As announced a while ago, we'll add a layer of scrutiny for emulator submissions. Since most runners have already adopted using speedrun-specific functionality of BGB or Gambatte-Speedrun, nothing changes for those. If you have not familiarized yourself with that topic yet, feel free to ask in the SML2 Discord or contact me. We're happy to assist you with setting up.

Here is the updated rule text that is binding starting now:

[quote]This board is for SML2 only. SML2DX has its own board.

Runs on the practice rom are not eligible for submission, including individual levels.

SGB and 3DSVC: both platforms run the game too fast, so times are converted for the leaderboards.

Emulator submissions: please include which emulator you are using! As of 2021, the only allowed emulators are BGB and Gambatte(-Speedrun) on their latest versions.

For BGB, activate speedrun mode in the settings and make sure that your submitted run starts from a reset that shows the speedrun mode splashscreen!

For Gambatte, either use Gambatte-Speedrun and make sure that your submitted run starts from a reset that shows the speedrun mode splashscreen

or, if using regular Gambatte, capture the entire emulator window and make sure that your submission starts from a reset!

If you have any questions, reach out to the community before you commit to recording runs!

We reserve the right to reject obvious joke submissions, such as runs that are clearly not speedruns. Please don't waste other people's time![/quote]

If you have any questions or remarks, please feel free to bring them up!

Hi @chiteiro1,

Your run is of R-Type II for the Game Boy, not R-Type DX. Both boards exist on the site:

https://www.speedrun.com/r-type_ii_gb https://www.speedrun.com/r-type_dx_gbc

Please submit your run to the correct board! This board will be deleted in 7 days. Let me know if anything is unclear.

Cheers.

Hello everyone!

It's about time we make the leaderboards and rules a little prettier. This thread is meant for discussing changes to the rules and this post (or the next one) will contain a work-in-progress version of what will eventually become the official phrasing. Everyone is invited to share their thoughts and make suggestion as the rule set takes shape.

Here's a little overview of what might need an update:

#timing method across all categories You may have noticed that, currently, there are 3 different timing standards for ILs, Any% and everything else.

• everything else: So far, times have been truncated after the first decimal. This decision was made several years ago when the average video material didn't justify anything close to frame accuracy.
• Any%: Times are given in milliseconds, based on video frames.
• ILs: Times are given in milliseconds, based on an approximation of game frames

The plan is to standardize the timing. The average video quality of submissions has become reasonably good enough to justify a higher accuracy, at least in my opinion. It would still be an approximation and it won't necessarily be 100% "correct", but it'll be consistent across categories and runs. My suggestion is that we use a time conversion based on actual GB frame cycles (59.7275Hz). This would not concern submitters, but only leaderboard moderators, but I'll still share how I'd suggest the conversion should work: [quote](time in seconds and milliseconds) x GB framerate = number of GB frames <- round to full frames <- divide by GB framerate <- round to 3 decimals Example: video segment: 12:25.637 = 745.637s x 59.72750057 = 44535.03434 frames <- rounded to 44535 frames <- divided by 59.72750057 = 745.6364250133898 <- rounded to 745.636 = 12:25.636[/quote]

#rules standardization across categories The site offers the opportunity to separate game rules and category rules now, so I'd like to make use of that. Rules should follow a common structure, so that differences between categories are easy to find and understand. The phrasing should be more specific in terms of what a category entails and what's specifically (not) allowed. #emulator rules The rules should state more clearly that we only allow BGB and Gambatte (Speedrun). Whether we extend those rules to always only allow the most recent versions remains to be determined, since several runners have already stated that there might be differences in performance between versions. As for the aspect of proof of legitimacy for emulator submissions, I suggest using either the speedrun-specific versions (Speedrun Mode enabled in BGB / using Gambatte-Speedrun) which will show a disclaimer upon each reset OR a full-window capture of the whole emulator. The former would be preferred (by runners, too, I'd assume) which would entail that every submission should include the reset before the run starts. Further, I think it needs to be specified that offline runs will be subject to more scrutiny and that further proof can be required in particular cases. So far, we've been lucky that there haven't been any questionable submissions yet (at least to my knowledge), but we should be ready just in case, imo. #category names This is a very minor point, but I think it's important to open it up to the community. There are varying opinions on category naming in various speedrunning communities. My own suggestion would be to rename "Any% Glitchless" to just "Glitchless". Other opinions and suggestions are welcome.

