N64 Emulator Controversy
pri...@flash.net
Yes, PC's are much too slow. It was calculated that a Macintosh (Which is
proven to be much faster in the graphics department than a PC, no offense
to PC's or anything) that is 300mhz (Or maybe it was 500MHz) could run an
N64 emulator, but it would be so choppy it wouldn't be worth playing it.
And the programmers who do emulators aren't lazy, they just don't feel the
need to do something so worthless. They would program one if it would
work. You don't need to blame it on them, they are the ones that brought
the SNES, NES, emulaters (and the others) to you in the first place. If
you want an N64 Emu, get off YOUR LAZY ass and learn to program.
Blake
(Please do not get in any PC v. Mac argument. This post was not meant to
start one. I have nothing against either. And those arguments are stupid.
They both have advantages. We can guess most of them. We don't need anyone
telling us them for the millionth time.)
Jaakan Shorter
pri...@flash.net wrote in message ...
>Yes, PC's are much too slow. It was calculated that a Macintosh (Which is
>proven to be much faster in the graphics department than a PC...
That is half true, but the Video Cards( 3DFX based,etc.. ) will be doing the
3D stuff.
-At under $2500, I would call a DEC Alpha 500 a PC.
-For what a Mac can do, Why isn't there a port of Softimage3D for them.
ttammi
pri...@flash.net wrote:
>Yes, PC's are much too slow. It was calculated that a Macintosh (Which
>is proven to be much faster in the graphics department than a PC, no
>offense to PC's or anything) that is 300mhz (Or maybe it was 500MHz)
>could run an N64 emulator, but it would be so choppy it wouldn't be
>worth playing it.
Was this Macintosh with or without a 3D(fx) card? BIG difference.
True, it would take awful lot of CPU time if EVERYTHING of N64 was
emulated on it, but the 3D card helps a lot with the "emulation" of
N64's 3D graphics effects.
Jason Rold
3Dfx cards are nice but the N64 sports some features the 3Dfx doesn't have,
bi-linear filtering vs. tri-linear filtering ect... I don't expect to see
any working playable N64 emulators for high-end consumer computers this
century.
ttammi <tta...@netlife.fi> wrote in article <VA.000007d0.002ed5e5@tlt>...
Ronnie Deschenes
OK, the problem with your argument is that emulators are supposed to make
one kind of hardware do something that another piece of hardware does.
Maybe the tri-linear filtering and anti-aliasing wll be handled by the
emulator... Of course, it'd be slow, but nonetheless...
--
Ronnie Deschenes -=- dm...@freenet.carleton.ca
Handler of renowned e-wrestler "Playboy" Ronnie D #32 in the 1997 RSPWF 200
Playboy's page: http://www.ncf.carleton.ca/~dm870/
As soon as they idiot-proof the world, a better idiot will come along.
ttammi
"Jason Rold" <jr...@NOSPAMtstonramp.com> wrote:
>3Dfx cards are nice but the N64 sports some features the 3Dfx doesn't
>have, bi-linear filtering vs. tri-linear filtering ect.
Wrong. 3Dfx Voodoo supports trilinear filtering, 3Dfx calls it
"advanced texture filtering" on their homepage (Gary Tarolli from 3Dfx
has confirmed that it means trilinear filtering).
Anyway, I don't think there is even one existing N64 game that uses
trilinear filtering anyway, because it halves the pixel fill rate of
N64 (half the speed for little more graphical enhancement, not worth it
in game developers' eyes).
>.. I don't expect to see
>any working playable N64 emulators for high-end consumer computers
>this century.
No one expected to see MAME or SNES9x either. I'll give it a benefit
of doubt. _If_ they were trying to emulate the whole N64 on a
Intel CPU only, I would be very sceptical as well. Now that they
are going to utilize PC 3D cards, I feel they at least have
a chance.
ttammi
dm...@FreeNet.Carleton.CA (Ronnie Deschenes) wrote:
>OK, the problem with your argument is that emulators are supposed to
>make one kind of hardware do something that another piece of hardware
>does. Maybe the tri-linear filtering and anti-aliasing wll be handled
>by the emulator... Of course, it'd be slow, but nonetheless...
First of all, 3Dfx Voodoo _does_ support trilinear texture filtering.
Also, 3Dfx Glide API supports polygon edge antialiasing as well, but
currently 3Dfx cards take a considerable speed hit from using it.
Anyway, if the author of the emulator found out that emulating e.g.
edge AA would be to slow on 3Dfx, he could simply forget about it
altogether. Like, you know, simply pass the instructions where edge AA
is applied? Where's the problem? Sure the N64 emulator games would have
jaggy polygon edges after that, but a) it wouldn't slow it down
b) it isn't like the that would destroy the games, it would only make
the N64 games that use edge AA (both of them) look a bit worse.
For example, it's not like current SNES emulators support ALL
SNES graphical tricks either, yet they work fine for most
SNES games.
AutismUK
In article <VA.000007db.000768bd@tlt>, ttammi <tta...@netlife.fi> writes:
>No one expected to see MAME or SNES9x either. I'll give it a benefit
>of doubt. _If_ they were trying to emulate the whole N64 on a
>Intel CPU only, I would be very sceptical as well. Now that they
>are going to utilize PC 3D cards, I feel they at least have
>a chance.
You ought to look more closely at the R4300. It is not simply
a matter of rendering the graphics to the display.
Also, it is not accurate to say "no one expected to see
MAME or SNES9x".
