about CPS
ZXNet echo conference «real.speccy»
From Leonid Mishankov → To All 7 October 1999
Why is it that in the S-Terminal (and not only in it) the CPS for reception is ~1500, and for transmission
much less - ~1100?
In Dikiy there’s a problem: there’s always 1100, I’ve
set the lockbox and everything else correctly.
S!ncerely Yours,
Leonid Mishankov.
From Evgeny Milun → To All 8 October 1999
Thursday October 07 1999 23:38, Leonid Mishankov ─── All:
LM> Why in the S-Terminal (and not only in it) the CPS for reception is ~1500, but on
LM> the transmission is much less - ~1100?
I can speak for the S-Terminal. :) Because there the transfer proceeds in a normal
way, and the exact time of transfer can be easily measured with INTs. When receiving, the reception goes through the NMI, which means the INTs are partially clogged, and their handler
works fewer times. In a word, for the program it turns out that when receiving the same number of bytes, fewer interruptions occur than when transmitting, i.e. Less time passes. How to compensate for this? That's right - enter a correction factor. It is in its
meaning that the error lies.
Good luck! Evgeny. (Aka 2:5057/18.73) ■ The night is shorter than the day... The day will kill me... The world of illusions burns down in it. (C) Aria.
From Leonid Mishankov → To All 9 October 1999
EM> I can speak for S-Terminal. :) Because there the transfer proceeds in the normal
EM> way, and the exact time of transfer can be easily measured with INTs. When receiving
[Skipped]
By the way, I’ve long wanted to ask, have you thought about implementing Z-Fast, like in Melon?
(this is when 10kb blocks fly). It turns out to be very effective.
Sincerily yours,LEO/SG.
From Evgeny Milun → To All 9 October 1999
LM> [Skipped]
LM> By the way, I’ve been wanting to ask for a long time, but you haven’t thought about implementing Z-Fast,
Worse - about implementing Z-Intelectual. It turns out that you can get a lot out of ZModem. The idea is that the receiver transmits the size of its buffer and flags to the transmitter. Based on this, the transmitter itself selects the transmission method (there is such a feature - 4-5 methods) and all the time adjusts the length of each block based on the quality of the connection.
S-Terminal and Melon (as far as I know) do not support these capabilities.
They do not support them yet. :)
From Leonid Mishankov → To All 10 October 1999
LM>> By the way, I’ve been wanting to ask for a long time, but have you thought about implementing Z-Fast,
EM> Worse - about implementing Z-Intelectual. It turns out that you can get
EM> a lot out of ZModem. According to the idea, the receiver transmits the size of its buffer to the transmitter, and
that is, at every opportunity, the window will grow to some final size?
P.S. I have some kind of garbage in Wild: at first the blocks are 256 bytes, after 2 seconds
- 512, then 1024, 2048, 4096 and then 8192!
Sincerily yours,LEO/SG.
From Evgeny Milun → To All 10 October 1999
LM> that is, at every opportunity the window will grow to some final
LM> sizes?
Yes.
LM> P.S. I have some kind of garbage in Wild: at first the blocks are 256 bytes, after 2
LM> seconds - 512, then 1024, 2048, 4096 and then 8192!
This is most likely what it is.
From Anton Kriwoshea → To All 12 October 1999
LM> 256 bytes, after 2 seconds - 512, then 1024, 2048, 4096 and then
LM> then 8192!
For me it’s even worse ;), first, in Melon, block 10240, then 256 bytes, then again
10240 etc... around the ring :(, another situation:
Melon issues: 10240....10496...20224...20770... etc... also in
loop :(. Does anyone know what kind of garbage this is? $k1v/M7H (orp.'99.
From Oleg Pritula → To All 14 October 1999
12 Oct 99 at 18:29, Anton Kriwoshea squeezed the keyboard, hoping that such words would reach Leonid
Mishankov about “about CPS”:
AK> p.s.: How can you see CPS in Melon? ;)
No way.
\\|// Hasta la vista !!! Curator from the prefecture:
o(@ @)o -=POWer/WDG/AF=-
.-----ooO-(_)-Ooo-----. E-Mail: po...@nts.hippo.ru
From Felix Knajzev → To All 16 October 1999
[Tuesday October 12, 1999] at [19:29:31] Anton Kriwoshea wrote a letter to
Leonid Mishankov.
They discussed the topic “about CPS”. Let's continue what we started...
[skip]
AK> My situation is even worse ;), first, in Melon, block 10240, then 256
AK> bytes, then again 10240 etc... along the ring :(, another
AK> situation:
AK> Melon issues: 10240....10496...20224...20770... etc... also
AK> in
AK> cycle :(. Maybe someone knows what kind of garbage this is? :(
On the “other” side they use ZModem, transmitting the so-called continuous
stream. Brake the boss to set the frame length for transmission to *8192* byte
and, accordingly, set a *confirmation* after receiving each frame.
From Leonid Mishankov → To All 16 October 1999
Only by asking a friend with Ammy/PC to see what kind of CPS he wants...
P.S. In principle, you can ask to send any file (having previously disabled
V42bis), then:
CPS = V/t, where
V is the volume of the file in bytes
t is the time it was received in seconds.
You will get some approximate value, of course it is difficult to determine very accurately,
fortunately CPS is constantly “floating”.
S!ncerely Yours,
Leonid Mishankov.