Sorry if I ticked the wrong platform. Anyway, my recent 300k run was done on SGB2 and doesn't need a time conversion. Feel free to remove this thread after fixing the run. Thx.

I know another thread on this topic already exists, but that one got off-topic so quickly and didn't reach the correct conclusion, so I decided to open the topic anew.

I feel really stupid for only realizing this now, but version 1.0 of the game has a few very relevant differences from the much more common version 1.1. First of all, you can tell the versions apart easily by checking which song is playing as A-Type: Minuet = 1.0, Korobeiniki = 1.1.

As mentioned in the other thread, but not to its full extent, v 1.0 allows for inputs during line clears. What wasn't stressed there is that that includes preloading DAS, which is of course relevant for DAS players. However, the advantages don't stop there: the level up progression is very different in v 1.0. Starting on level 9, your level progresses every 20 lines, i.e., level 10 at 20 lines, level 11 at 40 lines, etc. That, of course, also means that line clears are worth more earlier in the run, which is VERY relevant for the 300k category. Whereas in version 1.1 300k is equivalent to clearing 100 lines with only tetrises (ignoring pushdown), you have much more leeway in version 1.0 and can even realistically reach 300k before arriving at 100 lines (all tetrises would amount to 360k). Another small advantage is the speed increase that comes with the earlier level-ups, of course.

I didn't see any of the top runs making use of this, so I assume it's not really known or shared in the community so far. I'm baffled that I did never notice this myself before today.

I'm not sure how popular this would be. I'm indifferent myself and would just add it if demand is there. My idea would be...

• one open category where you choose whatever level and powerup you like to start with
• one category with the restrictions of Any% Glitchless since it's by far the most relevant category

Possible ruleset:

• Time starts when entering a level (1st frame when Mario raises his arm on the zone map)
• Time ends on the first frame of the fade-out after reaching the end of a level or defeating a boss
• It will be attempted to time runs down to the frame (hint: recording in 60fps is encouraged)
• Runs are restricted to the regular exits (exception: Macro 1 in the Any% Glitchless category)
• Invincibility stars from killing 100 enemies are not allowed (exception: Wario's castle in the Any% Glitchless category)
• Carry-over glitches are not allowed (pipe glitch, bubble glitch, bell glitch, star glitch)
• Zips may be exploited in the open category (not sure if this is even relevant)

Please let me know how you think about this and feel free to suggest amendments or changes to the restrictions. I admit that I haven't spent a lot of time to think this through. :-p

I'm not sure what exactly I want to do with this, so for now it'll just be a loose collection of documented "historical" milestones of SML2 speedrunning.

#SDA

[to do...]

#TSSB

###Episode 11

Any% in 3:42 by Mugg

###Episode 46

Glitchless in 27:44 by Mugg

###Episode 48

Glitchless in 27:27 (or 27:15?) by controllerhead

###Episode 49

Glitchless in 27:05 by controllerhead

###Episode 51

Glitchless in 26:50 by Mugg

###Episode 52

All Levels in 39:13 by Mugg

#On-site marathons

##AGDQ

Laplacier 2014

dadinfinitum (slashinfty) 2018

TASbot 2016

##ESA

Oh_DeeR 2017

LateLearner & Oh_DeeR 2018

#Online marathons

[to do...]

• number 1
• number 2
• number 3

FrankerZ

5 of the 6 games in this series have completely empty boards. Please either add runs or delete them. Adding games just for completeness is not a good practice.

This board has been up for over 3 years and yet there is not a single full-game run on it. Nowadays, a game with only individual level runs wouldn't even get accepted in the first place. I strongly recommend getting full-game runs on the board. Many mobile and web game boards are poorly set up and don't meet the standards of boards of more conventional platforms. Since you're even tracking the DS version here you should try to meet those standards even more. As it is now, the board sets a bad example for mobile/web games on the site.