I remember 14-15 months ago there were (I think) three
SNES emulators about (one of which was ESNES) & the main
missing component was sound (no-one knew how it worked).
It has a CPU with a <4Mhz Clock Speed and is relatively
simple.
MAME is a lot of arcade emulators glued together under one
roof, so to speak. Whilst this was not 'expected' there were a
fair few individual arcade machine emulators about (e.g. one for
Pacman, one for Galaxians etc. etc.), which MAME compressed
into one program. I believe most of its emulated processors are
things like 6502,6809,Z80,8086 etc.
Andreas Koslowski
Ronnie Deschenes wrote:
>
> OK, the problem with your argument is that emulators are supposed to make
> one kind of hardware do something that another piece of hardware does.
> Maybe the tri-linear filtering and anti-aliasing wll be handled by the
> emulator... Of course, it'd be slow, but nonetheless...
Of course emulation would be also working fine without anti-alias and
all the other tricky gfx stuff. But even emulating the N64 CPU at 90 MHz
is far too much for today's high end-machines. The current emulated
speed is about 1/10 of the original on faster x86 machines.
ttammi
auti...@aol.com (AutismUK) wrote:
>>No one expected to see MAME or SNES9x either. I'll give it a benefit
>>of doubt. _If_ they were trying to emulate the whole N64 on a
>>Intel CPU only, I would be very sceptical as well. Now that they
>>are going to utilize PC 3D cards, I feel they at least have
>>a chance.
>You ought to look more closely at the R4300. It is not simply
>a matter of rendering the graphics to the display.
Nowhere did I say so. But the fact that they can use 3D
accelerators for the emulator does take a lot of the burden
off the CPU.
>Also, it is not accurate to say "no one expected to see
>MAME or SNES9x".
It isn't that long ago when some people still claimed PCs cannot do
even natively what SNES can, let alone successfully emulate it.
>I remember 14-15 months ago there were (I think) three
>SNES emulators about (one of which was ESNES) & the main
>missing component was sound (no-one knew how it worked).
>It has a CPU with a <4Mhz Clock Speed and is relatively
>simple.
And the emulators do it all on the x86 CPU, unlike N64 emulator.
The question was whether the could be a working and useful N64 emulator
this century. With the 3D accelerators and new Intel CPUs (Merced and
whatever), who knows, maybe? The individuals who think they know it
is impossible are the same ones who thought SNES and Callus emulators
would be impossible on Pentiums too.
AutismUK
In article <VA.000007df.008ffd71@tlt>, ttammi <tta...@netlife.fi> writes:
>>You ought to look more closely at the R4300. It is not simply
>>a matter of rendering the graphics to the display.
>Nowhere did I say so. But the fact that they can use 3D
>accelerators for the emulator does take a lot of the burden
>off the CPU.
No, you're missing the point. A 266 Mhz CPU emulating a
93 Mhz CPU gives you say 3-5 instructions per emulated instructions,
giving a bit for pipelining and caching. Given the fact that the R4300 is
64 bit, and the Intel architecture is 32 bit there isn't much chance of
emulating the functionality.
>>Also, it is not accurate to say "no one expected to see
>>MAME or SNES9x".
>It isn't that long ago when some people still claimed PCs cannot do
>even natively what SNES can, let alone successfully emulate it.
Natively ? What you mean do a few bitmap layers and a few sprites ?
Probably PCs faster than 286s could do this, given competent coding.
Who are these 'people' ?
>>I remember 14-15 months ago there were (I think) three
>>SNES emulators about (one of which was ESNES) & the main
>>missing component was sound (no-one knew how it worked).
>>It has a CPU with a <4Mhz Clock Speed and is relatively
>>simple.
>And the emulators do it all on the x86 CPU, unlike N64 emulator.
And the CPU is 30 times faster, and the custom chips are more
complex, being virtual processors in their own right. It is true that
given sufficient compatibility between N64 3D and 3DFx 3D some
power could be reclaimed for CPU emulation but not enough.
>The question was whether the could be a working and useful N64 emulator
>this century. With the 3D accelerators and new Intel CPUs (Merced and
>whatever), who knows, maybe? The individuals who think they know it
>is impossible are the same ones who thought SNES and Callus emulators
>would be impossible on Pentiums too.
Well, they aren't in my case. You should learn more before entering this
sort of discussion. Nobody believes claims Intel make for speed, delivery
etc. any more. The claims for MMX were crap. The Pentium 2 is just a
hardware redesign to get around the problems of making a PPro.
Also, emulators don't respond to caching too well - though the code
being emulated may involve tight loops etc., the emulation code itself
is a bit like a big case statement...
Martin Korolczuk
>>>You ought to look more closely at the R4300. It is not simply
>>>a matter of rendering the graphics to the display.
>>Nowhere did I say so. But the fact that they can use 3D
>>accelerators for the emulator does take a lot of the burden
>>off the CPU.
>
>No, you're missing the point. A 266 Mhz CPU emulating a
>93 Mhz CPU gives you say 3-5 instructions per emulated
>instructions, giving a bit for pipelining and caching. Given
>the fact that the R4300 is 64 bit, and the Intel architecture
>is 32 bit there isn't much chance of emulating the functionality.
Can you tell me WHAT is 64-bit in the R4300? The 64-bit floats
are nothing compared to 80-bit PC doubles. Something else?
The bus is 64-bit? Nothing of so difficult or time-consuming onto
a PC, especially with a quick AGP 3D card... (Since the N64
doesn't have its own big texture memory as the 3dfx has, an
AGP card will be needed.)
