turboxray

FPGA or rather, Mister: for main systems like NES, MD ,PCE and a few others.. the emulation is constantly iterating and getting more accurate, while "software" based emulation tends to be stagnating. This might sound good, and it IS good, but that doesn't mean "software" emulation is somehow invalid. Direct to CRT support Support for original system controllers (this is important for PCE as turbo settings doesn't behave the way software emulations implement. Can easily switch between 6button mode and 2 button mode... TAP works with it). If FPGA evolves into some other more advance chip, the "vhdl" cores of the emulators are mostly portable. So that's not an issue Play it without "PC" issues. I mean, the R-PI can do this too, but you're always suspectable to something with the OS interrupting the core/app on PC. I absolutely hate any dropped frames. Emulators are concerned with supporting a wide range of people and that includes the least capable PC systems, so inherently the last part of reaching accuracy is going to cost a LOT on a software solution; with FPGA.. every thing can run separate and in parallel. That means a 200mhz FPGA can easily beat your new fancy CPU in your PC when it comes to emulation on a circuit level. Again, you probably won't notice.. but if you're the type of person that frets over this (and you probably own real hardware too), then this fulfils that need. I personally like the fact that it's usually more accurate, is continuing in development, isn't coupled to my PC, etc. For me, it's a win-win. I only use software emulators for debugging purposes (inspecting games, or looking at my own code for homebrew) - or capturing some data. Don't get me wrong though, if you're not into following every active emulation accuracy update.. and think a decade old software emulator is "good" enough.. then that's great! Don't get bogged down with Mister if it doesn't offer anything for you hahah. I personally think that it having an external bus to actual carts would give it some bonus; ala Analogue's products. But this probably will never happen. I also pair the Mister with the Tink 4k since it now supports HMDI, for all those lovely custom filters. I think that's probably overkill and out of range for what most people want to spend on these products. I dunno. It's just some other thing to play with related to retro gaming and ultimately that's what most people probably get out of it - it's a fun toy. s That all said, I play my real systems with a flashcart more often than Mister.

"Intro into the thick of it" at ~6:30 in the video.

Okay, if you won said millions from said lottery... and you put together a top tier AAA team to do the sequel.. how would you design it? What story line? Etc. And this would be a retro-dev release for the SNES too (cause why not hahah). Ohh I had forgot about this!

Wow... I stand corrected: I played every single one of those games and never used it. I don't even remember them in game, with the exception of CotM which that attack was nice because it's not implemented like the rest and don't consider it the same (didn't have that weird limp whip physics). I still maintain that's mostly useless. I never used it in any of those games, and in SCIV I rarely use it (I also dislike the timing of it too.. you can accidentally trigger Brandish whip mode which is annoying). They added that to Ritcher in SotN but none of the directional attacks (which is my primary post between the two).. also missing is the jump direction attack. But none of that changes what I said about SCIV design vs Rondo's design; Gameplay, mechanics, and design related to stage and enemy AI/pattern and how that impacts the style of Rondo design/feel is the direction the series took in later games. That's an evolutionary step over SCIV by comparison.

I wouldn't. I've seen errors/typos/etc in official manuals.. it's not surprising. These old official manuals are nice curiosities, but outside of that I never use them; there isn't really a need to use them as we have plenty of documentation and understanding that goes beyond anything these manuals can provide. It's possible the person who wrote that, or taken from another source, thought "AD" in ADPCM means "ADaptive". But if you've ever written an ADPCM encoder and decoder, you'd know it's nothing like BRR. The "D" in ADPCM is delta/differential (the D in DPCM is delta). The adaptive part is that the 4bit "delta" is not a direct delta value like it is in DPCM; it's a delta for an accumulated "index" and the current state/value of the "index" points into a table that has the real deltas - in which you can imagine that table is a non-linear ramp-up of delta values. There is no block or filtering or such. Some early/cheaper ADPCM playback chips doesn't even saturate.. meaning it can clip and wrap around (which sucks haha.. a cheap/easier/lame way for encoding was to use a lower bit-depth than the target device final output, so that it never clipped but you then had more bit-crush artifacts). The only thing I can see that they have in common is the data points are 4bits in length and so have similar compression ratios because it's 4bits per "PCM point" (ignoring the fact that you can have 3bit and 2bit delta widths in ADPCM for those specific variants. 4bit is the most common though). For the longest time, because people just parrot this and on sites, that BRR was just a variant of adaptive delta pcm.. I was pretty surprised just how different it was when I dug into the format of it.

