 2000 with a hook... ;)
(this is not a joke, I accidentally killed that letter)

10 Jun 99 07:46, Evgeny Milun wrote to Kirill Frolov:
EM> CPS is an indicator characterizing the speed of the LOGICAL channel
EM> transmission (simply speaking the DTE-DTE channel).
EM> Speed boxes & connection speed - indicators characterizing
EM> the speed of the PHYSICAL transmission channel (moreover, the box sets the speed
EM> of the DCE-DTE channel,
EM> the connection speed - DCE-DCE).

And where is data compression?

CPS is limited only by what You'll quickly be able to stuff
_identical_ bytes into the momed.

Kirill Frolov. [ZX]

11 Jun 99 23:13, Dmitry Sharikov ضءج ثخدذثة ؤجر Alex Noman:

DS> ،P O SپL Address پ Days/Time پ Bytes Bytes پ CPS ،
DS> ، پ پ Online پ received sent پMin.پMax.پAvg.،
DS> ،* 0 4پ‍2:5030/675.7 پ 0:06:30پ 334,050 83,652 پ344پ1987پ1176،
DS>

ل سزإؤخةت CPS ؤد×دجطخد اد×£خظت :(

AN>> ps/ 1600 c اءثدح - ذزإؤإج.

Hص حدضإش 1700 خء ةحذدزشخظب لôَ.

DS> عظ. ÷ظ×دؤ: "نةثةت" - زصجإع.

ïخ شد حدضإش ة زصجإع, خد ثءقإسش×د س×رعة خإ سحدشزر خء ش×دہ
ـج. لôَ - سصثسط.

َ ص×ءضإخةإح, àزء DOORSFAN
[ZX] [DOORS] [BLACK SABBATH]

AN>>> ps/ 1600 with a hook is the limit.
DS>> Why ?
AN> Do the math. Even the formula that EM gave shows
AN> this. Although there is an ideal, it seems. ;)

I already posted a piece of my boss’s log. CPS (maximum) was 1900 and then... day I saw in his log 2000 with a hook...
(but this is possible when I threw something large and unpacked)

DS>> That is. do you want to say that on a 14400 modem you can
DS>> get cps, roughly speaking 3000, and that’s not the limit
DS>> ?
KF> You can. But not always!

For some reason I doubt it very much...

DS>> PS. Conclusion: "Wild" - rulez.
KF> Have you seen what kind of CPS there are at the beginning of file transfer?

No, “wild” shows the CPU only after a minute of file transmission/reception.

DL> yes. on uncompressed data, the CPS limit is the lock speed,
DL> divided by 10.

I.e. It turns out that by increasing the speed of the link, I increase the transmission speed
?

Who measures the CPS like that? :) We don’t even know at what moment the maximum CPU was.
Maybe it happened on netmail? :) Then you are not 675.7. And who will give
a guarantee that the 675.7 modem is not 16800? ;))) There can be many excuses, but it all boils down to the fact that 1987 is a fairy tale. The DSP is also the average value.
If you have noticed, soapers show the DSP for the entire session or at least for the entire
transmitted file. In Internet programs, the CPU is also calculated, although not for the entire
time of file transfer, but for some fairly large interval (
10 seconds by eye ;)).

DS> That is. It turns out that by increasing the transmission speed, I increase the
DS> transmission speed?

Not always. But definitely on uncompressed data.

Kirill Frolov. [ZX]

DS> for some reason I doubt it very much...

Should I send you a file with a zero meter?

Kirill Frolov. [ZX]

no. you set the upper limit.

All the best.
Dmitry.

... LD..X-TRADE... [ZX]

AN>> ps/ 1600 with a hook is the limit.
DS> As you can see, it’s not the limit... ;)
DS> I don’t think anything is wrong with him...
Of course, it happens that the maximum CPS reaches 4 thousand, but you need to rely
on the average. And the average, as you can see, is far from the figure of 1600. Here's some information for you:
The average CPS for 14400 modems is 1638, and anything more is a matter of
hardware compression of the modem.
Sincerely, SP/PoS.

YY> And the average CPS is pretty shitty :(

These numbers should not be unconditionally trusted:
1. For short files (up to a second) writes a cps equal to their length in bytes.
2. The statistician has some documented bug: min & max cps is calculated by
files, and avg by the whole session (i.e. the average CPU is underestimated). Taking into account the fact that the soap has the slowest performance of all soapers I have ever seen, we can draw the correct conclusions ;) One of which is quite funny: in a session with one transferred file, the average CPU is according to this statistician. _always_
will be less than the minimum.