>Also, emulators don't respond to caching too well - though
>the code being emulated may involve tight loops etc., the
>emulation code itself is a bit like a big case statement...
The emulated code should be compiled at run-time and then
you'll spend less than 10 cycles per emulated instruction. The
N64 has a tens times smaller cache than a PII (16+16 KB of
L1 cache and 512 KB of very quick L2 cache). And before the
end of the century (January 1st, 2001), Intel will release 600
(or more) MHz PII and some IA-64. Some very quick 3D AGP
cards will help N64 emulators' authors to speed up their
programs too.
So, first good N64 emulators will appear before the end of this
century on high-end machines.
Jason L. Gurtz
>>>>You ought to look more closely at the R4300. It is not simply
>>>>a matter of rendering the graphics to the display.
>>>Nowhere did I say so. But the fact that they can use 3D
>>>accelerators for the emulator does take a lot of the burden
>>>off the CPU.
>>
>>No, you're missing the point. A 266 Mhz CPU emulating a
>>93 Mhz CPU gives you say 3-5 instructions per emulated
>>instructions, giving a bit for pipelining and caching. Given
>>the fact that the R4300 is 64 bit, and the Intel architecture
>>is 32 bit there isn't much chance of emulating the functionality.
>
>
>Can you tell me WHAT is 64-bit in the R4300? The 64-bit floats
>are nothing compared to 80-bit PC doubles.
Another point to note is the next generation of proccessers
from Intel will be 64 bit, not to mention that you can get a 64-bit
DEC Alpha for under $3000 right now and run NT on it as well
as Unix. (Of course, performance would be WAY better under
a native Unix)
Something else?
>The bus is 64-bit? Nothing of so difficult or time-consuming onto
>a PC, especially with a quick AGP 3D card... (Since the N64
>doesn't have its own big texture memory as the 3dfx has, an
>AGP card will be needed.)
Also, AGP is currently only using 1/4 of it's possible bandwidth.
The disscussion on this topic is lengthy, so, point your browser
to http://www.tomshardware.com/agp.html.
>
>>Also, emulators don't respond to caching too well - though
>>the code being emulated may involve tight loops etc., the
>>emulation code itself is a bit like a big case statement...
>
>
>The emulated code should be compiled at run-time and then
>you'll spend less than 10 cycles per emulated instruction. The
>N64 has a tens times smaller cache than a PII (16+16 KB of
>L1 cache and 512 KB of very quick L2 cache).
Also keep in mind that current Socket 7 boards can be equiped
with a 1 Meg L2 cache if it is a VP2 or VP3 chipset from VIA.
Also, current MB's for the Alpha proc. have not only a VERY
large L1 and L2 cache (bigger than Intel), arranged incedentaly
not unlike the PII L2 cache, but also has a huge L3 cache
And before the
>end of the century (January 1st, 2001), Intel will release 600
>(or more) MHz PII and some IA-64. Some very quick 3D AGP
>cards will help N64 emulators' authors to speed up their
>programs too.
>
>So, first good N64 emulators will appear before the end of this
>century on high-end machines.
>
Also worth noting is the fact that it has been proven that EDO
RAM can and does work at above its rated speed of 66Mhz
Many people have in fact gotten it to work on an 83Mhz bus.
So, what is one to think about the "rated" speed of SDRAM?
Manufacturers insist that it won't work at over 100Mhz, but, they
said the same thing about EDO and 66Mhz.... Now all we
need is a board that supports SDRAM AND a bus speed
100Mhz and above(next logical speed is 133Mhz). Intel,
awhile back, anounced that they were forming a partnership
with NEC to develop nDRAM, able to handel a freq. as high
as 1Ghz. While that product remains to be seen, it should be
on the street on or about the year 2000, as it was reported to
be closely associated with, "MERCED,"Intels 64-bit
PII succesor.
AutismUK
In article <6897su$8fh$1...@news2.isdnet.net>, "Martin Korolczuk"
<mart...@hol.fr> writes:
>The emulated code should be compiled at run-time and then
>you'll spend less than 10 cycles per emulated instruction. The
>N64 has a tens times smaller cache than a PII (16+16 KB of
>L1 cache and 512 KB of very quick L2 cache). And before the
>end of the century (January 1st, 2001), Intel will release 600
>(or more) MHz PII and some IA-64. Some very quick 3D AGP
>cards will help N64 emulators' authors to speed up their
>programs too.
>
>So, first good N64 emulators will appear before the end of this
>century on high-end machines.
Hello the the Intel PR department....
It is worth repeating again that the PC does not just have to be
a bit faster. It has to be a LOT faster. In the N64s case, a lot lot
faster. There is so much more to it than recompiling instructions
(which is much harder than you think !) and shoving a few polygons
out through some PC card. The problem is the lack of direct
compatibility with the target hardware.
Emulator authors are just about get a 3 Mhz SNES emulator
up to speed on Pentiums - do you think they are incompetent or
something ?
Paul Robson (auti...@aol.com)
PS: Another reason for the non-appearence of the N64 emulator is
that Nintendo will attack anyone who looks like producing one.
James Joplin
What about Intels chip under current development with Hewlett Packard? (code
named Merced) It doesnt even run on the x86 CPU standard. From what I have
heard its going to be a processor more similiar to chips of arcades
consoles etc. Would this chip have a better chance at running an N64
emulator?
-James
AutismUK wrote in message <19971227211...@ladder01.news.aol.com>...
