Posts for Alyosha

Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
I don't know much about the RAM map of this game sorry. The jump I got was to 02A0 and then it continued on until it hit 03A2 which contained undefined instruction 02. Everything from 02A0 to 03A2 was FF, which isn't a very helpful jump, I don't know if others are possible. EDIT: so far I can only get a jump to 02A0, when the game resets it is BNE past the bad command (by pure luck) and eventually makes it to a BRK. Overall doesn't seem useful at the moment. You can avoid a crash / reset by jumping through the ceiling before going left past the point of no return, but this doesn't seem to do anything useful.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Wow really interesting. Thanks for pointing that out archanfel I was not aware of it. A cursory look at a trace from a crash case shows stack corruption from a TXS instruction and a jump to RAM. Shouldn't be too hard to track down what's gone wrong and see if it's useful or not . I might not get around to it for a while though so good luck to anyone who wants to jump in, I hope it proves useful.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
wow, thanks for the show of support Zeuper : ) much appreciated! I pretty much agree with Warp on this one, comparing now to 10 or even 5 years ago isn't particularly meaningful, especially with the emergence of Bizhawk and TASEditor. About the only way someone could beat the submission / publication record nowadays is if they specifically set out to so and planned maybe 2 years in advance, at which point it would be a pretty artificial record. But anyway, onward to 3000!
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Thanks for doing this research jlun2. Looking those runs over its not really surprising why they got rejected before the vault was available, but at least it gives me something to look at if I get stuck or lose motivation on the runs from the original list.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
I'm with andypanther this is TAS of the year stuff! The set up in that video is really smooth too, the cpu's are right there when you need them, great work!
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
I'm always really impressed by the total control stuff especially when its done on real hardware, really amazing! The text in that encode gets garbled for me though after a little while, so maybe a higher quality encode is needed.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Since Arc did all the inputs for this run, I request that his name be listed first in the publication entry. EDIT: thanks!
Post subject: A-2600 Dragster
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
I noticed some graphical glitches in Atari 2600 Dragster, and am thinking there might be an emulation bug First, the car starts in the wrong spot on the screen. Here is bizhawk: And here is another A2600 emulator PCAE 2.7: watching some youtube videos all also show the correct car position at start. Also, after a certain point on the screen the car breaks up into sections for some reason: This also doesn't happen in youtube videos or in the other emulator. Any help on this would be appreciated, if there is a setting I should be changing I surely can't find it. The pics were taken from bizhawk 1.11.4 as well. Finally, I've really noticed big improvements in Bizhawk since I first tried it about a year ago, keep up the great work development team!
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
So in case this isn't already known, the race ends when $3A reaches #$61. 3A increments every time $42 passes FF. 42 is just adding your speed to itself each frame. So basically, the race ends when you have gone 24832 units along the track. The condition to check if the race is finished actually happens the next frame though, and time increments first, so your final time is actually the next time unit after you cross the finish line. So assuming switching gears never pauses increase in speed, and assuming speed is capped at 252, the theoretical best time you could get in this race is 5.47. Now since switching gears pauses your speed for a frame, this and you need to get at least to gear 4, and you have to switch gears by a certain speed, we can estimate the theoretical best practical time as 5.54. In scepheo's movie, this is possible if speed is never paused after 4th gear is reached. This is still very far away from 5.51, as you are actually already 252 distance units down from ideal time by the time you reach 4th gear, and 672 units away from ideal when your speed caps out. So as far as I can tell, the only way to get 5.51 is to somehow start earlier. EDIT: also I notice some graphical errors that aren't present in any gameplay videos I see, my car tends to split into 3 pieces even in normal play, and at the start part of the tire appears on the right side of the screen. Some sort of emulation bug? EDIT2: its also possible to have a speed of FD, but in order to get it I had to slow down 1 speed unit earlier on which is definitely not worth it in the long run. EDIT 3: so I looked a little more and something emulator related is definitely up here. I tried running the game in another emulator (PCAE 2.7) and the game displays correctly and my car does not stretch apart into multiple pieces. Unless there is some setting I need to change, I think there is an emulator bug.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Oh this is a neat problem, who could have guessed an atari 2600 game could hold this kind of a riddle. Well I briefly looked into the code just to see how the time is kept track of. Here is the bit of code:
--If the racer hasn't finished the race
LDA $D2,X 
BEQ $F344

