Steve Jobs: G4 is slower than Intel CPUs

[Hammer]

Karlos wrote:
Slower or not, lets just see OS4 and MOS running on 933MHz G4 systems and sit them side by side with XP or 2003 on the fastest x86 you can find...

Obviously for processor intensive work there's no contest but I've seen my OS3.5 on my 040 be more responsive than my friends AlthonXP 2800 many occasions :-)

Overall (WindowsXP/2003) speed is dependant on surrounding installed components i.e.
1. The amount of installed ram (important for XP/2k3).
2. Installed video card (and drivers).
3. Motherboard chipset.
4. Windows’s available free physical memory.
5. Boot'vis application.
6. the type of hard disk installed.

I run, Athlon XP @ 2.2Ghz/Geforce 4600 TI/nForce II 400 Ultra/1Gb PC3200 DDR SDRAM (dual channel)/2x 80Gb 7200RPM UDMA IDE RAID (boot drive). It’s faster than my old A3000/040@25Mhz (no gfx card).
Such a hardware setup should fly with AROS i.e. PC world’s brute force method with near Amiga like OS efficiencies.

With ‘1Gb PC3200 DDR SDRAM’, it reduces the WindowXP’s use of the slower virtual memory. RAID IDE setup accelerates disk access.

PS; I do have lesser AthlonXP/512Mb/VIA KT class machine…

[Hammer]

So you need all that hardware acceleration to make your windows box snappier than an 040 powered no-graphics-card amiga 3000 eh?

I don’t have access to Zorro based Gfx card anymore. The issue in regards to Windows GUI’s speed is an operating issue NOT a hardware issue.
If one is bias for relatively lightweight OS one could purchase QNX X86 (or download AROS, or even install Amithlon).

There are some things that the AmigaOS is unsuited at this time due to lack of ‘middleware’  infrastructure. This area is related to RAD development for revenue/income generation.

My point was, a friend has a 'half decent' AthlonXP 2800 400MHz FSB, Asus A7V8X-X mobo, 768M DDR400, 128M Radeon 9800 and ATA133 120G drive. It's a nice setup.

Ok. The hardware is fine (except for VIA KT600 class motherboard).  Configuration is another matter.….

However, running Win2K, XP or 2003 (uninstalled after virtually nothing worked  ) there were times when simply bringing up a popup menu caused a delay of a couple of seconds. For the power and acceleration available I'd expect a lot better.

I yet to encounter that 'issue' on my system due to different chipset/RAID-HD/memory/Gfx/driver setup (relative to your friend's setup). Secondly, there are the issues of deadlocks** and some issues with certain MS auto up-date patches.

**broken applications/utilities that plug itself in popup menus.

PS; I have set my WinXP-SP1's popup menu with fade-in/fade-out effect.

running Win2K, XP or 2003 (uninstalled after virtually nothing worked  )

MS Windows Server 2003 should have worked with
ASUS A7V8X-X (VIA KT600 based) class motherboard...

If an ageing 68K system with a third rate graphics card (the hardware features of which are totally under utilized by the RTG software as it is) can have a more responsive UI, what does that say for MOS, OS4 and as you point out, AROS?

The issue you pointed out is an operating system issue NOT a hardware issue…

[Hammer]

Even Intel know that the x86 is soon dead.

Not according to the Intel's road map. Note that they just announced 'Pentium 4 Extreme Edition'(refer IDF 2003)....

Other Pentium 4 cores;
1. Prescott (Q4 2003)
2. Nocona
3. Potomac
4. Tejas (2004/2005)

Backwards compatibility is key aspect for continuing dominance of Intel, AMD, MS, X86 Linux and X86 BSD.

Note that Motorola 68k was once the ‘Pentium’ during early 80s…

I would prefer a platform that is using a modern RISC CPU(the PPC is only 10 years old).

Note that modern X86 CPUs has a post-RISC like cores (i.e. they have a fancy HW translators/emulators to make them run X86 legacy).

Transmeta processor's (VLIW concepts)decoder/translator is combination software and hardware.

You will either upgrade your computer platform now, or wait 5-10 years when Intel and AMD say "Nope! That's enough! We're going on with another architecture.

With AMD K6/AthlonXP(K7)/Opteron/AthlonFX/Athlon64(K8) processors they decodes/translates X86 Instructions into RISC like instructions before they execute in their post-RISC  pipelines.

The PowerPC 970 (Power Series based core) has similar process for PowerPC ISA e.g. decode/crush stage.

The current IA-64 (VLIW/EPIC concepts) does have X86 ISA compatibility on HW, but it's only poorly implemented…

If you're worrying about these machines not using the old x86 hardware then you're not thinking about the long term future enough.

With AMD's case, they will just replace K's series post-RISC core with another RISC core and redesign the front-end translator/decoder.
 
Note that AMD's K9 is under development just for 6 months...

[Hammer]

Wain wrote:
The 80x86 command set  is an old, outdated, and (by modern standards) flawed set.  Intel is trying to replace it with their IA series of processor (slowly, but probably surely while they attempt to pry the backwards compatibility nuts off of the old set with a crobar).

Note that Intel refers X86-32 as "IA-32". IA-64 (VLIW) is just the inefficient version of Transmeta’s CPU.
"IA-32" refers to X86 ISA period of 80386 and later.

Note that Intel's 'Pentium Pro' has a very poor implementation X86-16, but it was fine on X86-32 (e.g. Windows NT code base). This is somewhat related to AMD's decision to delete the V86 during X86-64 mode.  

The 80x86 chipsets are being phased out,

Post-RISC core of an Athlon processor is substantially different to 8086 e.g. 8086 doesn't decode/translate X86 ISA into smaller RISC like instructions. AMD Athlon K7’s chipsets is based on DEC Alpha’s EV6 architecture i.e. non-X86 architecture.

and being slowly replaced by incompatible CPU's

Where did they state this?
Did AMD state that they will replace AMD64 ISA?

They’ll just delete particular compatibility modes (e.g. V86 while in X86-64 mode).

(yes AMD's still planning on true backward compatibility, but even they've mentioned plans to eventually phase it out)

AMD64 ISA(X86 ISA without V86 nasties) will be here for a long time.

So G4 isn't "brand new top of the line" as of what 4 months ago or something?  

Note that PowerPC 970 decodes/crush PowerPC ISA into smaller RISC like instructions i.e. a process should be similar to a certain X86 processor.

[Hammer]

Something that should be really clear to most people that I am not hearing people noticing on here is how Microsoft is shifting to a non processor specific architecture.

We probably see yet another WOW(Windows on Windows). Note that MS Window Longhorn (e.g. Build 40xx)runs fine on X86...

If you are following microsoft's .NET architecture

Software investment protection is one of the critical factors why X86 dominates the desktop markets...