Konami Gradius series

[tomaitheous]

I've seen some people mention that the Gradius series has slowdown built into the engine to help facilitate in getting out of those crazy situations. This was specifically mentioned in passing for the SNES (Gradius III). I kinda dismissingly laughed it off. Just some excuses of rabid fanboys insecure about their beloved console's processor speed. Then, recently, I saw this mentioned again on a dev forum. I questioned what kind of research was done to prove this. I wasn't given any links, but was told that a certain reputable member of the dev community actually looked through the code and found evidence that appeared to point to the fact.

I didn't think too much of it, until more recently. Watching Spida1a's review of Salamander for PC-Engine hucard. It jokingly mentioned of the slow down in the game. That reminded me that the Gradius hucard port also had slow down. And so did Gradius 2 on PCE CD. I'm thinking at this point, Konami are some pretty incompetent programmers.

So I did some research about the arcade hardware. For Gradius and Salamander, the main processor is a 10mhz 68k. For Gradius 2 and 3, it's a dual 68k (10mhz IIRC). Plenty of processor resource. And all the games have slowdown.

So either Konami is great at game design but shit at actual code implementation, or this fact that the slow down is built into the engine/game as part of the design to help it out. I'm suspecting it's the latter.

[Haoie]

I purchased Gradius III/IV for PS2 recently, hasn't arrived yet.

Is there much slowdown in that?

[Zapf]

Its there by design. A lot of shmups are designed like that - they even properly emulate the slowdown in mame and modern console ports (not a framerate drop, the game itself slows down some).

Sometimes when you have a shitload of bullets you need a second to take a gander and figure out where you need to be.

[Bloodreign]

I purchased Gradius III/IV for PS2 recently, hasn't arrived yet.

Is there much slowdown in that?

The slowdown is present, but can be removed by the wait option on the pause screen, on default the slowdown is there, I can't remember if the higher the number the lesser the slowdown, or the other way around. Gradius III full speed is a bitch, there's a reason that game has slowdown.

[dreamcaster]

The slowdown is present, but can be removed by the wait option on the pause screen, on default the slowdown is there, I can't remember if the higher the number the lesser the slowdown, or the other way around. Gradius III full speed is a bitch, there's a reason that game has slowdown.

So that's what that feature is! I could never figure it out - the manual doesn't explain what it did.

[Ed Oscuro]

tomaitheous, I think that the boss explosions are points where slowdown serves a "cinematic" purpose, so it's probably there by design.

You're probably better equipped to settle this for once and all, so pull out the debugger. I don't see the idle speculations of the forum crew really doing anything but fogging things up further. That said...

First, this recalls this classic discussion on Shmups Forum; excuse our friend THE, programmer of Lost Hope (who knows a thing or two about implementing bugs) and especially excuse the less helpful postings of others in that thread as well. There are at least a few posts in that discussion worth checking out, however, that at least give some idea what the consensus amongst and impact on gamers is.

Secondly, considering that Konami made their own hardware board designs - I've seen credits of some arcade games where at least one person was tasked (credited) solely with the board design - and even custom chips (even early on) it doesn't seem outrageous to consider that this could have been intentional, or at least not considered worth scaling back the game for. If I had to put money down, I'd guess they'd have a game design in hand, then they'd start implementing it...and when there was slowdown they said "aw hell, but not worth cutting the game down for." Which I generally agree with, especially if it meant more time for developing new games that otherwise might not have been made, like Surprise Attack I suppose. (Which I don't think has slowdown anywhere, incidentally.)

Gradius III SNES might be a special circumstance, however, because like Super Castlevania 4, or Castlevania Adventure and Nemesis on GB, it's an early game for the system it's on, and far surpassed, objectively speaking, by later efforts (for the SNES and GB, I'm thinking of Axelay, Belmont's Revenge, and Gradius Interstellar Assault). Of those early games for their systems, only Castlevania 4 is felt a real classic by most. Gradius III isn't the only arcade-to-home shooter move in the first batch of SNES games that was slow - so was Super R-Type (which has objectively gotten the crap beaten outta it by R-Type 3).

A lot of shmups are designed like that

Well, Cave (I think in particular) are known to cause the action to be slowed down at points to help the player by lengthening the reaction time window - think forced bullet time.

they even properly emulate the slowdown in mame

This has nothing to do with shmup design, because "they" in this case are MAMEdev, whose goal is to accurately emulate arcade games, including the slowdown. MAME is not intended to "improve" upon the originals, except as far as making the internals a bit more accessible than before. If MAME isn't emulating slowdown native in the original game, either you need to replace your thermal paste or MAME hasn't gotten very accurate on that particular game yet.

and modern console ports (not a framerate drop, the game itself slows down some).

