Faster than LDIR...
ZXNet echo conference «zx.spectrum»
Hello, All!
I heard somewhere that there are ways to copy blocks faster than ldir. Is this
possible? If so, how?
See you later...
Denis.
┌ [TEAM Agatha Christie] [TEAM DDT] [TEAM Legends of Rock] [TEAM DOS Navigator] ┐
└─────══════ [AKA 2:4613/204.77, E-Mail to sl...@kot.poltava.ua] ══════──────┘
... Everyone says that we are together. Everyone talks, but not many know which one.
Greetings, Denis!
────────────── ─────────────────────── ─────────────────────── ───────────────────
In ancient times (14 Dec 97), Ser Denis wrote on the scroll All,
Bequeathing “Faster than LDIR...”, and today (16 Dec 97) I decided intervene:
────────────────────────────────── ──────────────────────── ─────────────────────
DI> I heard somewhere that there are ways to copy blocks faster than ldir.
DI> Is this possible? If so, how?
Yes, there is, but everyone has their own situation! And LDIR is universal!
And if you need clearing or filling procedures, then there is even faster!
All the best. /──────────────────────\ [ Fantasy World - Rulez ]
Sincerely, │ Wanderer 'Sagacious' │ [ Getting up early - Must Die]
─────────────────── ────\──────────────────────/ [ Station 106.8 - Rulez ]
... It’s hard in the village without a revolver ...
Denis is that you?
...why did you tell All about Faster than LDIR... ?
DI> I heard somewhere that there are ways to copy blocks faster than ldir.
DI> Is this possible? If yes, then how?
....
LDI
LDI
...many LDI...
this is the first method.
the second is
POP HL
POP DE
...all registers - POP.
SP->new address
...all registers - PUSH
PUSH DE
PUSH HL
i.e. block transfer is done something like this: put the stack at the beginning, and load all the registers via POP XX. Then you rearrange
the stack to the place where it is needed, and do PUSH in the reverse order.
p.s. I hope there is no need to explain why this is so? :)
Good Luck...
... DEFINITION: Computer - A device designed to speed and automatic errors.
Hello, Denis!
It was like this: Denis Itskovich 14 Dec 97 at 15:28 was pushing speeches to All.
Well, I wrote it.
DI> I heard somewhere that there are ways to copy blocks faster than ldir. Is this
DI> possible? If yes, how?
Yes. LDI, but it's more in bytes. You can use a stack, but this is not always suitable.
╔══─══─═─═─── ── ─ ─ -
║ Max (aka LMS/Inside Law) [TEAM SPECCY rOOleZZZZ]
╚══─══─═─═─── ── ─ ─ ─
DI> Is this possible? If so, how?
For example, several LDIs (numbered according to the size of the block being copied).
[skipped]
Happy connections.
Roman.
18 Dec 97, already 00:19
AH> LDI
AH> LDI
AH> LDI
AH> ...
And why will it be faster?
See you later...
Denis.
─────══════ [ E-Mail to sl...@kot.poltava.ua ] ══════──────
VS> Yes, there is, but everyone has their own situation! And LDIR is universal!
Well, for example, I need to copy a block of 1536
bytes from address 64000 to 16384. Is there a faster method than LDIR?
VS> And if you need clearing or filling procedures, then there is another
VS> faster!
Well, which one is for cleaning?
RA> For example, several LDIs (numbered by the size of the block being copied).
You can loop several of them, but then the size
of the block must be the same to get something like this:
loop
ldi
ldi
ldi
ldi
jp pe,loop
for this example 18.5 clock cycles per byte, instead of 21 clock cycles with ldir
but the number of bytes must be a multiple of four...
you can, of course, use a stack... the maximum is 10.5 clock cycles, but
de-cranching is needed...
ld hl, <- two bytes to be transferred
push hl
...
if without de-cranching then 12.5 clock cycles per byte:
ld sp,source
pop af,bc,de,hl
exx
pop bc,de,hl
ld sp,destonation
push hl,de,bc
exx
push hl,de,bc,af
imho nowhere faster :(
▌▌║▌█▐│▌▌▐▐ WiTh The BeST wIsheS fROM
▌▌║▌█▐│▌▌▐▐ *C*R*E*A*T*O*R*
▌812/03.00▐
... the glitch is not a sparrow, it flew out - kill it!