It's been over 3 weeks since this board was accepted. All categories are empty and no rules have been defined. Please set up the board asap or it will be deleted.

Hey there! SML2's 26th anniversary is drawing closer and with that it's been at least 1 year since I and a few others came up with a few fun speedrun categories. They'd be listed under a "Misc" tab, together with Small Mario only. Some of them are at least as interesting as some of the main ones, imo. It turns out that if there's no leaderboard for them, nobody's even aware that they exist and that's the reason for this thread. I'm gonna list a few candidates, including a short explanation for each and then you can vote here:

http://www.strawpoll.me/16578828

Jackpot% ...is a short and (arguably) fun category similar to Any%. The goal is to get to 99 lives as quickly as possible, utilizing the 999-coins pipe in the casino. Glitches are allowed.

999 Coins (no level repeat) ...is by far the most diverse category. The goal is to collect 999 coins as quickly as possible without entering levels more than once and without using glitches. What makes this particularly fun to attempt is the wide variety of possible routes. I am faily certain that the optimal route has not been found yet.

Low% Coins ...is another glitchless category (because it would be trivial otherwise). The goal is to beat the game while collecting as few (regular) coins as possible. It is not possible to avoid coins altogether which is why I think this category should be timed dynamically with a fixed penalty of 10 seconds per collected coin.

No Shoot/Run Button ...is rather self-explanatory. It's also intended as a glitchless category and assumes that you never press the button that lets you run and shoot until you defeat Wario.

All Bosses ...could also be called Beat Wario or Any%(no credits warp). The goal is - surprise: to beat Wario after having gained access to the castle by beating all 6 bosses. Glitches are allowed.

Pacifist ...is another self-explanatory and glitchless category. The goal is to beat the game with exactly 5 kills on the kill counter (killing the 5 floating faces in the castle cannot be avoided, boss kills don't add to the count). Whether this category should be fixed to that count or a per-kill penalty should be added, comparable to the Low% Coins category, is up for discussion.

Feel free to suggest your own ideas and/or comment on my suggestions! And, of course, don't forget to vote in the strawpoll!

Hi, following up on my other post, I wanted to ask what the community thinks about adding the starting level as an extra column for the 100 lines category (and possibly the 300k category). I think it would be informative and provide a better overview of the board.

Hi there, I was wondering why I don't see any L9♥ runs on the board. I saw that the rules state "Player can start from any level 0-9." Is that on purpose or an oversight? If it's on purpose, what's the justification for excluding ♥ level starts? It should clearly be allowed, right?

Hi there,

we've just added a series page for V-Rally: https://www.speedrun.com/v-rally

@thebpg13, @CoopeZz, are you okay with adding the game to the series? Are you alright with bangerra as a series mod? One of you could be added as a series mod, too, if you desire.

As far as I can tell, there is zero speedrun-relevant content on this board (I didn't check every single category, so feel free to point me towards anything that I've missed). I strongly recommend reorganizing this board to be more speedrun-relevant. At least the main categories should be actual speedruns. Although, personally, I'd get rid of all the scores, you could still keep them under Misc., if really needed. Also, @JayKayGame, most of your social media links are expired. Please update them so users can reach out to you if they have questions or suggestions.

Hello there. I have recently taken over as a content moderator, and as such my focus lies particularly on the GB library.

Now, it's no secret that this leaderboard, in particular, is not easy to handle. There's the issue with the different platforms running at different speeds, there are runs potentially done on VBA, etc., many mislabelled runs and, last but not least, many runs without video proof.

The latter point is a philosophical one. I am aware that you require runs under 12 minutes to have video proof these days (which, however, is not universally applied to the current runs on the board). Personally, I find it rather outdated to not require video proof for ANY submission, since almost nobody should have trouble with at least pointing a webcam or a phone camera at a screen. Submissions without proof just carry less meaning and water down the integrity of the board overall. All that said, this is completely up to you as a community, though.