There has been a lot of drama over the years about arcade to console ports, but I've read Cave in particular have been adamant that their home system ports accurately reflect the arcade game experience with regards to timing, so it has the same feeling as the arcade game and players who know one system will not feel they are playing something totally different when they go to the other format. That's the theory anyway. In this case, they did have to go to extra lengths to get the timing down right. Looking at this part of your post again, I'm not exactly sure you are referring to this, but it's an extension of your theme. Incidentally, framerate drops are bad, almost as bad in classics as sprite flicker (disappearance) in games (i.e. in Daioh), and especially in dancing ("maniac") bullet shooters where not knowing where the bullet is going is as bad as not seeing it.

Making such "arcade-correct" ports is probably an extra-complicated task if the original game slowdown is related to saturating the CPU with bullet movement and collision checks, since the way the slowdown behaves may be determined by the quirks of that CPU. It wouldn't be impossible to portray this on a more powerful system, but it does seem to push the PS2 a bit as the arcade games being ported aren't '90s vintage.

Back to the Shmups Forum thread link, check out discussions of "intentional slowdown ramping" for lulz and profit.

My thoughts on slowdown...the typical "slow movement when boss explodes" is worthwhile (extra time to savor boss death, though I also enjoy quick & ungracious funny deaths like Koranot, from Super CV 4, again); bullet time type slo-mo can be worthwhile, but overall, I don't see anything wrong with going fullspeed the whole time and varying intensity the old fashioned way. That said, I would like it that any game's design vision be implemented as intended.

[tomaitheous]

Yeah, that's why I asked here. I figured there would be more information on this here.

I don't doubt that there is slowdown in the SNES version of Gradius 3 because of the slow rom used (results in slow cpu mode). But even hacked for fastrom mode, it doesn't eliminate all slowdown. And the slow down in that game is more than 50% (2 frames). Which is kind of absurd if you think about it. Under normal circumstances, the code missing a frame update because it's taking too long. Thus you get a immediate 50% slow, because it's waiting for the next 'end of frame' (modern games will drop the frame rate rather than drop the speed, but you can still get both). And the SNES Gradius gets into 3 frame slowdown at times IIRC (1/3rd of the original speed). That's ridiculous even for the SNES processor speed. My point being, most games only slow down to 50% - not less (50% reduction in speed is equivalent to a 200% boost in processor speed, hence why most games don't slow down past 50%).

Anyway, the idea of it makes sense.. I guess. Even disregarding the SNES versions/situations altogether. It was the PCE ports that had me wondering and now the arcade versions emulated under MAME. PCE processor is very capable and the arcade hardware much more so. But, I'm surprised that this wasn't common info, knowing us gamers If anyone has any official info on this subject, please post it.

You're probably better equipped to settle this for once and all, so pull out the debugger. I don't see the idle speculations of the forum crew really doing anything but fogging things up further. That said...

Meh, I've already have some projects I'm working on (don't need any more distractions at the moment). But I figured this would make a good discussion. Everyone loves shmups (The cool people do, that is)

[Ed Oscuro]

tomaitheous, to be honest I completely forgot about the various ROM speeds, but yes I also don't think it should be the major factor. Gradius III, and Super Argh Type, and some other games simply seem shoddily done. There also isn't much of an excuse to be made (wrong system I know) for Castlevania Adventure, though it's very playable; Nemesis / Gradius on GB is more competent but still a bit underwhelming.

I put it down to unfamiliarity with the specific hardware. That seems odd considering that the CPU is what, a bog standard Apple // CPU, more or less? Maybe they didn't have much exposure to the 65c816 but it seems quite odd all the same. Perhaps some early dev kits had poor code for interfacing with the new graphics abilities?

On the subject of info not being forthcoming, well, many developers in Japan are quite circumspect about even successful, well-respected franchises...and naturally more so when discussing abject failures. Even the topic of Cave's relatively successful experience with slowdown is the cause of much gnashing of teeth and drama manufacture, so I can't blame them for not saying a lot about how that came to be, other than the apparent part which is that they insist on putting it into ports.

[Ace]

I didn't think too much of it, until more recently. Watching Spida1a's review of Salamander for PC-Engine hucard. It jokingly mentioned of the slow down in the game. That reminded me that the Gradius hucard port also had slow down. And so did Gradius 2 on PCE CD. I'm thinking at this point, Konami are some pretty incompetent programmers.

Don't forget that the recent Gradius Rebirth for WiiWare suffers from slowdown as well, but that's intentional slowdown. The slowdown occurs on the early Loops in the game, but when you reach Loop 3(the hardest Loop in the main game), the slowdown is gone, which makes the game many times harder, not just because it doesn't slow down, but also because the enemies fire off A LOT of shots, and they fire off 2 shots when destroyed.