--If the countdown has reached 0
LDA $8D
BNE $F367

--add time
SED 
CLC

LDA $B7,X -- this is a counter that increments by 34 every other frame
ADC #$34 
STA $B7,X

LDA $B5,X -- notice that the carry flag can be set, so sometimes it adds 4
ADC #$03 -- add time for the current racer
STA $B5,X
One thing I noticed is that if you start from power on, your opponents time is incremented first, but after resetting, my time incremented first as the race started. I haven't looked too deeply into why yet, but I do wonder if this can influence the time you get.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
When I first saw the title of this game on the workbench I thought it was a rom hack, somehow I have never heard of this game before. It's a really nice looking game though and the TAS seems well done, too bad about that super jump discovery a little late but oh well no big deal. Thanks for introducing me to a nice game I had never knew existed!
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
I noticed that MESHUGGAH isn't added in the publication entry as co-author, and also that neither of us have the game added to our list of published runs, some kind of glitch in the publication process? Seems to be a stray entry in the player's page list too associated with this movie: http://tasvideos.org/Players-List.html EDIT: oh, that was fast, thanks for fixing that Spikestuff
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
I'm curious on how this is coming along, any updates? Find any more glitch warps like the one in level 4?
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Well its been a while since I could cross a movie of this list but now Snake Rattle n Roll is finally done! I'm pretty seriously stuck on Double Dragon, so I might put that one on hold for now. Probably Strider and Out to Lunch will be my next picks from here. Uh oh, but since Samsara edited it that one time it seems I can't edit the opening post anymore, can this be fixed please? EDIT: VVVV Thank you!
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Tompa wrote:
I need to wait another week at least before I'm able to watch your WIPs as well as helping out myself. As that's when my new computer will be up and ready. Just so you know =).
Cool, good luck with the new pc. I won't be doing much until after the holidays at this point, but in the mean time here is my current working file that completes the game with 101%. It has my most recent croctopus chase and I included some clever jump rolls that were in the nico video TAS as well. Now about 12 seconds faster then my last file, and 18 seconds to go to reach sub 43, not sure we'll get there though. http://tasvideos.org/userfiles/info/27617079760156308
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
So I thought I might post to say that I will no longer be actively looking into the stage end glitch. Everything I check comes up clear, so I am categorizing this as hardware error unless some new information comes up. The code suffers from a weakness in that flipping ANY bit in $003B will end the level immediately, with no sanity checking of any kind. So while still a rare occurrence, this is very susceptible to a random bit flip. There are few instances in the code that change 003B and none of them show up abnormally in anything I test. At the time the glitch occurs, nothing in particular is happening, and I have looked at lots of trace logs to try and find anything out of the ordinary. Level end is delayed by a timer at 002A. At most this can be FE and countdown, setting glitch initiation back a few seconds, but here to nothing of interest is happening. While I suppose something like DelayStageClear from megaman is still possible, I don't see anyway a bad bank swap would lead to the level ending. Even when I modified the ROM to purposely call the code I wanted, I still got results that were distinguishable from what happened in the video, because that block of code does other things like clear enemies and trigger sounds. I don't really see ACE potential either, but obviously I didn't check everything so who knows. So if anyone is still interested in this, I will gladly help answer any specific questions, but will not be investigating further. To anyone still hunting, good luck!
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Cool you are about half way now, keep it up!
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Hurray its done! I'm not sure I deserve to be co-author for this, all i did was work out a few tricks and the technical stuff about V9, but I'm really happy to have helped in the creation of this run. Seeing that platform work out just right sure was nice though! I'm also happy this run is such a big improvement over the (technically invalid) V9. I do like that this site tries its best to be close to the original consoles and strives for verifiablility, so having a more technically valid run of such a classic game is pretty important I say. Yes Vote!