Martin Korolczuk
>It is worth repeating again that the PC does not just have to be
>a bit faster. It has to be a LOT faster. In the N64s case, a lot lot
>faster. There is so much more to it than recompiling instructions
>(which is much harder than you think !) and shoving a few polygons
>out through some PC card. The problem is the lack of direct
>compatibility with the target hardware.
Recompiling Rx000 (R3000, R4000, R4300, etc.) instructions
is easier than you think. It's not an easy task, but it's absolutely
feasable.
>Emulator authors are just about get a 3 Mhz SNES emulator
>up to speed on Pentiums - do you think they are incompetent or
>something ?
No, I don't think they are incompetent at all. But the Rx000
processors are really much easier to emulate that many
others. And since the new consoles have instruction cache
and are programmed in C, self-modifying programs are not
common at all. (It wasn't the case on previous consoles or
computers.)
The actual SNES emulators emulate everything by software.
A quick PSX emulator will need a 3dfx card or something like
this, but a N64 emulator will absolutely need a very quick AGP
card in order to run at original speed because of this console's
uncommon texture memory...
I've already seen a very quick R3000 emulator working at
more than 60 MHz on my PII/233. (I've made the test myself
on a real R3000 machine and the emulated one.) So, with a
PII/600, it would run at 155 MHz. Not bad, don't you think?
An R4000 is not more than an R3000 with an arithmetic
coprocessor build inside and a few new instructions, etc.
(I don't know anything about the R4300i used in the N64.
I suppose they have built a matrix coprocessor on the
same chip, reduced power consumption, changed the
cache size.) Nothing a Pentium cannot do.
So, it should be enough power on a PII/600 to convert
the polygon list and send it to the 3D card and make some
other computations. It will be tricky to make everything
working together at this speed, but say 1/3 of that speed
would be nice.
Porting to the next-gen IA-64 should not be very difficult,
you'll need to change the x86 opcodes into IA-64 ones
and compile the emulator to work with the new processor
in order to benefit from the new architecture. :-)
So, I think the first really good N64 emulators will appear
before the end of this century. (There are 3 years left.)
>PS: Another reason for the non-appearence of the N64 emulator is
>that Nintendo will attack anyone who looks like producing one.
Agree. It will be a problematic. But I don't see why emulating
a hardware by software would be against law. Piracy is
prohibited, but not emulation...
michael james perry
"James Joplin" <rjo...@webzone.net> writes:
>What about Intels chip under current development with Hewlett Packard? (code
>named Merced) It doesnt even run on the x86 CPU standard. From what I have
>heard its going to be a processor more similiar to chips of arcades
>consoles etc. Would this chip have a better chance at running an N64
>emulator?
>-James
Yes and no. While the Merced is based on EPIC architecture (explicitly
parallel instruction-set computing), Intel has said that the Merced will
maintain full backwards-compatibility with x86 based 32-bit software. How
this will be done has not been specified. I would imagine that they
will be using a second hardware decoder and translating the x86 instructions
into the new EPIC format although with the rumored "entry level" clock speed
of 600 Mhz, partial software emulation is not out of the question although
the idea is to let software companies enjoy the increase in execution speed
without forcing them to port everything to the new architecture immediately,
kind of similar to the 6502 emulation mode of the 65816. I for one will
not look back to the x86 once the EPIC architecture is available. The
EPIC (IA=64) family line will have a single instruction set for all
models (unlike the x86), will offer 128 _general_ purpose integer registers
, 128 floating point registers (and we're talking ADDRESSABLE registers),
and predication (eg conditional move) of ALL instructions (or most).
Unfortunately, the Merced will be very expensive at first and not geared
towards the casual home user. Intel has said that they will be
continuing the x86 line at the same time that the IA-64 architecture
is available.
In short, an N64 emulator on the Merced is very possible, but even so,
is it a big deal? This is a processor that they expect to have at
1 gigahertz within a year of its debut, and thats just the clock speed
aspect. The architecture is totally different and the clock speed would
be three times faster than the fastest PII available currently. Thats
a LOT more than what the currently available technology can do. Most of
the people who have contributed to this post have argued that its
almost possible already with a 3DFX card and something "just a little
faster" than the PII-300 and thats dreamland. Anything can be emulated
given enough speed (and other factors of course) but thats not the
point. The point is that an N64 emulator is just not possible on
_today's_ (have to be careful when something new pops up every month)
x86 based PC.
--
// "C makes it easy to shoot yourself in the foot. C++ makes it
// harder, but when you do, it blows away your whole leg."
//
// -- Bjarne Stroustrup on C++
ttammi
auti...@aol.com (AutismUK) wrote:
>No, you're missing the point. A 266 Mhz CPU emulating a
>93 Mhz CPU gives you say 3-5 instructions per emulated instructions,
>giving a bit for pipelining and caching. Given the fact that the R4300
>is 64 bit, and the Intel architecture is 32 bit there isn't much
>chance of emulating the functionality.
Well then, that is exactly where the author of the existing N64
emulator has started, emulating the CPU of N64. So he is starting from
the very hardest part of the emulation then, I guess. This is the
release notes of the second revision of the emulator:
--- clip clip clip ---
Project UnReality Revision History
==================================
Vers / Date ł Features
ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄĹÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ
0.2a łImplemented COP0 registers: Random, Wired, PRevID, Index,
Dec 23, 97 łPageMask, EntryLo0, EntryLo1, EntryHi, EPC, and ErrorEPC.
łImplemented COP0/TLB opcodes: TLBWI and ERET.
ł
łImplemented COP1 opcodes: CFC1, CTC1, MFC1, MTC1, and
łCVT.S.W.