Calling it ADPCM is a misnomer. It's not really "adaptive delta/differential" that most ADPCM schemes follow. The BRR data points aren't deltas, like with ADPCM codecs/schemes. It's more of an adaptive "range" PCM.

The game came out in 1993. That's like 3-4 years after it came out. I dunno man, there's just soo much to unpack between the two games. Too much to go into detail here, while I do love SCIV (especially for nostalgia reasons), it has a lot flaws.. and is experimental in its game mechanics/controls. It's not a forward evolution to the model.. it's a side-step experiment. Me thinks you really haven't given this a truly objective look, and it's just nostalgia or bias getting in the way. To sum it up; SCIV limp rotation whip model is ridiculous and unnecessary - it was there to help differentiate it from the original 8bit games (and not really good design or even though out). The character is too big, and most importantly the whip is too long while at the same time being too over-powered. The whip range being too long means everything is kept at a safe distance - this directly affects the pacing and energy of the game.. paired with its simplistic enemy AI/patterns. And having the mechanic of diagonal and straight up attack just really adds to this, makes the combat more boring. By comparison Rondo has tight responsive controls because you need to be reactive against the enemy AI patterns - it puts you right into the action and is constantly keeping you on your toes (SVCI does not do this, and neither do the 8bit games); if you're not using the backflip often, you're playing the game wrong. The sub-weapons (which is a staple of the series) are mostly an after-thought in SCIV which is also a tell-tale sign that had trouble balancing the whole gameplay mechanic change from the 8bit ones. This isn't from memory - I just played through Rondo from start to finish this past Monday, and then SCIV. Everything that makes Rondo fast, intense, responsive, complex, and high-energy/hectic with balance and perfect controls.. is what is missing from Dracula X SNES port. Just to note: SotN didn't add "limp whipping" and diagonal/straight-up whip attack abilities to Ritcher from SCIV.. for good reason. SCIV not an evolution, because it stopped with that game. Konami knew what they were doing when they tweaked Rondo to a near perfect balance and experience. Tighter controls, smaller character, non-overpowered which allowed items to be part of the gameplay, and enemies that could actually do something other than walking back and forth (which is a lot of SCIV design) or rotate around blocks. SCIV is very dated in gameplay design and stage design for that gameplay, in comparison to Rondo. Having to get whip extensions via candles is 8bit. SCVI shares more with its design (especially enemy patterns) with its 8bit roots than Rondo. I mean I love SCIV. I love the atmosphere and the sound track (even though it's different). The artwork for the backgrounds is still impressive today when you consider it's an early title (and cart size). And despite SCIV simpler gameplay mechanics and enemy patterns, I still come back to it simply because it's different and my nostalgia for it. But I'm not going to pretend that it's somehow an evolution over a Rondo, or that Rondo is a step back. I would LOVE a remake of Bloodlines (on the MD too). An open-source CV engine would be great for that (and definitely new art assets).

The GBA ones are alright, but the DS ones are just better all around. I played them all in order, and OoE was my favorite of that genre for a while - was a fan of the sound track too. In between when the DS ones were coming out, I re-visited the last two GBA ones. I had less of an attachment on the second time playing through them again. I don't remember any of the GBA CV game soundtracks as particularly memorable or stand out to me - other than Harmony is poor and grates on the ears in some spots.