With best wishes, Vladimir.

YY> He may rule, but the quality of the connection despite
YY> your email. ATS is a bummer.

The connection quality is ok, but the fact that I have a momed plugged into the relish and the hydra is being used
has an effect, because there is no simultaneous transmission... :(

DS>> increasing the transmission speed ?
DL> no. you set the upper limit.

The maximum speed will be higher...

16 Jun 99 01:02, Dmitry Sharikov pressed the buttons for Yura Yakimuk:

DS> The quality of the connection is ok, but the fact that I have a momed plugged into the relish and the hydra
DS> is used affects, because there is no simultaneous transmission... :(

That is, the wild one does not work FULL DUPLEX? As I understand it the whole joke of Hydra lies in this
Then IMHO it is better to use Z-MODEM.

Sincerely, Yura DOORSFAN
[ZX] [DOORS] [BLACK SABBATH]

EM>>> formula for calculating the ideal CPU: CPS=0.95*SpeedConnect/8
AN>> Ideal-maximum? ;)
RM> And by the way, this is on HST. On other protocols, it seems, 0.91-0.93.

Did you see it there too?

RM> Sincerely, Roman.

With respect, Alex Noman.

KF> And where is data compression?
Like where? As a rule, at almost all levels of the ISO/OSI model. Both in
DTE-DCE and in DCE-DCE...

KF> CPS is limited only by the speed at which you will push _identical_ bytes into the module
KF>.
CPS is a characteristic of the speed of the DTE-DTE logical channel (DTE - Data
Terminal Equipment, DCE - Data Carry Equipment), and depends on applied
protocols both at this level (application) and all the following (representational, channel, etc.). To put it very simply, this is the speed with
which two procedures exchange with each other: a transmitter procedure that sends
SIGNIFICANT information down the stack of levels, and a receiver procedure that receives
SIGNIFICANT bytes from lower ones levels. I have nothing more to add to this...

P.S. I get the feeling that you are simply not familiar with both the seven-layer ISO/OSI model and the principle of the protocol stack...

DS> They immediately explained that there is a limit...

Yes.
And cps will also greatly depend on where the mailer puts the mail. I’m generally silent about the flop. I'm putting it in a ramdisk, because I have enough of it, but the screw
is still slower.

From CoModerator to All:
Let's move on to RU.MODEM with the bazaars about cps modems in general (not related to Spectrum
mylers)! There, these problems have already been solved a long time ago and they will give you
qualified answers, but here they will soon reward you.

All the best.
Sincerely, Roman.

Saturday June 12, 1999 16:53, Dmitry Sharikov wrote to Evgeny Milun:

DS> Wild thinks everything is correct...

Nobody argues with this... The Russians told you in plain English that it’s a matter of keeping track of
time. The wild one defends himself here... in the sense that nothing depends on him.

DS> With best wishes, Dmitry. [ZX]

With respect, Alex Noman.

Monday June 14, 1999 11:51, Dmitry Sharikov wrote to Dmitry Lomov:

DL>> yes. on uncompressed data, the CPS limit is the locking speed divided by 10.

Monday June 14, 1999 11:46, Dmitry Sharikov wrote to Alex Noman:

DS> I already threw a piece of my boss’s log. CPS (maximum) was 1900 s

We walk around and around. There are simply modem capabilities on 14.4. And they
are limited. ;)

DS> With best wishes, Dmitry. [ZX]

C уважением, Alex Noman.

Wednesday June 16 1999 08:38, Konstantin Samoylov wrote to Dmitry Sharikov:

KS> Here's some information for you: The average CPS for 14400 modems is 1638,
KS> and anything more is a matter of hardware compression modem.

Whatever is more? :) Then this is not the average, but the maximum. ;) For unpacked
data, of course. In general, the accuracy is amazing. I have exactly the same CPU on
14.4 most often.

KS> Sincerely
KS> SP/PoS.

With respect, Alex Noman.

Fatten you, cut your hair and you'll look just like Gaidar - "... during
devolution in terms of the second derivative...". :)

EM> P.S. I get the feeling that you are simply not familiar with both
EM> the seven-level ISO/OSI model and the principle of the protocol stack...

Does he need it? :) The person doesn’t know/didn’t understand where something is packaged, and you
immediately called him a lamer. ;)))

EM> Good luck! Evgeny.

With respect, Alex Noman.