The other points are less ambiguous:

1. I'd like to ask of you to discuss time conversions to a common standard (preferrably, the standard that virtually every other GB game community uses, which is the GB's refresh rate of ~59.7275 Hz, which is universal to all official hardware releases - except SGB - and the accurate emulators). I have provided you with a raw conversion model for how to potentially deal with certain emulator runs (Dream Collection and 3DSVC), based on swordsmankirby's video footage, but have yet not seen a lot of response to that. As a runner, I want to be able to compare my own achievement to the others on the leaderboard on an even ground. Everything else just creates needless tension. It is unfortunate for us that Nintendo's focus is not on accuracy for their emulators, but that's how it is. We, as a speedrun community, have our own priorities and are able to fix the issue for our purposes.

2. I shouldn't even have to mention that runs done on VBA should not appear on a GB leaderboard.

3. The integration of platforms is not exactly well-done on speedrun.com, even at this point in time. On top of that, many GB games were labelled as GBC by default, initially. So the mess we see here is mostly due to that. We're not stuck with this, though, and it's also not the sole reason of mislabelled runs. For Super Mario Land 2, I use a workaround with a custom variable to at least differentiate between the emulators used. It's still not pretty, but it clears up things a lot more. The original platform variable was meant to show the platform the game was released for. Naturally, it was later used to specify which platform the run was PLAYED on, which lead to confusion and to missing options for emulator submissions. A few examples of misnomers:

• GBC[Emu]: should probably be GB[Emu]. It's arguable whether Gambatte could be labelled GBC[Emu], since it uses the GBC palette by default. Going back to the initial point, the platform variable was intended to specify the release platform and both Gambatte and BGB, of course, use their respective GB core to emulate GB games. 2 strong points to support GB[Emu]. Too many words for this almost trivial point, sorry.

• GBP[Emu]: doesn't make any sense

• Wii/WiiU/3DSVC: should be Wii[Emu]/WiiU[Emu]/3DSVC[Emu]. It's software emulation. Period. Doesn't matter if official or something else.

• GBA[Emu]: that would make me think of VBA, probably not what was intended during the submission

• GBC: most of the time probably just left on the default option and/or forgot to tick the [Emu] flag.

So, now it's your turn as moderators and as a community to make this leaderboard a place where people can find accurate information and compare their own efforts against other's. I will help where I can and if you want me to do the cleanup myself instead, I will. Before that, though, I'd like to see an open discussion about the points I brought up above. Moderators, active and inactive runners, people from other communities: anyone's participation is very much appreciated.

Hey there! I manage a few Game Boy game leaderboards on this site and recently have come across a few run submissions done on the 3DSVC. It quickly became clear that that platform was running the games too fast. Sadly, Nintendo doesn't seem to care too much about accuracy for their various emulators (=Virtual Consoles). In my opinion, speedrunning should be about comparing your skill at a game with others and not a race of arms of who uses the faster platform. This is a general statement and not meant to discredit anyone's efforts! E.g., the Game Boy community handles runs done on Super Game Boy according to that comparability approach and times are converted following this example: https://kb.speeddemosarchive.com/Super_Game_Boy_timing Out of my own interest and after talking to eBloodyCandy about the situation on the KDL leaderboard, I had a look at some data provided by swordsmankirby, where he compared clockspeed and load times between GBP, 3DSVC and the Dream Collection (on WiiU?). You can have a look at my findings here:

https://docs.google.com/spreadsheets/d/13_Le7b0KgGA1tT_x-ta2Yfv5PrLWViauZ26XbVzgKhY/edit?usp=sharing

The speed differences on the 3DSVC are consistent with what I had previously encountered in submissions of Super Mario Land and Super Mario Land 2 so I'm confident that that data is correct. I'm not sure whether the Dream Collection footage was recorded from Wii or WiiU. A comparable recording from the missing platform would be nice. Data for the load times is a little slim and more and longer segments should be recorded to fill in the gaps. One last thing that I have not looked into yet (simply because it's so hard to quantify) is lag. Lag is a MASSIVE factor in KDL so a well-designed comparison could be interesting. All that said, I do not own any of the systems in question myself, so I can't collect any data. I just wanted to share my findings and let you, the KDL community, decide how to handle this. Cheers!

