Help wanted! System power scale

[Felsir]

I'm working on the BYOAC Wiki (together with PL1), a lot of work went into compiling the FAQ.

On of the questions that pops up a lot on the forum is "will my PC run emulator X or Game Y?" so I think that question warrants a topic in the FAQ. So I was thinking, how can we give that kind of information? So I came up with this idea:
Emulation vs CPU chart.

The idea is, if we collect some data points of CPUs people have and what they can run, one might be able to plot out their own system in the chart. For example, my cab has an Intel i3 3220 and it manages to run PS2 emulation quite well. I think it's the toughest I've thrown at my system so far.

So I'm looking for data points: what system do you have and what is the "heaviest" game/emulator you can run? I've added a rating with 1 being trivial to run and 31 being the hardest emulated game (it might be an idea to have a separate scale for MAME games).

Note that it is a relative scale, so my initial ranking of the PS2 emulation is not set in stone.

If you have a suggestion of how to display this information differently, feel free to comment!
 

[BadMouth]

I tried something similar once upon a time: http://forum.arcadecontrols.com/index.php/topic,130988.msg1342049.html#msg1342049

Emulators other than MAME, especially Demul rely heavily on the GPU.

[BadMouth]

All my current systems are running 3.4Ghz Athlon X3 with 9800GT video cards and 64-bit OS.

They can run Gamecube on Dolphin and Dreamcast on NullDC.
I've never tried Dreamcast on Demul, but the arcade games are hit or miss.  The 3D shooters have slowdown and sound skips when there are a lot of enemies, so I'm running those in Makaron.
Chankast is older and probably has lower required specs.

I haven't tried a PS2 emulator, but I've read that the 2D games aren't that demanding and the 3D games are very demanding.

For the sake of preventing things from getting muddy, I'd pick the most popular emulator for each system and only use that as a reference.

[paigeoliver]

Different emulators and different versions of the same emulator can have vastly different hardware requirements without any noticeable difference to the end user.

[Haze]

Different emulators and different versions of the same emulator can have vastly different hardware requirements without any noticeable difference to the end user.

Until the game blows up on the last boss or something, due to a bug that was fixed years ago...

this is very true of the emulators that have developed rapidly and are for complex systems, or where protection is involved, the more mature but slower versions or newer emulators tend to be a lot more dependable in the long run.

if of course all you want is to be able to boast a huge game list and wow people with the first couple of levels of a game, that's not sometihng you should need to consider, but if anybody dare tell me that the last levels of Fighters History are broken (for example) they will get a very evil stare. :-)

[PL1]

I understand and agree with Haze's position that the latest version is the best to use when setting up a system, but people who aren't arcade purists or are on a tight budget might not want to buy a computer powerful enough to run the latest version.

Regardless of the reasoning, the core question is still either "What CPU do I need to run MAME version X?" or the reverse "What version of MAME runs on this CPU?"

Maybe list examples of various generations of processors (AMD+Intel) and the highest version of MAME that they will run at 100% for three general categories.

Sample entry (data is not accurate)

AMD Athlon X2  3.0 GHz
1.) Classics  - 0.152
2.) Majority (95%?) of games - 0.142
3.) All games - 0.107

I wish there was a program that could run automated tests like Passmark's CPU Benchmark.   


Scott
    Post #3000 already??   

[ark_ader]

Maybe a drop down wizard to compare CPU and Emulator would be better?

[kahlid74]

I like the idea but I would want to see numbers.  I'm a numbers guy, so I would want to see a frame rate and a rating system.  Like 200 FPS with Good in green next to it.  That way for the basic user they could see a good/green and be set to roll and for a more advanced user they could see frame rate and decide accordingly.

Do also remember, that motherboards and memory settings should be collected.  You can have the same CPU and different numbers due to subpar/superior motherboard and memory speeds.

As has been mentioned above, different versions of emulators do different things so if we just started collecting information and kept building the database we could have a pretty robust database the longer it is allowed to gather/grow.

Also, a standard should be created for testing, not "Oh, it looks moderately good to the naked eye" but something anyone can do to test.

Scott
    Post #3000 already??   

Nice!

[PL1]

Also, a standard should be created for testing, not "Oh, it looks moderately good to the naked eye" but something anyone can do to test.

IIRC, the easiest way to measure if a game is running at 100% is F11, but a downloadable benchmark program that allows you to select an emulator/version from a pulldown menu, run the tests, then see and/or upload the results would remove quite a bit of human error.

Not sure if this type of benchmarking utility would be worth the time investment for someone with the coding chops (definitely not me  ) to write it.   


Scott

[stigzler]

This is a great idea and wish I'd had something like this when building the pc for the cab. Instead, it was a bit hit and miss + now have numerous motherboards, cpus and GPUs sitting around!

[SGT]

I'm working on the BYOAC Wiki (together with PL1), a lot of work went into compiling the FAQ.

On of the questions that pops up a lot on the forum is "will my PC run emulator X or Game Y?" so I think that question warrants a topic in the FAQ. So I was thinking, how can we give that kind of information? So I came up with this idea:
Emulation vs CPU chart.

The idea is, if we collect some data points of CPUs people have and what they can run, one might be able to plot out their own system in the chart. For example, my cab has an Intel i3 3220 and it manages to run PS2 emulation quite well. I think it's the toughest I've thrown at my system so far.

So I'm looking for data points: what system do you have and what is the "heaviest" game/emulator you can run? I've added a rating with 1 being trivial to run and 31 being the hardest emulated game (it might be an idea to have a separate scale for MAME games).

Note that it is a relative scale, so my initial ranking of the PS2 emulation is not set in stone.

If you have a suggestion of how to display this information differently, feel free to comment!

  I really hope you can make this happen.  It is sorely needed especially if it can be maintained as Mame and CPUs evolve.

[Felsir]

I'm still thinking of a way to collect the data that is easy for everyone...

[stigzler]

How about compiling 4/5 games per emulator with lowest taxing to highest (also allowing a mix of CPU tax vs gpu tax)
For example, for PCSX2, I'm guessing Gran Tourismo 4 is quite demanding.
Guess you could go to the individual emu forums to get your "list of 5."

Then you could ask users to rate each game for things like:
- Framerate consistency vs Freeze
- Audio/Visual burps vs high fidelity

Lastly, user could submit their PC specs via CPU-Z validator to summerise their specs (and also verify - not that anyone would ever post false specs or anything *coughs* yooootube)

 Dunno how you'd collate this in a wiki - guess you'd need some kind of database functionality behind it.

So, for me at present, can run all PCSX2 games (inc GT4 and Shadow OT Collossus) and all Gamecube Dolphin games on:



Haven't tried Wii properly yet.

Hope you get this working. Like I said - it would have been invaluable 8 months ago!