I was trying to tell a person about dosbox cycles.
I want to explain how slow 1000 to 3000 cycles is to him.
Is this not equivalent to a 386 running at 8mhz to 25 mhz ?
I want to explain it to him in a way that makes sense.
So what would be the speed in Mhz of 1000 to 3000 cycles.
Conquer the FCC GROL, & Amatuer Radio test
They are not really related...so, 1000 is slower, 3000 is faster. There you go.
This is something i also wonder sometimes... i mean what is the computer equivalent of the cycles is? I am sure it would be easy to add the option to show it even if accurancy is not the goal of dosbox, at least show what it roughly compare to.... like like when cycle down/up in dosbox it nor only would show the cycle but also the system speed, so like "***** cycles" equal ~386 25mhz or something.
But i guess you can ge that info by using a benchmark utility... although lazy people (like me) would appreciate that it is shown by dosbox 😀
This definitely needs to be stickied...
The way Dosbox works you cannot really say that x cycles approximates x mhz. It depends too much on the host PC AND the game being run. If the game is very taxing in graphics or whatever else it can dramatically shift the ratio.
When benchmark x says dosbox on your machine is like a Pentium XXX, another benchmark on another PC with the same cycles might claim something else entirely...
Hence why there is no such chart given by the devs.
Ya, the MHz equivalent of the CPU Cylcles in Dosbox depends on the host CPU.
My Core i7 720QM notebook shows the speed of the CPU as around a 486DX "66"MHz when the CPU Cycles is set at 18000 and the CPU Core to Dynamic in Dosbox 0.74.
Used PC Tools 9.0's System Information Professional, which I use in my actual 486 machine.
I'm not sure if the OS and other components (RAM, Cache, etc.) in the host can affect this perceived MHz reading.
it doesn't depend on the host.
The maximum number of cycles depends on the host, but a cycle itself not.
cycles are number of emulated instructions per millisecond.
If you want to know how that relates to MHz, then look up the instruction specific timings of a specific cpu that you have in mind.
Water flows down the stream
How to ask questions the smart way!
IIRC DOSBox is reading/executing any x86 instruction in 1 cycle. Vanilla hardware use 1-4 cycles per instruction...
So - you can't translate cycles to MHz even in theory.
wrote:IIRC DOSBox is reading/executing any x86 instruction in 1 cycle. Vanilla hardware use 1-4 cycles per instruction...
Could be much more actually, depending on the processor (DIV can be 150+ cycles on 808x)... but yeah, not very comparable. Plus a system's overall speed depends on much more than the CPU.
Lassar: TOBENCH is good for doing what you want, since it compares the measured performance against a database of results from real systems, and gives you close matches (which update in real time as you adjust the cycles in DOSBox). Though as you will see, you still won't get a simple cycles <-> MHz correlation.
wrote:IIRC DOSBox is reading/executing any x86 instruction in 1 cycle. Vanilla hardware use 1-4 cycles per instruction...
So - you can't translate cycles to MHz even in theory.
This means we can very accurately tell a range 😀
