How much faster is an iMac than an A500?

[Starglider 2]

Hi, for a future video I'm interested in knowing how many multiples faster this iMac processor is than an A500 with its 68000:

3.6GHz 8-core 9th-generation Intel Core i9 processor

I'm guessing x2000 but I'd love to work out a way to know for sure.

[saimon69]

hmm, 7.16 MHz single core versus 3600 MHz eight cores, we are in the "old Fiat 500 vs Millennium Falcon" speed comparison here...

[Hewitson]

Wouldn't be difficult to come up with a simple test. I don't have time to look into it at the moment though.

[Starglider 2]

Originally Posted by saimon69

hmm, 7.16 MHz single core versus 3600 MHz eight cores, we are in the "old Fiat 500 vs Millennium Falcon" speed comparison here...

Okay that's a start. Would it be as simple as:

3600 MHz x 8 cores = 28,800
7.16 / 28,800 = 3,840 times faster?

[Aladin]

ryzen is much powerfull

[Hewitson]

Originally Posted by Starglider 2

Okay that's a start. Would it be as simple as:

3600 MHz x 8 cores = 28,800
7.16 / 28,800 = 3,840 times faster?

Definitely not. Clock speed is not relative to CPU speed at all these days. In fact it never was really, as different CPU's take more cycles to execute instructions than others.

Originally Posted by Aladin

ryzen is much powerfull

What a useless post.

[Starglider 2]

Okay any idea of the answer?

[khph_re]

It's also multi threaded 64 bit vs 16/32. will be hard to work out.

[Starglider 2]

Originally Posted by khph_re

It's also multi threaded 64 bit vs 16/32. will be hard to work out.

I know, that's why I've come to the pros

[meynaf]

One cannot tell cpu x is n times faster than cpu y. It all depends on what said cpus are doing. F.e. it could be 10000 times faster for task a and 100 times faster for task b.
In the same way, having 8 cores doesn't make it 8 times faster. Actually, it won't make the cpu faster at all for programs that aren't explicitly written to use that feature. Similar story for being 64 bit.
So there is no precise, definitive answer to this question.

In addition, the amount of work they have to do isn't the same.
68k is slow cpu running fast code, todays x86 is fast cpu running slow code.

[DrBong]

Originally Posted by Starglider 2

Hi, for a future video I'm interested in knowing how many multiples faster this iMac processor is than an A500 with its 68000:

3.6GHz 8-core 9th-generation Intel Core i9 processor

I'm guessing x2000 but I'd love to work out a way to know for sure.

Originally Posted by meynaf

One cannot tell cpu x is n times faster than cpu y. It all depends on what said cpus are doing. F.e. it could be 10000 times faster for task a and 100 times faster for task b.
In the same way, having 8 cores doesn't make it 8 times faster. Actually, it won't make the cpu faster at all for programs that aren't explicitly written to use that feature. Similar story for being 64 bit.
So there is no precise, definitive answer to this question.

In addition, the amount of work they have to do isn't the same.
68k is slow cpu running fast code, todays x86 is fast cpu running slow code.

Exactly. As Meynaf has said, the answer (and any type of calculation involved) is by no means trivial.

[rothers]

If you're using Javascript in the browser it's not a whole lot faster than a bare metal Amiga

[Bruce Abbott]

Originally Posted by Starglider 2

Hi, for a future video I'm interested in knowing how many multiples faster this iMac processor is than an A500 with its 68000:

How long does it take to load a game from floppy disk? My A500 takes 2 seconds.

[rkidd7952]

Originally Posted by Starglider 2

Okay that's a start. Would it be as simple as:

3600 MHz x 8 cores = 28,800
7.16 / 28,800 = 3,840 times faster?

As others have mentioned, the value depends on what the benchmark is, clock speed isn't comparable, yadda yadda.

But I think this calculation is good enough for a ballpark. You could throw in another factor of two or four to account for the difference in word size. Divide by eight in some cases to recognize that not every task can be split across cores. There are so many variables that reporting more than one significant digit is meaningless. Calling the iMac 3-4 orders of magnitude faster is about all you can do.

Now if you load up Lightwave on both and measure render time, that would be interesting.

Robert

[Pyromania]

But does it blend? I believe it would be a lot easier to extract the juices of an iMac in a modern blender vs an A500 which would be much harder to blend.

[Starglider 2]

Originally Posted by meynaf

F.e. it could be 10000 times faster for task a and 100 times faster for task b.

Well I think finding an average figure would be a reasonable outcome. In your example the average would be 5,050.

This is for a chess match and I want to set UAE to run at - on average - the same speed as the Mac.

[Mojost]

What about running something like povray, there is a version on aminet and you should be able to compile the same version on the Mac.

[paleboy]

I would ignore multi-threaded performance and see if you can get some kind of synthetic single-core benchmark numbers of each. You can just qualify it with "and that's just single-core performance - with a multithreaded workload it'd even faster".

[Rotareneg]

Originally Posted by Starglider 2

Well I think finding an average figure would be a reasonable outcome. In your example the average would be 5,050.

This is for a chess match and I want to set UAE to run at - on average - the same speed as the Mac.

What computer do you plan on running UAE on that you expect a chess program running on the emulated 68000 CPU to be as fast as a chess program running directly on an i9 9900K?

[nogginthenog]

Originally Posted by Mojost

What about running something like povray, there is a version on aminet and you should be able to compile the same version on the Mac.

That's a bad test. An A500 does not have a FPU.

I thought about BogoMIPS but that's not great. Dhrystones maybe?