Misunderstandings
ZXNet echo conference «code.zx»
From Aleksandr Majorov → To Ilya Razuev 29 June 2001
Hello Ilya!
28 Jun 01 22:19, Ilya Razuev -> Stanislav Udin:
[skipped]
SU>> processor? What causes the gain in speed?
IR> Previously, there was also an axiom from 1982 -> that there is a program in the lower addresses
IR> works faster (separate memory fields) #0000-#8000 !
Not entirely true ;)
#0000-#3FFF - ROM, it's fast there.
#4000-#7FFF - slow RAM, the video controller is in charge here.
#8000-#FFFF - fast RAM
In the case of 128K, banks 0 - 3 have no brakes, 4 - 7 are slow.
Alexand
From Ilya Razuev → To Aleksandr Majorov 30 June 2001
Greetings, Aleksandr
You're all about Misunderstandings
I decided to add some heat to this dialogue.
SU>>> processor? What causes the gain in speed?
IR>> works faster (separate memory fields) #0000-#8000 !
AM> Not entirely true ;)
Hmmmmm...
AM> #0000-#3FFF - ROM, it's fast.
Ok!
AM> #4000-#7FFF - slow RAM, it rules here
AM> video controller.
Not quite right!!!
AM> #8000-#FFFF - fast RAM
????????????? How ?
AM> In the case of 128K, banks 0 - 3 have no brakes, 4 - 7 -
AM> slow.
You know there was such a company HEWSON!!!!
So she did it like this #5B00-#8000 main program codes
and the rest are sprites and stuff, even read the 40 best procedures
It's written there in black and white!!!!
Even read the reviews for 96-97, people write that
some programs like DIIZY 2, ZINAPS work
unnaturally faster than others!!!!
Well, we will!
.·∙°░▒▓█ Clan BOUNDLESS █▓▒░°∙·.
▀■▄[ZX]√[GIRLS][CvIEW]√[BEER]▄■▀
(500:3432/13.666)
From Aleksandr Majorov → To Ilya Razuev 2 July 2001
Hello Ilya!
30 Jun 01 23:44, Ilya Razuev -> Aleksandr Majorov:
[skipped]
AM>> #4000-#7FFF - slow RAM, it rules here
AM>> video controller.
IR> Not entirely true !!!
AM>> #8000-#FFFF - fast RAM
IR> ??????????????? What?
The company Specke-48 installed two lines of 16K RAM.
(And at the very first - generally 16K current.)
So, although the video controller only uses addresses #4000-#5AFF,
but since the RAM is 16K, then all 16K is slowed down.
And the second line is not touched by the video controller - percent
It doesn't slow down when turning there.
AM>> In the case of 128K, banks 0 - 3 have no brakes, 4 - 7 -
AM>> slow.
IR> You know there was such a company HEWSON !!!!
IR> So she did it like this #5B00-#8000 main program codes
IR> and the rest are sprites and so on
These are her personal problems.
And I myself ran into brake banks.
Initially, in MMD 2.20, the modem driver was in the 4th bank.
In St. Petersburg, machines with separate memory fields somehow
I can't see it - so I didn't check it.
And then someone launched a program on a machine with separate
fields, and the modem connection refused to work - because
Bank number 4 is _slow_. We changed the bank number to 3 - everything worked.In the case of the branded 128, there are two lines of 64K RAM.
The video controller accesses one line and reads the screen
data (bank 5 or 7). And since he accesses the RAM, then the percentage
When I try to access banks 4-7 it slows down.
IR> even read the 40 best procedures
IR> it says !!!! in black and white
Well, just quote what is written there...
Here is the book "System programming for ZX-SPECTRUM",
it is written in black and white that it _cannot_ be placed in register I
values in the range #40-#7F. We'll get a glitch on the branded Speck.
IR> Even read the reviews for 96-97, people write that
IR> some programs like DIIZY 2, ZINAPS work
IR> unnaturally faster than others !!!!
This IMHO has nothing to do with separate memory pages.
I can write you two programs (in the same addresses)
which will work at different speeds.
Alexand
From Valerij Kozhevnikoff → To Stanislav Udin 8 July 2001
Hello, Stanislav!
27 Jun 01 22:46, Stanislav Udin -> All:
SU> Somehow a question came to my mind. There is an axiom that on the Spectrum
SU> a program located in ROM runs faster than one in RAM. Why is this
SU> happens because teams here and there use the same number
SU> processor cycles? What causes the gain in speed?
The ROM does not have to be shared with the video controller; this memory area belongs to
exclusively to the processor, so the whites do not arrive there.
When I connected a 32k cache to Scorp, I was convinced of this. Turbo acceleration
200%. By the way, I was desperately stuffing half of the drivers into this cache, a cloud was freed up
main memory, and it started flying faster and faster.
WBR, Jason.
[Team Obituary][Team Friday 13th][Team Evil Maniacs][Team Huge Axes]