Most of you know that there is a CPU benchmark of the game Legend of Zelda: Wind Waker.
The thing is that I thought I could cooperate in the Dolphin community making a CPU benchmark of that game (Luigi´s Mansion). I´m also planning to make one for Mario Kart: Double Dash.
But first, I would like to wait for experts´ ideas and opinions (such as if it´s a good idea, if it´s the correct moment to do this...) and then start officially the benchmark.
Biggest question in my mind is "why?" We already have a benchmark that's given us some pretty amazing results. Adding different games wouldn't really add a lot of value, since the results will show us what we've already discovered with the Wind Waker CPU Benchmark.
What would be useful would be a GPU benchmark like we use to have using MK Wii. It would help since we have a lot of newer GPUs available that weren't able to make it in the old GPU benchmark.
I think if another benchmark is going to be done, it should be done after the next official stable release, and with a game that is extremely demanding. That is, if we need another benchmark at all.
I don't think it should be done personally for the same reasons that shonumi stated. A GPU benchmark would indeed be more useful. But if it were to be done this needs to be planned out in detail the same way starscream's was. That means build, settings, OS, game, memory card locations, and a method of organizing the results. Preferably use dolphin 3.5 and use the built in benchmarking feature this time.
And with a demanding game as Starscream said. I guess SMG (1 or 2) would be a good choice
(08-26-2013, 08:20 AM)Starscream Wrote: [ -> ]I think if another benchmark is going to be done, it should be done after the next official stable release, and with a game that is extremely demanding. That is, if we need another benchmark at all.
The main reason that I see for another benchmark would be to remove the human error from the results recording. Perhaps have the bench mark run for a set amount of time and log the FPS to a file instead of trying to get a screenshot of what the user thinks is an average FPS.
It would be more accurate. But not by much since the variation between min and max fps was very small in the WW benchmark. You still have different drivers and background tasks effecting performance to account for unfortunately. But I agree that it's definitely an improvement over our current method.
From my experience, the WW benchmark tends to vary about 10 fps depending on what's on-screen
Yep. About 5%. And about half of that is from background tasks in most cases.