I've thought about the optimal compromise for a bit and here is what I suggest for timing SML runs:

We've found out that the initial load time between pressing start and gaining control over Mario differs between SGB2 and all other systems. The difference is 8 frames in favour of SGB2 (see the other thread for how this number was determined). However, I wasn't happy with settling for one system as a standard over the other. I also didn't want to introduce a new standard in general. Luckily, the solution is very easy and we can apply an already existing standard to SML because the current timing has already been very close to that standard all the time. I'm talking about timing character control as used by SDA. We had already used the frame on which Tatanga explodes (or any of Mario's projectiles unload, alternatively) as the last frame. That frame coincides with the loss of control over Mario as per SDA standard, meaning the end point doesn't change one bit.

On SGB2, character control starts 8 frames after Start is pressed. On all other accurate systems it starts 8 frames later than that. The time between gaining control and the ingame timer ticking down from 400 to 399 is equal on all systems, of course. The timer switches to 399 exactly 41 frames after the player gains control over Mario. That's 1 / 59.7275 ¤ 41 = 0.686 seconds. So for uniform timing I'd use the 1st frame the timer shows 399 as the start point. The end point is the frame on which Tatanga explodes. Then you just add 0.686 to the time of that segment to get the total length of the run.

Another topic is timing accuracy in general. Currently, speedrun.com offers resolutions of either full seconds or milliseconds. I'm in favour of using 1 decimal since due to encoding artifacts (we're timing encoded material from Twitch or Youtube after all and not actual game frames) it's not reasonable to give the times in milliseconds or even only down to centiseconds. I'm fine with keeping the resolution at full seconds, too.

That said, I'll leave all of this open for discussion again. So please discuss! :-)

Edit: This is mainly interesting for board moderation. Runners can handle their splits however they want. If you're running on BGB, Gambatte or GBP and want to incorporate this into your live timing then you just have to subtract 0.268s from your timer offset and half of that if you're running on SGB2. Or just don't change anything because a moderator will time your run when verifying your submission anyway. :-)

Over the next couple of days I'll try to update the leaderboards, i.e. remove runs on inaccurate emulators, correct times if necessary and add information about which particular platforms were used for the individual runs. You can help with the last part by posting here in this thread which platform/emulator you used for your current best run on the board. That way a moderator can just add the variable and you don't have to re-submit anything. If you have reliable information about a run that is not your own, I'd appreciate if you shared that information as well.

Cheers! ö/

Let's dig up a rather old topic and look at it in a different way: As documented by 2 threads in this forum, back in the day mugg was comparing Tycon's run to his own and came to the conclusion that Tycon's was recorded at a wrong framerate. I did some comparing on my own and came to the conclusion that it is mugg's run that is sped up. For the 3 segments that I compared so far, mugg's recording is running 0.8% too fast. I used the in-game timer as a marker, comparing the segments between the frame the timer switches to 399 and the timer switching to x for 1-1, 2-3 and 4-3, x being 361, 268 and 206 respectively. I compared the runs of callumbal, TheTycon, arkanias566, darbian and 3 of my own runs to that of mugg and arrived at the following results:

The empirical evidence clearly shows that mugg's run should be retimed as it doesn't seem to fit the speed of any other used platform (SGB2, GBP, BGB emulator). I believe this is a recording/encoding issue, rather than a platform issue so it should be safe to assume that the speed-up is consistent over the whole run. I will still compare more segments to make sure that this is the case. By dividing the correct consensus time for a segment by mugg's time for the same segment we get a coefficient. For the segments I timed so far, this coefficient would be ~1.008. With that we can recalculate mugg's time (12:45 x 1.008) and would arrive at a time of 12:51.7. I'd like to hear a few opinions on this matter. Cheers.