>Why Z180 and not Z380? Because Z180 is faster as
>Z80 than Z380 in Z80 compatibility mode. Z180 at 33Mhz has a 3 cycles
>per instruction against 4 of Z80 (actually five, on MSX hardware). This
>makes Z180 at 33Mhz about 11/.714 ~ 15.4 times faster than normal 3.57Mhz
>Z80. Z380 at 18Mhz (the faster it can be) has a 2 cycles per instruction.
>It will give us about 9/.714 ~ 12.6 times faster than normal 3.57Mhz Z80.
>This makes Z180 at 33Mhz something like 22% faster than Z380 at 18Mhz
>(again, we are talking of Z180 working as Z80 and Z380 working as Z80).
There are two points I would like to make here. The first:
The calculations you perform are for a NOP or other simple instructions.
The Z380 is a powerful processor, like the R800 and some instructions are 8
times faster than a Z80 on the same MHz, that is over 30 times faster at
14Mhz! (Which is the speed the LPE-Z380 by mr. Padial runs at)
Also, on many instructions the use of IX/IY or its 8bit parts IXH/IXL does
not come with any speed penalty.
If one uses your reasoning, a Z380 at 14MHz runs as fast as a R800 at 7MHz.
Obviously the R800 is a VERY powerfull processor, since many instructions
can be performed in a single clockcycle, where the Z380 takes 2 or even 3
(for EX instructions). But the Z380 gains speed on the more complex
instructions, so that it's faster anyway.
(BTW, I'm not sure what effect ROM/DRAM mode has on R800, does anybody know
that?)
And the second point is the most important one:
Isn't the whole point of creating a new MSX, to support NEW software?
In that case it's not the performance of the Z80-compatible mode that is
the most important.
It doesn't take a wizard to understand the Z380 is a FAR FAR FAR more
powerfull processor than Z180, even if run on a lower clockspeed, if you
take a look at the extended instruction set (all are instructions the Z180
DOES NOT have):
16 bit loads (LD r16,r16)
16 bit loads from/to memory (LD r16,(r16), LD (r16),n16)
16 bit loads from/to memory with IX/IY or SP(!) index (LD r16,(SP+d))
8 bit EXchange (EX r8,r8)
8 bit EXchange with alternate (EX r8,r8')
16 bit EXchange (EX r16,r16 not just EX DE,HL)
16 bit EXchange with alternate (EX r16,r16' not just EX AF,AF')
16 bit shift/rotate
16 bit block instructions (LDIR/OTIR etc.)
16x16=32 bit MULTiply (7 times faster than R800!)
16/16=16.16 bit DIVide
The list goes on. Now also take in consideration the hugely expanded
register set, there are 4 banks with each AF,BC,DE,HL,IX,IY and shadow
versions (yes IX/IY have shadows too). That makes for: 90 8-bit registers,
or 41 16-bit registers.
The 16-bit registers can be accessed as 32 bit (41 32-bit registers!), and
the Load and Arithmetic instructions can operate on them. This without any
speed penalty! Since the PC is 32bit the Z380 can access up to 4GB RAM
without bankswitching!
Immediate values and addresses can be given in 16, 24 or 32 bit. Also,
IX/IY indexing and SP/PC-relative can be 8, 16 or 24 bit! With the addition
of the CALR (CALL Relative) instruction, you can write truly relocatable
programs, because A JR/CALR can jump up to 8 MEGABYTES forward or backward!
I'm sorry if this message turned into sort of a rant or Z380 advertisment
;) I just want to make clear the Z380 is the more superior CPU, even at a
lower clockspeed. For me, using anything less than a Z380 for the next
generation MSX hardware is NOT an option!
Oh, by the way, that ADVRAM thingy sounds very cool :)
Greetz,
Patriek
****
MSX Mailinglist. To unsubscribe, send an email to [EMAIL PROTECTED]
and put "unsubscribe msx [EMAIL PROTECTED]" (without the quotes) in
the body (not the subject) of the message.
Problems? contact [EMAIL PROTECTED]
More information on MSX can be found in the following places:
The MSX faq: http://www.faq.msxnet.org/
The MSX newsgroup: comp.sys.msx
The MSX IRC channel: #MSX on Undernet
****