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
^ well my understanding was that warpless was deemed more interesting. But a lot of those runs were done years ago with bad roms and emulators, so maybe now there is room for it? Personally I'm not all that interested in warps, but if someone wants to modify this TAS into a warps one maybe I could help along the way. It's basically just level 1,8,9,10 I also have a temp processing but it might not be ready for a while so thanks ahead of time for that Spikestuff. VVV oops, I always forget about that bug, when I make submissions. Can an editor please change it to NES? EDIT: thank you!
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
This is a very impressive TAS! Glad you had the will to restart after the desync and new tricks being found.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Nice! Will be waiting to see this on the workbench.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
http://tasvideos.org/userfiles/info/27511624942686207 finally made it to the end! Currently 2048 frames ahead. Level 10 needs some work still and I still need to work on the last boss a bit, but this is pretty close to final. If anyone has any ideas for entertaining stuff to see at the end of the game there are about 7 seconds of down time. I could end input early by just standing next to the spaceship, but that would be too boring even for me, so I'll stick with convention as ending input on the last frame input is accepted before boarding the spaceship.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
http://tasvideos.org/userfiles/info/27491139845120584 Got some good manipulation in level 8. Not sure how much better is theoretically possible, but I think its pretty close, and is almost 11 seconds ahead of Nitsuja in that level. I am now ~1550 frames ahead overall, aiming for sub 7. But I noticed at the start of level 8 Nitsuja's run desyncs in both FCEUX 2.2.2 and FCEU 98.15. I have a different ROM then the published run, so I'll need to make sure I don't have a bad ROM before continuing to level 9. Should be done fairly soon though. Also I am back ahead of the one plaer run by quite a bit now too after falling behind waiting for the scuba powerup. Now 474 frames ahead at the start of level 9. EDIT: well if I have any idea what I am doing I seem to have a proper rom, and according to stiltzkin on the previous page the ROM used in the published run is a bad dump. Aglar's run also uses this ROM but it syncs properly for whatever reason.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
^ Not sure, I didn't look through the code to determine how it works. Getting it working was enough for me. http://tasvideos.org/userfiles/info/27467007092971882 Ok, level 6 couldn't be improved much because both important factors are timing based, namely the enemy to bounce on and waiting for the scuba power up. The scuba power up gets shot out every 8th shot, so there really is nothing to do to speed it up, which is unfortunate. Level 7 is now ~150 frames ahead following Aglar's route up the geysers (which was very troublesome to reproduce with both snakes.) Anyway moving on to level 8 and 905 frames ahead. Levels 8-10 are about half the movie so I'd say there a chance of finishing under 7 minutes. I also tried investigating PJ's floating above the spikes trick, but so far nothing. Even if you get off the spikes, I suspect the game doesn't think you have 'landed' yet, so you would still die from fall damage, but I can't yet get past the spikes to see.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
http://tasvideos.org/userfiles/info/27438050487102963 I got a much better manipulation in level 3 and am now ~90 frames ahead of Nitsuja's attempt. I also added in the missed shortcut in level 4 which saved a few seconds. But now I am stuck in level 5 at the jump off the flying carpet. In the current run Nitsuja waits an entire cycle to jump up to complete the level, but in Aglar's one player run he is able to jump right up. In my attempts I am always 1 height unit too low to clip up onto the ledge, and am not sure how to get another unit of height. Since this will be a pretty big time saver I will try for a while to see if I can get it before moving on. EDIT: The current run is 352 frames ahead of Nitsuja's run at the start of level 5. EDIT: Got the jump. Not sure entirely how but jumping or not jumping at previous points in the level seems to impact it. 718 frames ahead at the start of level 6.