Rondo feels like a step back? What else is there to compare it to when it came out in '93? SCIV? Rondo stage design was top tier when it came out. The thought and tweaking that went into the stage design/layout/etc was the best at the time for any CV game. The enemies are incredibly varied with AI patterns, and capabilities.. again, above anything out for CV at the time. Matter of fact, they're re-used in SotN. People like to talk about "sectional" sprites for bosses.. but Rondo has sectional sprites for the enemies.. resulting animation that is 10x over of anything that came before it. Not to mention the particle and palette effects. Even Dracula X on SNES, re-imagined, doesn't even come close to Rondo's production assets and animations. The bosses are also so much above and beyond anything CV out at the time; animations, patterns, etc. The animation of detail that went into the boss designs, artwork, patterns is pretty crazy. Ritcher has much tighter/faster response and controls compared to anything out at the time for CV. The backflip adds a whole new level of strategy to the gameplay. Items and item crashes, etc. "Items" are sooo much of an integral part of the gameplay mechanics and especially for Ritcher. So much more thought went into this game than simply adding a "diagonal attack". This game is better for it. Hell, even the stair case ability to jump on/off is part of the gameplay mechanics and is built into stage design and flow (it's not just a party trick). And Maria being a completely different style, gameplay, strategy. People like to say she's just an "easy" mode.. but she's much more than that.. and while her attacks are more powerful (because you can attack at twice the rate), she takes much more damage than Ritcher. There's just soo much here in the gameplay mechanics. I never appreciated it back in the day when it came out (PCE CD) just how much they put into this game - just how polished the entire thing is in stage design, AI patterns, and gameplay mechanics - all balanced for an incredibly polished experience. Once you've learned all the controls, the game difficult is balanced fairly well too. It's not CV NES hard or cheap. I actually thought Rondo was a little more on the easier side when I played it back in xmas '93. I would say there's nothing backwards about Rondo. Right up until its release (and even SotN release), Rondo pushed everything forward. We are talking about Bloodlines, right? The colors are extremely contrast-y to the point where it looks like a simple color expansion/upgrade to an NES game. It's kinda bad. I was like WTF when picked up the game when it first came out. The artwork is low tier for stage design (lacking nice details and only has like 2 shades per color, etc). The stages lack detail for the most part. I remember my friend gushing over the statue head the breaks and falls.. but I was like yeah okay, but the rest of the entire stage is just some "lines" for tiles you walk on.. no detail. When there is detail, a lot of times it has that NES section design to it.. where colors are clearly isolated into 16x16 cells.. and are just shades of a single hue, or 3-4 color contrast-y look. The enemy design/look is also closer to the amateur side in design and pixel art-work, but animation is also NES level - a lot of 2-3 frame awkward animations (a lot of the walk animations in the game are just goofy looking). I don't think the bosses look that great either - the "pillar" boss looks soo dumb/goofy/amateur. The bat monster at the top of the pillar has something wrong with its tail physicals - look wrong/unpolished. The last boss is such a let-down. This was a '94 release!! The Gen/MD was capable of sooo much more than this subpar production from Konami. It's okay to say, "I like it anyway. It's different - warts and all" or whatever. But let's not pretend this isn't a subpar effort compared to CV games that already set the bar or just make up stuff because you like the game. I feel like people that gush over the graphics/design/animation of this game have low expectations of the Genesis or something. The sound track is definitely nice (and a few songs are top tier).

Unpopular opinion: This game, particularly the first game, is weird to me for a few reasons. I play and beat this game when it first hit the shelves. My brother had bought a copy that I quickly borrowed and finished. I remember thinking the game was pretty short at the time (not sure why, replaying it later on it felt normal length). I remember thinking the game was only "okay". By the time the first game released, I had other more capable systems. I also had a "gaming" PC (I built it myself, all the latest fastest parts/cards/ram/processor.. none of that slow ISA card junk) - expectations were higher. There wasn't anything next gen here from my experience back in the day. I thought the enemies and bosses looked weird/goofy/cringe-y in design. The graphics/colors didn't have that high color look of Jag, 3DO, or PC VGA.. they looked very contrast-y to me. I.e. lower color patches with blown-out details - this was the first thing that stood out to me (mostly the sprite models).. even from the opening scene. Maybe it's because I was already doing/editing/creating graphics and such back then. That, paired with the weird/uninspiring enemy and boss art. Gameplay mechanics and stage design felt somewhat flat as well - especially for a "Nintendo" game. A lot of enemies that just simply walked back and forth, and are basically stage "fillers", or just have plain generic patterns seen in lesser tier games from the beginning of the console's life (or even 8bit). It just felt like the entire focus on was on graphics. I mean, swap all the graphics for the original SMW graphics.. and then compare it to SMW.. DKC just doesn't do it for me and was a step back to even SMW in terms of design and mechanics - which was a release game mind you. I didn't hate the game, but for me it was mostly a "meh" affair; it did start out fun/new/fresh, but then just quickly became less fun as I progressed into the game (more on the repetitive and boring side; I hated the barrel launching stuff). I just saw it as a continuation of the trend in the 16bit generation where near the end I didn't care for a lot of the games coming out. So yeah, I got some enjoyment out of it.. but never understood the hype (back then and now). I never bothered playing any of the sequels when they came out. I played them years later in emulation. DKC3 is a horrible game IMO. I've beaten it, but that a real low point in the series. Baby Kong? Really? smh