ł
łImplemented PI registers: DRAM_ADDR, CART_ADDR, WR_LEN, and
łSTATUS.
ł
łImplemented VI registers: CONTROL_REG, ORIGIN_REG, and
łWIDTH_REG.
ł
łImplemented MIPS Interface register: VERSION.
ł
łImplemented 64-bit registers for 64-bit operations (LD, SD,
łDMULTU, DSRLV32, etc).
ł
łImplemented 64-bit opcodes: LD, SD, DMULTU (currently
łreturns a 64-bit result, not a 128-bit as defined in the
łR4K spec.
łMultLO = result, MultHI = 0), DSLL32, DSRL32, DSRA32, and
łDDIVU.
ł
łImplemented CPU opcodes: LWC1.
ł
łImplemented 2D video emulation for the following modes:
ł - 320x240 ARGB1555 NTSC
ł - 640x480 ARGB1555 NTSC
ł - 320x240 ARGB8888 NTSC
ł - 640x480 ARGB8888 NTSC
ł
łFixed minor memory overwrite in debugger.
ł
łFixed bug in NOR opcode.
łFixed bug where UnReality would crash on the next opcode if
łthe memory debug window was changed to a location that
łwasn't mapped.
ł
łAdded *real* DMA capabilities (ie: DMA takes time to
łtransfer rather than doing it all in one opcode).
ł
łAdded support for loading "Compiled Binaries" that end in
ła .BIN extension.
ł
łAdded support for registering filetypes and storing of
łconfiguration data in the registry.
łAdded support for Frontends by passing the name of the ROM
łimage on the command line.
ł
ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ
0.1a łInitial public release.
Nov 29, 97 ł
ł95% of R4300i (N64's main CPU) emulation complete.
ł20% of COP0 emulation complete.
ł
łCOP1 and COP2 to be implemented...
--- clip clip clip ---
When you say "266MHz CPU emulating the R4300", the question was not
whether today's PCs could run a N64 emulator now at the same speed, but
that whether there could be an useful N64 emulator this century.
Since the deadline is still years away, and P266 hardly is the
fastest and most advanced Intel CPU then, I'm giving it a benefit of
doubt. Don't act as if you knew what future will hold, you don't.
>Natively ? What you mean do a few bitmap layers and a few sprites ?
>Probably PCs faster than 286s could do this, given competent coding.
>Who are these 'people' ?
I think they are living under rocks these days. Nevertheless, that was
another example of a group of people saying "no can do". People who
claimed they know it for sure because they are "experts". Another such
example was how some other "experts" claimed PCs would "never" be able
to emulate Amiga 500 successfully because of all the fast
co-processors, yet I am already running most Amiga games successfully
on both UAE and Fellow emulators on my P233.
>And the CPU is 30 times faster, and the custom chips are more
>complex, being virtual processors in their own right. It is true that
>given sufficient compatibility between N64 3D and 3DFx 3D some
>power could be reclaimed for CPU emulation but not enough.
So you have tested it yourself? And what is "enough" in this context
anyway? We'll see in 2001. ;^) Like I said, I'm giving it a benefit of
doubt till the end of century.
Brennan Tallack
There is several N64 emulators in the workings, all promising to tun on
Pentium 166's with good speed. The N64 obviously isn't 2 hard to emulate
seeings how there is 3 organizations (to my knowledge, could be more)
which are all close to full emulation.
-B Tallack
Martin Korolczuk <mart...@hol.fr> wrote in article
<6897su$8fh$1...@news2.isdnet.net>...
> >>>You ought to look more closely at the R4300. It is not simply
> >>>a matter of rendering the graphics to the display.
> >>Nowhere did I say so. But the fact that they can use 3D
> >>accelerators for the emulator does take a lot of the burden
> >>off the CPU.
> >
> >No, you're missing the point. A 266 Mhz CPU emulating a
> >93 Mhz CPU gives you say 3-5 instructions per emulated
> >instructions, giving a bit for pipelining and caching. Given
> >the fact that the R4300 is 64 bit, and the Intel architecture
> >is 32 bit there isn't much chance of emulating the functionality.
>
>
> Can you tell me WHAT is 64-bit in the R4300? The 64-bit floats
> are nothing compared to 80-bit PC doubles. Something else?
> The bus is 64-bit? Nothing of so difficult or time-consuming onto
> a PC, especially with a quick AGP 3D card... (Since the N64
> doesn't have its own big texture memory as the 3dfx has, an
> AGP card will be needed.)
>
>
> >Also, emulators don't respond to caching too well - though
> >the code being emulated may involve tight loops etc., the
> >emulation code itself is a bit like a big case statement...
>
>
> The emulated code should be compiled at run-time and then
> you'll spend less than 10 cycles per emulated instruction. The
> N64 has a tens times smaller cache than a PII (16+16 KB of
> L1 cache and 512 KB of very quick L2 cache). And before the
> end of the century (January 1st, 2001), Intel will release 600
> (or more) MHz PII and some IA-64. Some very quick 3D AGP
> cards will help N64 emulators' authors to speed up their
> programs too.
>
> So, first good N64 emulators will appear before the end of this
> century on high-end machines.
>
>
>
>
AutismUK
In article <01bd1679$72293900$e3f6...@cafe.supernet.ab.ca>, "Brennan Tallack"
<btal...@geocities.com> writes:
>There is several N64 emulators in the workings, all promising to tun on
>Pentium 166's with good speed. The N64 obviously isn't 2 hard to emulate
>seeings how there is 3 organizations (to my knowledge, could be more)
>which are all close to full emulation.