-+- SMM version 1.05
Hello Max.
16 Dec 97 21:16, Max Romanoff wrote to Denis Itskovich:
[...]
MR> i.e. block transfer is done something like this: you put the stack at
MR> at the beginning, and load all the registers via POP XX. Then you rearrange
MR> the stack to the place where it is needed, and do PUSH in the reverse order.
Yesss! This is exactly what they call it - coding with gusto ;)
A really cool option;)
Oleg AKA Olav
... A beautiful lie? Attention! This is already creativity.
DI> And why will it be faster?
LDIR spends 21 clock cycles to transfer one byte, LDI - 16.
Best wishes! Michael KLUG
DI> Well, which one is for cleaning?
In SP you place the address of the end of the block that needs to be filled, plus two; in HL, DE
or BC you put two bytes of code that need to be filled, or 0 to clear, and
you do as many PUSH as you need.
LD SP,#addr_end
LD HL,0
PUSH HL
PUSH HL
PUSH HL
...
Best wishes! Michael KLUG
See you later...
Denis.
─────══════ [ E-Mail to sl...@kot.poltava.ua ] ══════──────
... Everyone says that we are together. Everyone talks, but not many know which one.
MR> ....
MR> LDI
MR> LDI
MR> ...many LDI...
MR> this is the first way.
MR> the second is
MR> POP HL
MR> POP DE
MR> ...all registers are POP.
SP->> new address
MR> ...all registers are PUSH
MR> PUSH DE
MR> PUSH HL
DI> ldir. Is this possible? If so, how?
LD HL,SOURCE 10 3
LD DE,DESTIN 10 3
LD BX,COUNT 10 3
LDIR (21*COUNT)-5 2
total 21*COUNT+25 clock cycles, 11 bytes
LD SP,SOURCE 10 3
POP HL 10 1
LD (SOURCE),HL 16 3
POP HL 1
LD (SOURCE+2),HL 3
...
POP HL 1
LD (SOURCE+COUNT/2-2),HL 3
total 13*COUNT+10 clock cycles, 2*COUNT+3 bytes
When copying the screen LDIR=145177(11), STACK=89866(13827)
(1.6 times faster and 1257 times longer code)
Best regards
┬ ┬┌┐┬┐ ┐┬┬
╨╜╨╙╜╨╙ ╨ ╨
DI> Well, for example, I need to copy a block of size
DI> 1536 bytes from address 64000 to 16384. Is there a faster way than LDIR?
DI> It was a long time ago... 16 Dec 97, the clock suddenly struck 15:12, my radar
DI> turned towards 2:5032/7.44, and I spotted: Artem Kharitonov said a word
DI> to Denis Itskovich about Faster than LDIR....
OB> Yesss! This is exactly what they call it - coding with gusto ;)
OB> A really cool option;)
Yeah. Spec's coder goes to great lengths to achieve stunning results. :) And they sometimes succeed.
Any PiTyushnogo coder who is more or less familiar with the Spekovsky world will say that the Specovsky coder is several times better than PiTyushnog. I heard it myself.
DI> Let me explain: I need to transfer a block of ~2048 bytes in size. How can this
DI> be made into a subject?
Who's stopping you from doing a series of POP and PUSH in a loop, and even using
the second set of registers?
Best wishes! Michael KLUG
ML> Yeah. Spec's coder goes to great lengths to achieve stunning
ML> results. :) And sometimes they succeed.
You also tell the people how to climb tablets in a stack and display sprites
. This is the real thrill, and cleaning is so primitive...
ML> Any PiCyushnogo coder who is more or less familiar with the Spekovsky world
ML> will say that the Specovsky coder is several times better than PiCyushnog. I heard it myself.
And I was convinced of this myself. But this is already off-topic.
I wish you happiness and... good luck.
Andrey.
My respect, Denis!
Not so long ago Denis Itskovich did not consider it difficult to write Max Romanoff
(I even remembered the date and time: 19 Dec 97 at 08:18):
DI> I explain: I need to transfer a block of ~2048 bytes in size. How can
DI> make this a subject?
Have you not moved in? Set ~1024 POP and ~1024 PUSH...
Really cool, but this is already a classic. Max Iwamoto only relied on his skillful
construction of such jokes and held on.
It is worth noting the following: POP and PUSH, in addition to the fact that they work faster, also
with two bytes at once.