I mean you can do more than a static screen. But yeah, it puts a huge limitation on vram when you have unique detail. You have to be really careful.. and weigh the trade-offs. Here's a use case example of a boss: (some random AI generated or whatever image I found scraping the net) Left: is the 8bpp version using 120 colors. It would allow two 16 color palettes for a second 4bpp BG layer. Right: is the 4bpp version using 79 colors with 6 palettes. It would allow two 16 palettes for another 4bpp BG layer, and another 2bpp layer.. could be used for transparency effects. So the 8bpp boss head floating around in game. 8bpp version takes up roughly ~35k of vram. Its size is 152x272 pixels (storage is less because of the purple area). So this scenario is doable and does leave enough room in vram for some sections of the head to be animated (mouth, eyes, etc). And it leaves enough room in vram for a 4bpp layer (mode 3), and sprites. Sprites have all their 8 palettes. So the 8bpp allows for more gradient to be kept in the original image than the 4bpp version.. simply because of how dividing into palettes like this can cause a lot of waste. You might be tempted to say that the 4bpp version doesn't look as good because it's 79 colors vs 120 colors. But here's an 8bpp version using 79 colors for comparison: Left: 8bpp 79 colors Right: 4bpp 79 colors, 6 palettes. So 8bpp still gives you better color density and fidelity (especially in gradients). Is it worth the trade off here vs the 4bpp version where you still have a 2bpp layer for transparency/color-math? The 4bpp version would also allow for twice as much animation for the same space too (rom storage).

"Palette cycling as opposed to".. direct color.. which can't. Glad to see you've finally come around. If you consider 1/4 the res a minor concession:

No you can't. Not without it being 4x lower resolution. You only have access 256 tiles in mode 7. Or did you not know?

Everyone that has Mister, Analogue Duo, or SHD3. And if Krikzz releases the expansion bus version of the new TeD Pro, then those people as well.

8bpp tile modes. It allows you to have more colors in a single tile. If a tile is 8x8, then it has 64 pixels total. Normal 4bpp mode means you can only have 15 unique colors for all those 64 pixels. You use palettes to choose which 15 colors to use. The tilemap assigns this "palette" number. Because art tends to be continuous outside of just 8x8 tile boundaries, you waste some "palettes" for colors that need to transition between two palettes. There are only 8 palettes for ALL the BG layers to share. That's not very much. So 8bpp allows you to more easily and freely use colors were ever.. without these segmented limitations. Of course, 8bpp tiles take up TWICE the space as 4bpp. Because of this, and the limited 64k of vram on the snes (which has to be shared with other things like tilemap, sprites, sprite tables, etc) - it doesn't make much sense to use 8bpp unless you're trying to show really highly detailed images (i.e. "digitized" images/art to borrow a phrase from the 90's). And in that, even though you still have "tiles", they're most all unique creating a pseudo "bitmap" type image. Again, this takes up a LOT of vram to do this. So much so, that it doesn't leave much room for sprites - so you have to be extremely careful. Because 8bpp is twice the size, it's also twice the bandwidth for the SNES sPPU's to fetch.. which means you lose some layers. Advantages; create some really highly detailed images with color fidelity and density that you couldn't with 4bpp traditional modes, but at the cost of putting huge constraints on vram (and rom too) - and not being able to do 3 layers. If you're doing large unique detailed images, you'll most likely have to revert to using a 2bpp second BG layer (which is quite limiting).. and gives a weird/jarring juxtapose between the two levels of details (8bpp BG + 2bpp BG). There isn't enough vram to use 8bpp BG + 4bpp BG in this way, which would have been ideal. There's also the issue of 8bpp mode (the more useful one that is "indexed color") will use the entire palette space if you try to use all 256 colors.. which means it will use up sprite palettes too. So having sprites means either dialing the color back from 256 to something less to make room for sprites, or having colors for sprites to be "shared" with the 8bpp palette. Same for 2bpp or 4bpp layer situation. But there is "direct color" for 8bpp mode, but it's mostly useless. Don't let Kirk tell you otherwise haha. And it's not "2048" colors either for direct color mode (he has no idea what he's talking about).