Well, they are all bloody liars then.
AutismUK
In article <VA.000007ea.010d42be@tlt>, ttammi <tta...@netlife.fi> writes:
>Well then, that is exactly where the author of the existing N64
>emulator has started, emulating the CPU of N64. So he is starting from
>the very hardest part of the emulation then, I guess. This is the
>release notes of the second revision of the emulator:
No you don't get it. The processor isn't the hardest part - often its the
easiest part. But for a CPU this fast its the most speed intensive part.
It's the hardest part to get running at speed. The listing below is irrelevant.
<snipped>
I also vaguely remember (if its the same emulator) that he estimated it
as running about 10% of the real speed with no graphics or sound and
"75% optimised".
>When you say "266MHz CPU emulating the R4300", the question was not
>whether today's PCs could run a N64 emulator now at the same speed, but
>that whether there could be an useful N64 emulator this century.
>Since the deadline is still years away, and P266 hardly is the
>fastest and most advanced Intel CPU then, I'm giving it a benefit of
>doubt. Don't act as if you knew what future will hold, you don't.
No, but you don't either. Given Intels fatuous claims about MMX "30-40%
faster" and their attempts to pretend Slot 1 is unconnected with their
inability to produce Pentium Pro's cheap enough I'm inclined to take their
claims less than seriously. If they can produce these amazing CPUs
are they likely to be affordable ? Especially if the stop anyone else
producing competition, because lets be realistic, CPUs would me far
more expensive if it wasn't for Cyrix and AMD.
>>Natively ? What you mean do a few bitmap layers and a few sprites ?
>>Probably PCs faster than 286s could do this, given competent coding.
>>Who are these 'people' ?
>I think they are living under rocks these days. Nevertheless, that was
>another example of a group of people saying "no can do". People who
>claimed they know it for sure because they are "experts". Another such
>example was how some other "experts" claimed PCs would "never" be able
>to emulate Amiga 500 successfully because of all the fast
>co-processors, yet I am already running most Amiga games successfully
>on both UAE and Fellow emulators on my P233.
Ermm.. same question - who are these people ? I don't recall any posting
saying Amiga emulators or SNES emulators were "impossible". Especially
as the A500 doesn't have any "fast coprocessors", just some nifty hardware,
so anyone making this statement is not an expert by definition.
This is an irrelevant line of argument - because some unnamed people
present a view I've never heard anyone present, and were "wrong" then
any 'expert' is also wrong.
Please name an "expert" who made either of these claims and tell me what
they said. (this reminds me of Creationism...)
>>And the CPU is 30 times faster, and the custom chips are more
>>complex, being virtual processors in their own right. It is true that
>>given sufficient compatibility between N64 3D and 3DFx 3D some
>>power could be reclaimed for CPU emulation but not enough.
>So you have tested it yourself? And what is "enough" in this context
>anyway? We'll see in 2001. ;^) Like I said, I'm giving it a benefit of
>doubt till the end of century.
So because I haven't written an N64 emulator I can't make any comment
on its feasibility ? Why not ? And if I can't, why can you make any judgement
about it at all ?
Paul Robson (auti...@aol.com)
ttammi
auti...@aol.com (AutismUK) wrote:
>No you don't get it. The processor isn't the hardest part - often its
>the easiest part. But for a CPU this fast its the most speed intensive
>part. It's the hardest part to get running at speed. The listing below
>is irrelevant.
Wow you keep changing the argument. This is what you said, and I quote:
"Given the fact that the R4300 is 64 bit, and the Intel architecture is
32 bit there isn't much chance of emulating the functionality."
You implied that it cannot be emulated on 32bit Intel CPUs, period.
You were wrong.
>I also vaguely remember (if its the same emulator) that he estimated
>it as running about 10% of the real speed with no graphics or sound
>and "75% optimised".
Three years to go.
>>doubt. Don't act as if you knew what future will hold, you don't.
>
>No, but you don't either.
And nowhere have I said I do. All I have said is "I'm giving it a
benefit of doubt", you keep claiming it cannot be done. I hope you
understand the difference there.
>Given Intels fatuous claims about MMX "30-40%
>faster" and their attempts to pretend Slot 1 is unconnected with their
>inability to produce Pentium Pro's cheap enough I'm inclined to take
>their claims less than seriously. If they can produce these amazing
>CPUs are they likely to be affordable ?
What are you talking about? The question was not whether you can run a
N64 emulator at 100% speed on a P100, but whether it will be possible
for a N64 emulator to run at respectable speed and emulation (so that
it is practical) on _any_ PC system in the end of this century.
I'm saying "maybe, maybe not, we'll see", you keep saying you know for
sure it cannot be done, and keep changing the argument.
>Especially if the stop anyone else producing competition, because lets
>be realistic, CPUs would me far more expensive if it wasn't for Cyrix
>and AMD.
That has no relevance whatsoever to this discussion. Intel is not
currently, and not going to, concentrate only on the lower end home
market. They are also aiming at the more expensive CPUs on graphical
workstations, Internet servers etc.
>>example was how some other "experts" claimed PCs would "never" be
>>able to emulate Amiga 500 successfully because of all the fast
>>co-processors, yet I am already running most Amiga games successfully
>>on both UAE and Fellow emulators on my P233.
>Ermm.. same question - who are these people ? I don't recall any
>posting saying Amiga emulators or SNES emulators were "impossible".