It makes sense to expand the cycles
It makes sense to delve into the pearls of the aforementioned Max Iwamoto and above Rst Seven, despite the fact that Max's patented protection (the last one) is difficult to break (STS saved me), so if you don't know the protections, it's worth delving into his older programs.
Good bye. Vadim
because LDI, unlike LDIR, does not decrement and check
the counter.
Good Luck...
... E Pluribus Modem
Denis is that you?
...why were you telling Vitaly Samoylov about Faster than LDIR... ?
DI> Well, for example, I need to copy from address 64000 to 16384 a block of size
DI> 1536 bytes. Is there a faster way than LDIR?
I wrote that either LDI or a stack...
DI> Well, which one for cleaning?
do you know how to clean the screen with LDIR? Is the analogy visible?
Good Luck...
... Car service: If it ain't broke, we'll break it.
SH> Yes, there is, but each to their own situation! And LDIR is universal!
SH> And if you need clearing or filling procedures, then it’s even faster!
If memory serves, the fastest way to clean the screen is this way. Stack on the screen and
PUSH PUSH PUSH..... LDIR is faster in... I don’t want to count right now it’s 3.34
nights and mail 4 meters has arrived, I need to clear it :-(.
Sincerely Shor Kan.
(2:5078/22.9, Sun December 21 1997 02:22 )
Because LDI runs in 16 clock cycles, and LDIR in 21 clock cycles.
Moreover, the advantage becomes clearly noticeable on relatively large blocks.
PS. Does my soap come to you? I hesitated to write 8-E~~~
Denis is not intelligence officer Isaev at all, but Duke Nukem!..
Ilya /STRG, December 18, 97 AD, 00:33.
... Do you want CHEAT (Y/N)? ..But no
ML> heard.
:) It’s better to see than to hear 100 times. Look at INSULT or SATISFACTION and see what people did at 3.5 MHz. And on PC everyone is just lazy:
Why optimize the demo if instead of 386SX/33 you can scratch 486DX4/133?
You would be in 1941... ;)
Good bye. Vadim
ML>> Yes. LDI, but it's more in bytes. You can use a stack, but this is not always
ML>> suitable.
DI> ^^^^^^
DI> Yes, I understand. But I read that there are ways to copy _large_
DI> blocks, subject
LDIR is the only block copy command on the Z80, and a lot can be transferred
a little at a time - if you need it quickly, then in a stack of 2 bytes in an expanded loop -
it will fly.
Good bye. Vadim
Well, for example, through the “PUSH” command, you put the stack at the highest address and forward!
You fill two bytes of memory in 11 clock cycles.
DI> Denis.
All the best. /──────────────────────\ [ Fantasy World - Rulez ]
Sincerely, │ Wanderer 'Sagacious' │ [ Getting up early - Must Die]
─────────────────── ────\──────────────────────/ [ Station 106.8 - Rulez ]
... It’s hard in the village without a revolver ...
Denis is that you?
...why were you telling me about Faster than LDIR... ?
DI> Yes, that’s understandable, but it will be faster if you transfer a very short
DI> block.
it will be faster in any case. If you need to make a long block
, you can make a loop - that’s once, or you can make a reversal
of the code.
DI> I’ll explain: I need to transfer a block of ~2048 bytes in size. How can
DI> be done with a subject?
either 2048 LDI or a stack. If it's a stack, then do a reversal, or
use one of the registers as a counter.
Good Luck...
... CanaDOS: Yer sure, eh? (B)eauty! (N)o way! (T)ake off!
Denis is that you?
...why were you telling Max Lanskih about Faster than LDIR...?
DI> Yes, I understand. But I read that there are ways to copy _large_ blocks,
DI> subject.
damn, well, are you a programmer, or what?!
Make a loop:
LD HL,FROM
LD DE,TO
LD C,128
loop:
LDI
LDI
LDI
LDI
LDI
LDI
LDI
LDI
DJNZ LOOP
rts
this piece copies 1024 bytes from HL to DE. If you make LDI=shek more
and reduce the cycle accordingly, it will be even faster.
Mlyn, can't you think of that?
Good Luck...
... VirusScan: MS Windows found.
Michael is that you?
...why did you tell Denis Itskovich about Faster than LDIR... ?
MR>> start, and load all registers via POP XX. Then you rearrange
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
MM> And who’s stopping you from doing a series of POP and PUSH in a loop, and even with
MM> using the second set of registers?