I got a chance to play the demo. It's a lovely game. Feels polished. Game design is pretty good. Hidden stuff, gain new abilities, map system for back tracking. And of course, I took a look at the game on a technical level too. So it looks like the game developers took into account the weird SNES cell reverse-priority dropout mechanism. They put the explosions and such above/in-front of the player and enemies, so that they drop out first rather than the player and important details. Normally this isn't a good idea because it basically covers up the things you need to look at, but since the game is pretty much guaranteed to drop these cells when it gets heavy with sprites - it works out. I.e. the game basically relies on the sprite drop out happening in heavy sprite use areas. It's an interesting approach. I did see some slowdown.. one of the bosses (I won't spoil it, but the one with the "thrust" coming down at you). Strange, because the rest of the game seems to keep up.. with seemingly more visual stuff happening on screen in other areas than this boss. Sprite setup is 32x32 and 8x8. A bit weird, but most of the sprite work is designed around single 32x32 sizes - this include bullets/explosions from the main character as well as enemies. Saves on some processing time too vs doing 16x16|8x8 setup. The sound FX are nice (with some bass to the explosions). The music is fitting, but nothing that stands out, or blows you away / would listen to outside the game. Art style is pretty good. It makes good use of the 15bit palette, but without all the 'pillow' shading and such. Minimal color usage paired with good use of the 15bit palette.. a modern retro look. I like it! Has some hdma "copper" effects thrown in, too (that's an Amiga reference that made that effect super popular).

So 10 people could play it? 😅

Now that is something I'd be interested in. As long as it had high production values, I think these stages would definitely benefit from modern realtime 3D/Raytracing.

Don't ever do that haha. That is such a waste of a channel.

It was Nicole Express' game made in C for PCE homebrew did it as more of a technical checkbox and doesn't really do anything with it. But technically, yes. But yeah, anything the PCE has.. the SGX can use as well. Yeah. But it has to be the back expansion bus. SHD3 Pro already did this for PCE; they added SGX support via the expansion bus - though in a slight hacky way. But yeah.. everything you need to create an SGX is on the crazy pinout of the expansion bus. You can even give the PCE 15bit (even 16bit) color palette! It can detect layers and which palettes are being used (on the digital pixel bus on the back plane) and could easily add SNES style color math.