So I did some research about the arcade hardware. For Gradius and Salamander, the main processor is a 10mhz 68k. For Gradius 2 and 3, it's a dual 68k (10mhz IIRC). Plenty of processor resource. And all the games have slowdown.

Gradius III NEEDS slowdown, but even with slowdown, the game is an absolute nightmare. It's the one Gradius game that gives me so many problems, I can't even get past Big Core Mark III(Stage 3's boss). Play Gradius III on Gradius III and IV for the PlayStation 2 with slowdown removed, see how far you can get without slowdown.

There is one Gradius game, however, where the slowdown has to do with the CPU, and that's Gradius II for the Famicom. I had recorded a playthrough of the game, and on Stage 2, the slowdown was so severe that even the music slowed down. Then again, it's not your typical Famicom cartridge, as it contains Konami's custom VRC4 mapper chip, the reason why it never came to North America(and also why Life Force is downgraded compared to Salamander, which contains a Konami VRC3 mapper. No such thing on Life Force, which means one less Option on screen and a static Konami logo after you beat the game. In Salamander, you get different endings depending on how many times you continue). See for yourself between 5:08 and 5:16 - http://www.youtube.com/watch?v=NH5vXMsFD_k

[Zapf]

For some reason I dont remember there being much slowdown in gradius rebirth, other than near the end of stage 2 with the horde of pink globs coming up behind you. I'm going to have to play that again.

edit: my bad about the mame thing, maybe I was having a brainfart and thinking of them like other emulator programmers - noone minds an n64 game running at 60fps, you know what I mean?

[Arkhan]

the slowdown exists in basically every version ever, so I would think it is intentional. it cant be coincidental that they were inefficient on every CPU

[tomaitheous]

I can't believe you guys are complaining about the slowdown in those great games. What is next? Are you all going to switch to Airraid mode complaining how the Super Grafx version of GnG has grass that doesn't cover Arthur's feet??

Reading comprehension fail No, it's not about complaining at all. It's about the perception of slowdown normally being applied to unskilled programmers/team and/or underpowered hardware. Never once did I read that slowdown was there intentionally, BITD. Yet, this appears to be the case. I find that interesting. And Konami was apparently doing 'bullet time' way before the Wachowski brothers

Maybe they didn't have much exposure to the 65c816 but it seems quite odd all the same.

Konami did a lot of coding on the NES/famicom. The '816 is a direct extension of the 6502. The majority of optimization and technique of programming on the '816 processor, is just like that of the 6502. They would definitely feel at home with the SNES processor going into development. But some companies do/did put a higher priority on the higher level of logic, than the lower level stuff (because you can waste a lot of time constantly rewriting optimizations for small changes in the high level logic). Stuff like game design and such. Which is a good thing IMO, but softs often suffer on a technical level because of that. I think it's kind of rare to have a company or team that excels at both.

[Thrillo]

In SNES Gradius II, if you have your ship upgraded to the extreme max (4 options spinning around you, full shields, twin shots, wide-beam lasers) and fire in open space on the beginning of a level, the game slows down even if there's no enemies on screen. There's no reason for that to be intentional. I always took that as confirmation that the CPU was constantly being taxed quite a bit. That, and the fact that there's a relatively large amount of sprite flicker.
edit: though I guess it could be caused by the ROM speed...

The slowdown in Famicom Parodius is pretty strong, and the game feels extremely sluggish at times especially when there's not much activity on screen. I wonder if that was due to Konami putting more focus on making the game look pretty, or maybe they approached the port with different optimization techniques; who knows.
And that's some cool info about NES Life Force vs. Famicom Salamander. I always wondered why Konami took out those awesome endings; I never realized that they use different mappers.

[Orion Pimpdaddy]

I have played Gradius III on the SNES and I thought the slowdown was so bad that it ruined the game. When I played the other two games on the original NES, the slowdown was less frequent. I just find it hard to believe the programmers of Gradius III wanted the game to slow down that much.

[Arkhan]

I have played Gradius III on the SNES and I thought the slowdown was so bad that it ruined the game. When I played the other two games on the original NES, the slowdown was less frequent. I just find it hard to believe the programmers of Gradius III wanted the game to slow down that much.

*plays these games on MSX and doesn't experience slowdown*

*does experience SCC music, which rules*

[tomaitheous]

I'm not saying there isn't slow down normally on Gradius III because the slowrom mode they're using. That's obvious. But there's still supposedly (here say) still code in the game that will slow the game down in tight areas. Overclocking the SNES (fastrom hack or via emulation) gets rid of the majority of slowdown, except for the areas that are apparently built in. Weird.