Read dejanews for comp.sys.amiga.advocacy if you want to. This was a
hot topic back then, and one of the reasons why one/some people decided
to start working on UAE, to show that it is both possible and practical
to emulate Amiga on PCs. The Amiga side used exactly similar arguments
as you are using now, and kept saying "no can do".
>This is an irrelevant line of argument - because some unnamed people
>present a view I've never heard anyone present, and were "wrong" then
>any 'expert' is also wrong.
Just goes to show that when some 'expert' claims "no can do", maybe
people shouldn't take his words as the truth.
>>So you have tested it yourself? And what is "enough" in this context
>>anyway? We'll see in 2001. ;^) Like I said, I'm giving it a benefit
>>of doubt till the end of century.
>So because I haven't written an N64 emulator I can't make any comment
>on its feasibility ? Why not ? And if I can't, why can you make any
>judgement about it at all ?
In case you cannot read, I said: "I'm giving it a benefit of doubt".
You, on the other hand, kept saying it is impossible to emulate it
at all (which of course is rubbish, take for example SPIM), and act
like a total ass saying anyone that doesn't agree with you "should
learn more about it". Grow up.
There is a big difference in saying "it cannot be done" and "we'll see
whether they can make it happen".
AutismUK
In article <VA.000007fa.00373e82@tlt>, ttammi <tta...@netlife.fi> writes:
>>No you don't get it. The processor isn't the hardest part - often its
>>the easiest part. But for a CPU this fast its the most speed intensive
>>part. It's the hardest part to get running at speed. The listing below
>>is irrelevant.
>
>Wow you keep changing the argument. This is what you said, and I quote:
>"Given the fact that the R4300 is 64 bit, and the Intel architecture is
>32 bit there isn't much chance of emulating the functionality."
at CPU speeds 3 years down the line at a playable speed....
You are trying to change my argument. This is not a discussion about
possible vs impossible, but a discussion about the possibility in the
next 3 years of a workable N64 emulator. You are ignoring the context
of the thread. In fact, in this whole post, there is very little else but
repetition of this misrepresentation.
Presumably it is necessary to add 'in the next 3 years' onto any
sentence I write , otherwise you will assume it means 'for ever'.
BTW: You have ignored the fact you misunderstood the post. The
'hardest' part can be defined in terms *other* than coding difficulty.
The 'hardest part' for the CPU is getting it quick enough - don't you
understand this ?
>You implied that it cannot be emulated on 32bit Intel CPUs, period.
>You were wrong.
Well, your reading of what I implied is incorrect. I have stated the
Turing Machine argument several times. Read the earlier stuff in
this thread about cycle availability etc and read arguments in
context. I didn't feel the need to add "in the next 3 years" to
every sentence. I thought it was obvious.
>>I also vaguely remember (if its the same emulator) that he estimated
>>it as running about 10% of the real speed with no graphics or sound
>>and "75% optimised".
>
>Three years to go.
You really believe CPU power will be 8-9 times faster in 3 years ?
You might be right. But it violates Moore's law. I don't think that
technology has speeded up that much in 3 years before - I'd
expect a 4x speed boost at best, and we all need motherboards
that offer some sort of comparable speed.
>>>doubt. Don't act as if you knew what future will hold, you don't.
>>No, but you don't either.
>And nowhere have I said I do. All I have said is "I'm giving it a
>benefit of doubt", you keep claiming it cannot be done. I hope you
>understand the difference there.
Nowhere have I said I do either. What I am saying is that I don't
believe that CPUs will be fast enough by the end of the century.
Incidentally, it is a perfectly reasonable "implication" that you
are arguing "for". You give it the benefit of the doubt. I doubt it.
(very strongly).
[Incidentally, it is polite to write <snip> to point out a truncated
statement or some similar indicator.]
>>Given Intels fatuous claims about MMX "30-40%
>>faster" and their attempts to pretend Slot 1 is unconnected with their
>>inability to produce Pentium Pro's cheap enough I'm inclined to take
>>their claims less than seriously. If they can produce these amazing
>>CPUs are they likely to be affordable ?
>
>What are you talking about? The question was not whether you can run a
>N64 emulator at 100% speed on a P100, but whether it will be possible
>for a N64 emulator to run at respectable speed and emulation (so that
>it is practical) on _any_ PC system in the end of this century.
>I'm saying "maybe, maybe not, we'll see", you keep saying you know for
>sure it cannot be done, and keep changing the argument.
Incorrect. I am saying that I don't think it can be done. You are actually
saying you are giving it "the benefit of the doubt" which isn't the same
as "I don't know".
Incidentally, the point of this Paragraph you've split is that Intel claims
for performance and delivery are not necessarily accurate.
Please understand that discussions are in context. The above statement
needs to be taken in the context of the above quotation (e.g. it means in
the next 3 years not ever..).
>>Especially if the stop anyone else producing competition, because lets
>>be realistic, CPUs would me far more expensive if it wasn't for Cyrix
>>and AMD.
>
>That has no relevance whatsoever to this discussion. Intel is not
>currently, and not going to, concentrate only on the lower end home
>market. They are also aiming at the more expensive CPUs on graphical
>workstations, Internet servers etc.
Sorry ? I don't understand the point of this at all. I was simply explaining
Intel's track record in honesty is not brilliant.
>>>example was how some other "experts" claimed PCs would "never" be
>>>able to emulate Amiga 500 successfully because of all the fast
>>>co-processors, yet I am already running most Amiga games successfully
>>>on both UAE and Fellow emulators on my P233.
>>Ermm.. same question - who are these people ? I don't recall any
>>posting saying Amiga emulators or SNES emulators were "impossible".