Rondo is a nice looking game and does do some nice effects for PCE, but it's not pushing the hardware. And definitely isn't pushing the PCE in the color department. It's not pushing the PCE in the sprite department either (plenty of room in the Sprite Table because PCE can do bigger sprites than SNES). Matter of fact, a LOT of the animations are sectional sprites, palette effects, and particles. A combination for all of it. That's something that doesn't take up as much space as dedicated 'bitmap' animation (because the SuperCD 3.0 CD RAM couldn't hold much). It might be a tell-tale sign why Konami dialed that back on the SNES port of Dracula X, even though it has the rom space for it. Rondo is weird in that it has stages are mismatch in art quality; a few areas are really nice, while others are plain and very tile-y looking. Almost as if different areas were developed earlier in the game's development time. Or they had a mixture of different artists with different skills, but it's a mismatch of you start looking at the game. This is an area the PCE could easily improve over its own; no snes needed. If the SNES was pushing sprites like the PCE does, it wouldn't look good when dropout happens. Ritcher could disappear instead enemies or bosses, etc. The SNES can't do the boat scene as-is because of this (MD could tho). Generally, the congested sprite usage would need to be cut back a little bit and/or more thought put into the sprite routines to keep up on the SNES (for the areas that need it, general areas wouldn't be too much of an issue). I.e. you wouldn't want to put it as-is, and but at least you'd get some background enhancements. But really, a better port would be the PC-Engine SuperGrafx CD with Arcade Card.. not the SNES. Way more vram, way more sprites per line, way more total sprites, way more detail for tiles, and has additional backgrounds. You could even make areas with 3 FULL backgrounds with FULL 4bpp color. The original Rondo doesn't even come close to pushing the PCE's color capabilities (both palettes and colors), so the SGX version would easily have an opportunity for an upgrade there. SGX ACD would be a better platform if we're talking about more capabilities for an upgrade. SNES might have 15bit color, but the original game didn't even come close to pushing the PCE's 9bit VCE palette to begin with. There's no difference between a PC-Engine with a CD attachment, or the all-in-one PCE CD system. They are the same.. not graphical upgrades or anything. Same old tech '87 tech (core and CD), packaged into a single product - no CD speed increased, not audio or video upgrades, etc. Weird how you claim to be a snes graphics expert but can't wrapped your head around simpler concepts the PCE pulls off. Why don't you try running it through Mesen and figure it out for yourself? All PC-Engine games are PC-Engine games. The CD adds a different storage medium, but the interface is primitive - there's not caching or DMA, or anything. There are no enhancements, except a single ADPCM channel. The CPU does all the manual heavy lifting for the CD interface (which is why they added a second process in the Mega CD). The original PCE CD system is STARVED for ram, and only allows simplistic games. The SuperCD 3.0 card added some ram, but games advanced by then and and by the time the SuperCD ram expansion was released.. it was still starved for ram relatively speaking to SNES and MD carts.. it only adds 192k more ram. CD RAM "emulates" a cartridge.. so at any point in time, it's like having a tiny cart (and lots of load times). The Arcade Card fixed the ram issue and is in-line for its release date (comparing it to carts from other systems), but only a few titles use it. And of the titles that do show it off, they still don't use all of the available ram. Some games, like Mad Stalker, use it just to load the entire game into ram at the beginning, which is not how you optimize for that kind of ram (or for a CD game in general). Mad Stalker is basically the MSU or MD+ equivalent back in the day. Again it amazes me that you pretend to be a "graphics expert" and are just now finding out these simple effects??? FYSA: I know that SNES smw hack video with mode 7 and sprites for the foreground layer you posted is supposed to be in response to my PCE Bonk demo with its "obect layer", but the SNES is more limited in this area. The weird reverse cell drop-out priority order means the SNES is the least capable in this area for this type of effect. The PCE can easily let things dropout behind sprite sections/areas and you wouldn't even seen it. The Bonk demo is a simple demo, meant to show of a library for the PCE.. it's not the full effect of it. The full effect is something the SNES couldn't do without using the SA-1, doing sprite analysis to manually drop out sprites, and still going to need some SA-1 power to do real time sprite compositing to combat the effect. The "stock" PCE doesn't need any of that and neither does a "stock" Genesis (it can also do this trick because it follows the same rules as PCE). The "stock" SuperGrafx could put off FULL 4 BG layers with FULL 4bpp support using its 32 palettes. Maybe I should make an extended Bonk demo for the "stock" SGX to show it off. If you specifically love that, I've got something you're going to love on the PCE hahah. 2024 is going to be a nice year for PCE demos.

I'm beginning to hate this forum. This isn't even an argument, debate or discussion. There is no ambiguity.

It has literally been in popular lexicon for like decades. What composers are you working with???? Sample: vernacular; a digitization of audio into a digital storage format. Typically a wave file or uncompressed format, but nowadays any container or compression to hold the "digital audio". "Sample that audio", "use this short sample in your mix", "play a sample", "use a sampled sound FX", "I have a collection of drum samples", "I ripped samples from the CD", "sample pack for sample-based synthesis", etc. Anyone from the music and audio world should know this terminology.

That's true if the point represents some sort of "delta" and not direct value. That's not true for direct PCM (direct level) samples.. i.e. typical "wav" files. As in, if any point in the sample describes the direct level of amplitude (from base 0).. then it doesn't describe pressure. You would need at minimum two PCM points to describe pressure at that point in time.

Why would it be a "point in the waveform". If you're talking about a single point in a waveform.. that's just an offset. You're making this more confusing than it needs to be. The common understanding everywhere, outside of people that have gaps in their knowledge and like to re-define terms, is that a "sample" is a collection of PCM points (or DPCM, or ADPCM, etc... all irrelevant). And the attributes for a "sample" or waveform are size and playback rate rate (if it is a fixed playback rate). That's all you need: sample, size, and playback rate. These are universally used terms.. as Trekkies has pointed out. If you're talking about "sample-based synthesis" ala SNES, which is not the same thing as waveform synthesis, then the "playback rate" of the sample describes a reference for "resolution" for its frequency capability in relation to its range (is it muffled or clear, etc).. ala something "note C, octave 3" is at 8000hz. If it's a collection of samples or how it's played back (loop points.. linear, ping-pong, etc), then that's a level above the sample data itself.. that's an "instrument" definition. What johannesmutlu is talking about is the playback rate, the streaming rate to the hardware DACs, for the final mixed output. Some GBA games mixed at an output rate of 8hz, 10hz, etc. From what I understand, the highest known commercial rate is 16khz (16000 number he's trying to communicate because doesn't understand what it means).