Mister, it looks like you’re reading doesn’t know how, or think... everything chewed up
wants.
Good Luck...
... After a few years in space, even Worf started to look good...
Andrey is that you?
...why did you tell Max Lanskih about Faster than LDIR... ?
AT> You also tell people how to climb tables in a stack and display sprites
AT>. This is the real thrill, and cleaning is so,
AT> primitive...
sprites in a stack - this is a fairy tale! And the signs are the best!
Good Luck...
... If hackers ran the world there'd be no war. A lot of accidents, maybe...
DI> Let me explain: I need to transfer a block of ~2048 bytes in size. How can
DI> be made a subject?
if the block is static, then
LD SP,WHERE
LD HL,n1
PUSH HL
LD HL,n2
PUSH HL
...
LD HL,n1024
PUSH HL
10.5 clock cycles/byte. It doesn't get any faster. if the block is not static and is memory critical, then
LD B,n
L1: PUSH BC
LDI
LDI
LDI
...
LDI
POP BC
DJNZ L1
n and the number of ldi-sheks to taste. will be slightly more than 16 clock cycles/byte. There are more
options, faster - through push/pop and pop hl/ld (nn),hl, but they need to be promoted like
first.
──══─════┤ *_Jeweller_* aka EX-VfNG/NEW at home. [Echo], Brest, RB. ├════─══──
Hello Max.
21 Dec 97 20:29, Max Romanoff wrote to Michael Markowsky:
[...]
MR> Mr., it seems he doesn’t know how to read, or think... he wants everything chewed up
MR>.
Well, why is it so cruel, maybe a person is just learning... :)
Oleg
... Often a roof over one’s head interferes with people grow.
Oleg is that you?
...why were you telling me about Faster than LDIR... ?
OB> Well, why be so cruel, maybe a person is just learning... :)
yes, it’s never too late to learn, just don’t slow down unnecessarily in
this.
Good Luck...
... Madness takes it toll. Please have exact change.
Hello, Max!
It was a long time ago... Dec 20, 97, the clock suddenly struck 23:36, my radar turned towards
2:5030/221.16@fidonet, and I spotted: Max Romanoff says a word to Denis
Itskovich about Faster than LDIR....
MR> Denis is that you?
MR> ...why did you tell Artem Kharitonov about Faster than LDIR... ?
See you later...
Denis.
─────══════ [ E-Mail to sl...@kot.poltava.ua ] ══════──────
... Everyone says that we are together. Everyone talks, but not many know which one.
Denis is that you?
...why were you telling me about Faster than LDIR... ?
DI> Well, that goes without saying... And if I copy in a loop via LDI, will it
DI> be faster?
If you install more LDI, then yes.
Good Luck...
... Backup not found: (A)bort (R)etry (T)hrowup
DI> OK. Please, someone give me a list of commands with a strategy.
I have... I have this in my book... I think someone will throw it in.
Good Luck...
... Unix: the operating system of the future of the past.
Max, you've probably become a complete navel on the Amiga, LDI does
decrement BC and checks it for zero! result in
overflow/parity flag. if you don't believe it, check it out.
▌▌║▌█▐│▌▌▐▐ WiTh The BeST wIsheS fROM CREATOR product
▌812/03.00▐ Powered by Cyrix 6x86L 200+/16DIMM/S3TrV2 2MB/1.6GB
... the glitch is not a sparrow, it crashed - kill it!
-+- SMM version 1.05
Hi Max !
DI>> Yes, I understand. But I read that there are ways to copy _large_ blocks,
DI>> subject.
MR> damn, are you a programmer, or what?!
MR> Make a loop:
MR> LD HL,FROM
MR> LD DE,TO
MR> LD C,128
what are you from 040 overheated? need to be in BC,1024
MR> loop:
^this is a spec, don’t forget.
MR> LDI
MR> LDI
MR> LDI
MR> LDI
MR> LDI
MR> LDI
MR> LDI
MR> LDI
MR> DJNZ LOOP
register B you are not installed, LDI decrements BC
need JP PE,LOOP
MR> rts
MR> this piece copies 1024 bytes from HL to DE. If LDI=shek is more
MR> do it and reduce the cycle accordingly, then it will be even faster.
MR> Damn, can’t you think of that?
with such ideas, a person will not understand what you wrote at all.