>Read dejanews for comp.sys.amiga.advocacy if you want to. This was a
>hot topic back then, and one of the reasons why one/some people decided
>to start working on UAE, to show that it is both possible and practical
>to emulate Amiga on PCs. The Amiga side used exactly similar arguments
>as you are using now, and kept saying "no can do".
No, I'm saying 'no can do with the next 3 years technology'.
....
I can find lots of stuff arguing about whether UAE is any good or not
but not anyone saying it was impossible. As you clearly had an expert
in mind (otherwise you couldn't have claimed expertise for him/her) who
was making this claim.
No wait, I've found some in 1995. Some chap said PCs weren't fast enough.
In 1995 that was almost certainly correct. One chap seems to have got
flamed. Most of it seems to be arguing about multitasking, then about
porting AmigaOS. Interestingly, someone has pointed out fairly early on
that such a beast exists (early UAE probably), and they are debating its
speed (on early 95 hardware).
Ah, now I've found someone who would be expert (Marat) but he doesn't
seem to say much.
Having looked through the 95 names on this thread, no-one i see apart
from Marat is an expert on emulation. As your expert claims that the
Amiga has coprocessors (Copper List :-) ? ) then they aren't an Amiga
expert. A name please, and a post for both your SNES and Amiga claims,
and some idea of why you considered them expert, and in what field.
(What a dull way to spend ten minutes)
>>This is an irrelevant line of argument - because some unnamed people
>>present a view I've never heard anyone present, and were "wrong" then
>>any 'expert' is also wrong.
>Just goes to show that when some 'expert' claims "no can do", maybe
>people shouldn't take his words as the truth.
No it doesn't. This is incorrect. These 'newsgroup posters' are automatically
'expert' simply because you want them to support your argument.
Noone is saying it is impossible to emulate an N64 period. No emulator
author would say anything was impossible - (unless there are no working
machines or documentation or software available, then creating an
emulator would be impossible).
I believe you are just flinging 'experts' around. You clearly don't have
a particular post in mind from someone who has demonstrated
expertise in emulation. All it really means is "someone said it".
What you are doing is trying to generalise from a spurious argument about
"experts" on some newsgroup that because experts say something is
unlikely or impossible, then expert views should not be taken seriously.
There is no reason to make this generalisation. Experts are sometimes
wrong,but usually less often wrong than those with less expertise.
>>>So you have tested it yourself? And what is "enough" in this context
>>>anyway? We'll see in 2001. ;^) Like I said, I'm giving it a benefit
>>>of doubt till the end of century.
>>So because I haven't written an N64 emulator I can't make any comment
>>on its feasibility ? Why not ? And if I can't, why can you make any
>>judgement about it at all ?
>In case you cannot read, I said: "I'm giving it a benefit of doubt".
>You, on the other hand, kept saying it is impossible to emulate it
>at all (which of course is rubbish, take for example SPIM), and act
>like a total ass saying anyone that doesn't agree with you "should
>learn more about it". Grow up.
Aahh. Resorting to abuse now eh ?
I am not saying it is impossible to emulate it at all. I am saying that I
don't believe it will be done by the end of the century. If you want to
attack this fine, but don't move the goal posts by trying to get me to
defend a view I don't hold. (Must have written this half a dozen times
by now).
It is nonsensical to claim you are taking a neutral stance on this
issue. Are you really claiming you are making no claims about the
feasibility of the emulator (other than it is technically possible in
some long term scenario) ?
BTW: Don't bother quoting "It can't be done" quotes out of context.
But if you want to, there's one above, a couple of paras or so.
>There is a big difference in saying "it cannot be done" and "we'll see
>whether they can make it happen".
This is true, but that isn't actually what you are saying is it ? You are
giving it the "benefit of the doubt" - i.e. you think it is more likely to
happen than not ?
Why do you think this ? Just because some 'expert' said an Amiga
emulator was impossible ? [what about the SNES BTW ?].
Do you have any arguments to support your viewpoint ?
Mine are (briefly summarised) :-
- the machine is too fast given my estimation of CPU speeds at that date.
(even allowing for non-trivial dynamic recompilation).
- using Ensoniq Sound or 3DFX GFX cards still leaves a large translation
problem, emulating pecularities of the hardware etc..
- Nintendo will kill anyone (possibly literally) who produces such a machine.
Regards, Paul Robson (auti...@aol.com)
[I do believe you wrote something about 64 bit floats compared against
80 bit floats on a Pentium (it may have been someone else). This
ignores the translation requirements between the two formats].
OS-Dev
>3Dfx cards are nice but the N64 sports some features the 3Dfx doesn't have,
>any working playable N64 emulators for high-end consumer computers this
>century.
HAHAHAHAHA! Well my 3dfx does bi-linear and tri-linear filtering, must be a
freak!
Oh, and _unlike_ a n64, it actually allows for textures bigger across than
the average IQ of the
trolls in this group. (ie 64)
-Josh
flare
On Fri, 2 Jan 1998 13:11:15 +1100, "OS-Dev"
<os-...@effect.nospam.net.au> wrote:
>
>-Josh
GWA HA-Ha
Take that, trolls!
And, in less than a year, Voodoo 2 will be out!
Immagine that, up to more than FOUR times the current power of the
Voodoo chipset!
TAKE THAT N64!!!!!!
Regards,
Flare
Prey
Acutally, idiot, read the facts before you post, no four times here, average
about 3 times....
flare wrote in message <34b5ca6a...@news.idt.net>...