Loading time
6 years ago
France

Hi,

I compile data and find an interesting result. Now this is a small sample size, need to be confirm with time.

Here it is, I simply deduct total split time with total racing time (like time trial) (I used an online chronometer).

48 tracks loading time based on system :

  • EU digital : 19m05s (taken from Mystwalker)
  • Jap digital : 19m05s (estimated time based on Toshiki bonus tracks)
  • EU cartridge : 20m00s (taken from GlockeLP)
  • Jap cartridge : 20m50s (taken from Tossy_XD)

Also Luigi_fan2 loading time is 19m22s (Canada digital) . Not sure if this info is relevant (he starts his run on rainbow road which might affect some loading times).

edit : calculation on jap digital

Edited by the author 6 years ago
AeonFrodo and MystwalkerMX7 like this
Netherlands

Thanks for the data! Loading times are still random in terms of loading new tracks. I made a post about loading time on MK8 a long time ago. They're on the forums if you're interested.

France

An update :

Here are Mystwalker last 48 tracks No items runs : I'm not timing the loading time, but each 3 laps races only (as if it was a time trial). Total time on the left, loading time on the right

01:28:08,00 - 19:17,02 01:27:57,00 - 19:08,29 01:27:06,60 - 18:59,55 01:27:02,00 - 18:53,16 01:26:57.00 - 18:48,37 (edit on July 17th)

Based on those data, we can assume that the switch is getting faster to load tracks. A theory could be that there are lesser and lesser slow loading tracks (you know once in a while there is a track that loads around 4 seconds more than usual).

I can't seem to find a pattern on why there are slow loading tracks sometimes. If anybody have a theory or ideas to share, thanks.

Edited by the author 6 years ago
Portugal

The only instance of slow loading I've noticed was the transition from Music Park to Yoshi Valley, where the game stays stuck for a few seconds before entering the loading screen. It happened 2-3 times to me so far.

It's pretty weird how the amount of idle time in the speedrun is getting lower and lower like that, though... I really don't see how I'd save 24 seconds at all besides perhaps me getting better with mashing/timing A to skip the intro cutscenes and "Next Race" button, but I don't feel like there were many instances of that in the older runs.

Also, I do believe there's variance in loading times from run to run, but besides the loading time of the very first track (quitting on first track and restarting VS Race seems to give slightly better loads) I've rarely felt affected by loading differences in No Items before. So yeah, quite intrigued by those results...

Maybe the recent Nintendo patches helped improve downtime a bit?

Oh right, speaking of loading times, here's Quinn37's small spreadsheet about the difference between his Cartridge run and my Digital run (from June 23rd): https://docs.google.com/spreadsheets/d/1zbBwgEfNo49H9owiY7P0nmsVrQnvEoqjFklW_7jsfq0/

The data here's a tad different (timings go from A press at "Next Race" to A press on track cutscene), but it seems to show there's roughly a 1 minute difference between Cartridge and Digital when accounting all the tracks in the game. It doesn't account for any possible PAL/NTSC/JP differences besides this one specific comparison, though.

Edited by the author 6 years ago
France

Yes, I believe that the decrease of loading time is due to nintendo's patch or firmware update. I can't seem to properly time the same way Quinn37 did. My times are inaccurate and few tenth of seconds can be misleading when you multiplied that by 48 tracks.

With my method, I end up with a gross total loading time that isn't accurate either (player sometimes press A late). Also, timing that way give a margin of error around 5 seconds give or take in total.

Loading time is a technological part of the speedrun and should be studied to reduce it to the minimum.

British Columbia, Canada

so times are 1:09:xx. we should have w/o loads times

Portugal

^ Any ideas on how to implement that in LiveSplit then...? It's not as straightforward as it seems.

And yeah, I helped Quinn with the timings and I had a really hard time getting accurate measures, even with VoDs running at 0.5x or 0.25x speed. A presses felt really inconsistent to time. Maybe counting downtime instead of loadings might be the way to go, from Next Race to GO!.

France

After a quick search on google, it looks like we need an auto-splitter for MK8D. Not sure how to do that, but there are many games on speedrun that have auto-splitter available. That way livesplit will pause for each loading time. That would mean also to add a column for "no loading" speedrun.

I wish a coder can develop an auto-splitter for this game, because I like that idea of "no loading" speedruns.

edit: after further research, looks like auto-splitter are almost impossible for console games. Makes sense.

Edited by the author 6 years ago
France

Added latest Mystwalker WR run to first post.

Here is the total loading time between the tracks. First one start at Ok and finish at GO in N64 RR. All the others start at finish line untill GO.

00:17,86 00:22,15 00:23,90 00:23,64 00:23,33 00:23,46 00:23,41 00:23,14 00:24,07 00:23,94 00:23,62 00:24,00 00:24,71 00:24,59 00:23,33 00:24,12 00:23,75 00:23,59 00:23,78 00:23,66 00:23,52 00:23,25 00:24,04 00:23,53 00:23,97 00:23,93 00:23,74 00:23,49 00:24,07 00:23,98 00:23,54 00:24,15 00:23,96 00:23,35 00:23,39 00:23,59 00:23,52 00:22,74 00:23,10 00:23,47 00:23,34 00:23,22 00:23,46 00:23,95 00:23,98 00:23,19 00:23,54 00:23,22

Edited by the author 6 years ago
Pennsylvania, USA

That's disappointing the cartridge load times are that bad. I kinda don't want to have to trade in my physical copy for the digital one.

Belgium

Loading time is awful because i've got the worst loading time in my run 48 tracks xD

Portugal

The difference in loading times on an track-by-track basis isn't as bad as it seems like, but yeah, it's a bunch of small numbers that do add up greatly the longer the run is.

Here's a lowdown on the loading differences I've noticed so far. Bear in mind that these numbers are estimates, they aren't written in stone and thus can vary somewhat in-between runs:

-> On first track load, Digital gains about 3 seconds over Cartridge. -> On subsequent track loads, difference seems to be around 1 second per load or slightly more depending on track loaded. -> The loading time difference between mine and Quinn's 48 Track No Items run, using a different timing method from Haji's, accounted for about 60 seconds. -> That means that for a 16 track category, Cartridge might lose about 18-20 seconds compared to Digital. -> And about 37 to 40 seconds for 32 tracks.

You can use these values as a reference to compare Digital to Cartridge runs on the leaderboard, and vice-versa. But again, it's a rough estimate!

That said, making the switch from Cartridge to Digital is something that I'd probably only recommend if you're hitting a limit with Cartridge and really want to optimize the time even further. Otherwise Cartridge is perfectly fine to work your way up. :P

Lastly, these numbers are for No Items. I'm unsure if the same exact estimates would apply for Items runs. I do have the suspicion that loading times are faintly worse in Items runs, likely since the game loads additional objects like Item Boxes into the track.

Edited by the author 6 years ago