MR> rts
I'm having fun! And where is register B set for the DJNZ command?
And how can you do this when the LDI command makes BC <- BC-1 ?
Showing:
LD HL,FROM
LD DE,TO
LD BC,1024 ; the number must be divisible by 8, otherwise the program will go into a loop
Hello Vladyslav!
22 Dec 97 22:21, Vladyslav Yakubovich wrote to Denis Itskovich:
VY> if the block is static, then
VY> LD SP,WHERE
VY> LD HL,n1
VY> PUSH HL
VY> LD HL,n2
VY> PUSH HL
VY> ...
VY> LD HL,n1024
VY> PUSH HL
VY> 10.5 clock cycles/byte. It doesn't get any faster.
It happens, because the block is static, which means you can
form a forwarder based on its contents:
If two words are the same, then in the second LD HL,nn is not needed
if one byte is different, then you need LD H,n (L,n) and so on.
Of course, these are all special cases, but very common.
Yours sincerely
┬ ┬┌┐┬┐ ┐┬┬
╨╜╨╙╜╨╙ ╨ ╨
VY> It happens, because the block is static - which means you can
VY> form a transfer agent based on its contents:
yes, I forgot to mention. for some reason I had a completely different block spinning around in my head
:) thanks for correcting me.
zy:
by the way, about your emyl: I have the feeling that sometimes the whole
channel doesn’t play, and sometimes there’s no envelope :( never looked at it. card - OPTI931. on
my monitor (Samsung SyncMaster 3NE) at 49HZ the screen twitches, and at 53HZ a stripe is visible in the middle of the screen. If you did all this in XMODE and you have more than 360 points on X, then I’m not surprised :( maybe it would be worthwhile to provide support for SVGA? and the rest is emul - just lovely.
zyy:
I guess you were interested in information on programming in PMODE? Then go ahead.
──══─═══┤ *_Jeweller_* aka EX-VfNG/NEW at home [Echo], Brest, RB. ├═══─══──
IM> I'm having fun! And where is register B specified for the DJNZ command?
I’m also crazy, but I just showed with an example how you can send
a lot. The idea is clear. Of course, it’s been a long time since I remember all the subtleties of the Z80, because... I don’t work with him.
IM> And how can you do this when the LDI team is doing BC <- BC-1?
how, how, how to top! This was an example.
Good Luck...
... DOS Tip #17: Add DEVICE=FNGRCROS.SYS to CONFIG.SYS
PF> Max, you've probably become a complete navel on the Amiga, is LDI making you annoyed? Became and became.
PF> decrement BC and checks it for zero! result in flag
PF> overflow/parity. If you don't believe me, check it out in sts'e .
yes? Well, Lana. however, LDI works faster.
Good Luck...
... Hidden DOS secret: add BUGS=OFF to your CONFIG.SYS
PF> with such ideas, a person will not understand what you wrote at all.
ideas are just right. it just shows the idea of what needs to be done _CYCLE_.
p.s. Damn, am I a megabrain to write bugless programs by mail? :)
Good Luck...
... All wiyht. Rho sritched mg kegtops awound?
Get Msg, Paul!
25 Dec 97 00:00, Paul Falcon reported to Max Romanoff about {Faster than LDIR...}:
PF> Max, you've probably become a complete navel on the Amiga, LDI does
PF> decrement BC and checks it for zero! result in flag
êëíçç Û Hœ◊ŸÕ ÁœƒœÕ, Max ! ççíëê
çíëê êëíçíëê êëíçíëê êëíçíëê êëíçíëê êëíçíëê êëíçíëê êëíçíëê êëíçíëê
ç Îœ«ƒ¡-‘œ, Tuesday December 30 1997 00:38, Max Romanoff
ç –…”¡Ã ƒÃ— Ivan Mak:
MR> ...⁄afi≈Õ ‘Ÿ “a””Àa⁄Ÿ◊aà Ռ≈ –“œ ‚Ÿ”‘“≈≈, fi≈Õ LDIR... ?
...
MR> — ‘œ÷≈ ¬aÃ≈¿, a –“œ”‘œ –œÀa⁄aà Œa –“…Õ≈“≈, ÀaÀ Õœ÷Œœ –≈“≈”Ãa‘ÿ
MR> ÕŒœ«œ. ȃ≈— –œŒ—‘Œa. “a⁄’Õ≈≈‘”—, — ƒa◊Œœ ’÷≈ Œ≈ –œÕŒ¿ ◊”≈»
MR> ‘œŒÀœ”‘≈ Z80, ‘.À. ” Œ…Õ Œ≈ “a¬œ‘a¿.
