68060 equivalent to 486 DX or Pentium in performance ?

[bundyboy123]

Always wondered if my 68060 performs like a 486 DX or pentium .
I think from what it does its very 486ish.
Why did the 68k stop at 060 ?
worth a thought

 

[Sardine]

060 is a very good cpu, so is the 486,

the problem is the Amiga is missing the hardware Planar bit mapping functions that most decent pc vga cards had at the time making games like doom very smooth and playable even on a 486 sx-25

now if you watch an 060 on an amiga playing doom purely in software bitmap mode then that's impressive,, install a vga card with no Planar support in a 486 and watch the fps drop below 10

 

[diskers]

Making short and not too technical the MC68060 is rather / was rather the pentium from Motorola. As the Pentium from Intel it was the first superscalar processor (instruction level parallelism) available for the end user from Motorola (however they weren't the first ones on the market - ex: the Motorola MC88100).
PS. There's a quiet good explanation about supercalar architecture on wikipedia.

 

[Akiko]

The 68060 is a superscalar (2 issue) and pipelined (4 stages) architecture. Most of the instructions can be executed in one cycle, which are characteristics of a RISC CPU. Actually it is a RISC core with a CISC frontend. The 68060 has a higher integer performance than a pentium, but it's floating point performance is much lower (about 50%), because the FPU of the 68060 is not superscalar. Funfact: Until today the 68060 has the fastest integer multiplaction logic ever built into a CPU. It takes 2 cycles and there is only one CPU out there, that have 2 cycles in an ideal situation, the VIA Nano 3000. Every other CPU takes 3 to 27 cycles.

A 486DX 25 MHz has the same performance like a 68020 25 MHz in the real world (about 8 MIPS). The 486 has a somewhat higher IPC, which starts to show on higher MHz numbers.

 

[Deling]

68060 = pentium 75-90 depend of mhz

 

[diskers]

A 68020@33 with a 68881 or 68882 FPU
But rather a 68030 is more comparable to a 486.
Without it I will rather compare it to a 386DX33/40 or an 486SX (however the low speed SXs were slower than 386DXs @33 and 40 Mhz - as I remember well the SXs were without FPU) but with an advantage for the MC68020 / 68030.
By the way a small off topic, theoretically a 2 CPU server board with an 486DX4 120 or a AMD5x86 133Mhz would be in my opinion more efficient in arithmetics than a single Pentium pro server board. For exemple on a database server it would be cheaper and more efficient.

The 68060 is a superscalar (2 issue) and pipelined (4 stages) architecture. Most of the instructions can be executed in one cycle, which are characteristics of a RISC CPU. Actually it is a RISC core with a CISC frontend. The 68060 has a higher integer performance than a pentium, but it's floating point performance is much lower (about 50%), because the FPU of the 68060 is not superscalar. Funfact: Until today the 68060 has the fastest integer multiplaction logic ever built into a CPU. It takes 2 cycles and there is only one CPU out there, that have 2 cycles in an ideal situation, the VIA Nano 3000. Every other CPU takes 3 to 27 cycles.

A 486DX 25 MHz has the same performance like a 68020 25 MHz in the real world (about 8 MIPS). The 486 has a somewhat higher IPC, which starts to show on higher MHz numbers.

 

[koshman]

Not a very scientific test but playing Quake on 060 at 80Mhz I think the experience was comparable to a Pentium (say < 90Mhz one) - very fluent at lowres even with the planar graphics disadvantage.
For gaming generally I would say definitely more Pentium than 486.

 

[diskers]

I know that I will open the pandoras' box but for the 68060 as well as for modern i386 compatible CPUs it's not a RISC core with a CISC frontend but rather a CISC core with internal RISC like instruction capabilities. A RISC core will obviously need a external layer to execute any CISC instructions what make no logical sense to it. Moreover with modern CPUs what gain you will get with a CPU with only reduced instructions sets, speed? Nowaday not really noticable. Ok, maybe for small embedded system or low power consumption ones.Apart those it will only add more complexity in programing and compilation.

Actually it is a RISC core with a CISC frontend

 

[sax]

I think 68060/50 MHz can bring the speed of the Pentium90/100 MHz. Maybe it just depends on the better usability of the Amiga OS, but under Win95 I could barely play an MP3 on a P100, and even a Word text editor would sometimes stutter. The real breakthrough on PC came with the MMX instruction set, a P166 MMX was already very fast, but the PPC 604 / 603 was faster. P166 MMX came in 1997, 68060 relase date is 1994. (same as Pentuim 100)

Motorola 68060 50 MHz -> 67 MIPS
Pentium MMX 200MHz -> 227 MIPS (I think, it is real)
Pentium 100 MHz -> officially 188 MIPS (based on measurements, it's no more like only 112, but you can see MMX is az fast at the same clock speed... ha-ha )
486 DX4 100 MHz officially 70 MIPS (based on measurements, it's no more like only 50)
486SX/33 -> 16 MIPS (no FPU)
PPC 603e 240MHz -> 295 MIPS (123 MIPS at 100 MHz)
PPC 604 400 MHz -> 512 MIPS (100 MHz -> 128 MIPS)

On paper, the Pentium is faster at the same clock speed, but in reality I think the 060 is 2x faster.