PowerPC emulation on x86 possible?

[Hammer]

xeron wrote:

mdma wrote:
A 3GHz x86 CPU emulating a PPC has got to be faster than a 600MHz G3 or near enough same speed though.  Amithlon for OS4 anyone? ;-)

Your maths are likely to be somewhat wrong. The PowerPC is a relatively modern processor. Remember, clockspeed is not an indication of how powerful a processor is.

Note that modern X86 processors (e.g. K6/K7/K8 and Pentium Pro/II/III/VI) translates X86 ISA into smaller RISC like instructions via HW decoder/translator before pumping it into the Post-RISC pipelines architecture. Think of Transmeta software/hardware translator done in pure HW acceleration. This techniques enable X86 CPU manufactures to change the cores without changing the ISA.

To reach reasonably high Mhz speed (in region of 1.8Ghz at this time**), IBM's PowerPC 970 employs a similar action (i.e. decode/crush stage) with PowerPC ISA.

[Hammer]

I wonder what level of decomposition is used for the PPC ISA? The instructions are far simpler and more orthogonal than x86 ones to start with...

Well, IBM has decided to employ crush/decode way and 16 depth pipeline stages for reaching reasonably high Mhz (compared to Motorola’s PowerPCs).  

X86 translation/decode HW has this thing called "zero cycle decoding" for for certain instructions. With AMD64, they have sawn off the 16bit stuff in the AMD64 mode.  

[Hammer]

when people starts a thread like this , just when things starts to look better, well its just like amithlon in 2001, it made people LEAVE! and they wont come back ..why? because they belive os4 will be emulated asap because of a thread like this, and everyone knows that by emulating a ppc will make the x86 even more popular and especially WINDOWS! .

Note that Linux X86 has leveraged the X86 desktop dominance to fight against the traditional Unix servers.

linux never was a contender and never will be, it will remain as a geek os just like what it was designed to be.

Tied Lindows 4.0?

Red Hat’s recent comment is just a cop out since their distro is just crap for desktop use.

anyone remeber the "UAE can do 3d?" thread that appeard here some time ago? , well did you notice who started it? and did you know that warp3d is out for uae because of it or hehe something?

Note that it didn't throw out Warp3D compatibility since it's just a wrapper for Window's 3D subsystems for WinUAE.

X86 = windows , linux = geek , amiga = emulators , beos died and became an x86 os which is kinda nice but i think that only because it DIED or heh..the ppc plattform around it.

I don’t think PowerPC alone would be an absolute guarantee in keeping out MS Windows since there is already a Windows NT 4.0 PowerPC edition.  Please note the X-BOX 2. The Windows NT foundation for PowerPC is already has been completed. All it needs is a reactivation. We know the X-BOX is powered by a cut down Windows NT5 i.e. an OS built on NT4.  

[Hammer]

MarkTime wrote:
well I think its fairly obvious that people want PPC emulation so that they can a) illegally run Mac OS X, and b) in this community anyway, illegally run OS 4, and c) possibly write an illegal emulator for the Gamecube

Note that Microsoft is now aggressively selling VirtualPC for MacOS PPC market i.e. thus enabling them to sell more Windows products to this particular market.

[Hammer]

That RISC chips are more efficient than x86 is not exactly new news...

One of the aspects of RISC concepts are the fix length instructions. Ever since AMD's K6 processor, it translates variable length CISC X86 instructions into fix length instructions “RISC86” (using AMD's words).

But you don't have to take my word for it read

Arstechnica's articles
RISC vs. CISC: The Post-RISC Era
G4 vs. K7: an architectural comparison
AMD's 64-bit alternative: x86-64

Chip Architect's
http://www.chip-architect.com/news/2003_09_21_Detailed_Architecture_of_AMDs_64bit_Core.html

The links provided addresses most of Paul DeMone’s points.

[Hammer]

In terms of your laptop, efficiency is a big deal.

What about the efficiency in using the money you have?

If the CPU is effieient it'll give you longer battery life. Do you think a 50W difference is worth it for perhaps 2X the speed? The G4 uses 20W and thats an old version, the modern low voltage G4s go down to 7.5 Watts at 1GHz, Compared to the P4 which uses 70 Watts.

Try against “Pentium M” i.e. Intel’s X86 processor designed solely for light and thin markets. Note that “Pentium M” @1.6Ghz integer performance is  roughly equal to Pentium VI @ 2.5Ghz. Intel is offering two different X86 cores in the market place i.e.
1. X86 core that is IPC bias e.g. Pentium M.
2. X86 core that is clockspeed bias e.g. Pentium 4.

PS; It didn’t stop PC vendors in using Pentium M for blade servers.

And for your application the fact the G4 has Altivec will make quite a difference, on Altivec code the G4 it's quite probably to outgun the P4 even at 3GHz - because the design is less efficient.

Note that the Pentium VI was built for clock speed first than IPC second.  The Altivec doesn’t rescue PowerMac G4 @1.4Ghz in every cases from being last in graphic extensive applications e.g. modern 3D games.

References;
http://www.barefeats.com//p4game.html

Note that next generation PowerMac G5 should have beaten the superseded AMD K7 Athlon MP@2.1Ghz(DDR266 FSB) due to G5’s  “superior” DDR1000 FSB speed.

For completeness and optimised for gamming purposes, the Athlon MP’s chipset in that case was not even
1. AthlonXP3200+/NVIDIA’s nForce2 400 Ultra
2. Athlon FX51/64 3200+
3. Pentium VI@3.2Ghz/Extreme Edition.
 
PS; There are two types of Athlon MP @2.1 Ghz i.e.
1. Athlon MP 2600+ @2.1Ghz (use in barefeats test)
2.  Athlon MP 2800+ @2.1Ghz

Reference for Athlons MP PR ratings;
http://www.pathwayexpress.com/catviewL.cfm?cid=CPU

If the CPU has Out of Order execution the smaller number of registers is going to have a big impact on the design of that stage making it considerably more complex.

Register renaming schemes is the way to expand the limited registers of the X86-32 limitations.  

[Hammer]

The P4 and Athlon are both very fast CPUs, but in order to get that speed they have to do a lot of work

Note that IBM's PowerPC 970 and AMD's Athlon 64 3200+ are both clocked at 2.0Ghz. AMD's Athlon 64 3200+ pretty much rivals/beats the Pentium VI @3.0Ghz in games (needs PIV @3.2C Ghz ).

In general performance; the Athlon 64 3200+ > Athlon XP 3200+/nForce II 400 Ultra**.

**Chipset type is important with AMD's K7 processors (e.g. in general, signal channel nVidia nForce2 400 > VIA KT600 > KT400A).  

and consume a lot of power.

Note that the IBM PowerPC 970 has ~52~55million transistors, which is in the ball park of AMD’s Athlon  XP-M Barton Core in regards thermal characteristics (without taking into an account of Speed-Step like technologies).  

If IBM put the effort into design they could produce a faster processor in the same silicon technology. I don't kow if the 970 reaches that goal (at least with current compilers) but I expect the next gen (due next summer) may do so.

In terms of overall benchmarks results and even with superior FSB speed (in the case of K7); IBM's PPC 970 is struggling to beat/surplus the similar clocked late model AMD processors.