˜≈”ÿ ∆œÀ’” ◊ ‘œÕ … ¬ŸÃ, fi‘œ fi≈Ãœ◊≈À ”–p¡€…◊¡Ã –pœ _ÙÔHÎÔÛÙÈ_ Z80.
MR> Good Luck..
...
êëíçç Û ’◊¡÷≈Œ…≈Õ, Ivan. ççíëê
TR-DOS 5.04Ei
Get Msg, Max!
30 Dec 97 00:38, Max Romanoff reported to Ivan Mak about {Faster than LDIR...}:
MR> a lot. The idea is clear. Of course, I no longer remember all
MR> subtleties of the Z80, because... I don’t work with him.
You can’t forget what you didn’t know :))
Warm greetings, Slava!
IM> The whole trick was that the person asked about the _THINKS_ of the Z80.
no. They asked how you can send a block faster than
LDIR. It was answered. It was asked again about how to send a block, not 1-2 bytes. there was a figurative example that prompted the idea that it is necessary
to do a tsitsk.
maybe I should also write programs for it? :)
Good Luck...
... CONGRESS.SYS Corrupted: Re-boot Washington D.C (Y/n)?
VM> You can’t forget what you didn’t know :))
Don’t just try to make jokes here, OK? From time to time
I ran into similar cases, and multiplied the number of passes by
the number of ldi. I just forgot about it now, as it is not needed
in the household.
Good Luck...
... Help Wanted: Telepath. You know where to apply.
MR> Õœ÷≈‘ ÕŒ≈ ⁄a Œ≈«œ ≈›£ … –“œ«“aÕÕŸ –…”a‘ÿ? :)
Îá””Œ¡— …ƒ≈— ! Hœ p¡¬œ‘¡‘ÿ ‘œfiŒœ Œ…fi≈«œ Œ≈ ¬’ƒ≈‘ :(
IM> Cool idea! But definitely nothing will work :(
of course, I’ll look at you as you write useless programs in a text editor for a soaper
...
p.s. damn, why didn’t you answer the question FIRST? Why show off your
erudition now?
MR> ≈›£-¬Ÿ, –œ”Õœ‘“¿ — Œa ‘≈¬—, ÀaÀ ‘Ÿ ◊ ‘≈À”‘œ◊œÕ “≈ƒaÀ‘œ“≈ ƒÃ— ÕŸÃ≈“a
MR> ¬’ƒ≈€ÿ –…”a‘ÿ ¬≈⁄¬a÷ŒŸ≈ –“œ«“aÕÕŸ...
Ï…fiŒœ — …» –…”¡‘ÿ … Œ≈ ¬’ƒ’. Ìœ≈ «Ã¡◊Œœ≈ ƒ≈Ãœ - ÷≈Ã≈⁄œ.
MR> –.”. ¬Ã…Œ, –œfi≈Õ’ ‘Ÿ ÂP˜˘Í Œ≈ œ‘◊≈‘…à Œa ◊œ–“œ”? ˙afi≈Õ ”≈ fia” ”◊œ¿
MR> ‹“’ƒ…√…¿ ƒ≈ÕœŒ”‘“…“œ◊a‘ÿ?
Ò fiœ -‘œ ’÷≈ ⁄¡¬ŸÃ, œ fi≈Õ ∆Ã≈ Õ.. :)
MR> Good Luck...
IM> Personally, I won’t write them. My main business is iron.
Then you won’t pop up like that anymore? :)
IM> I somehow already forgot what the flame was about.. :)
excused myself.
Good Luck...
... Beauty is in the eye of the beer holder...
░▒▓██ I am glad to welcome you, Max! ██▓▒░
█▓▒░ ░▒▓█▓▒░ ░▒▓█▓▒░ ░▒▓█▓▒░ ░▒▓█▓▒░ ░▒▓█▓▒░ ░▒▓█▓▒░ ░▒▓█▓▒░ ░▒▓█▓▒░
█ Once upon a time, Saturday January 10 1998 00:58, Max Romanoff
█ wrote for Ivan Mak: