pouët.net

Guys, thanks for all the kind words -- it really means a lot to reenigne, VileR, Scali, and myself. We worked on the code for 6 months together, but separately it was roughly 100 man-years of appreciation and deep research into the capabilities of the machine and the video card over the past three decades. You grow up with this stuff and it leaves a deep impression on you.

It was also a deep honor to compete against such very strong entries in the compo, which you can see in the results -- it was practically a 4-way tie! If you haven't checked out the other three, please do so, they're great (although I don't know why Oxyron complained about only one disk... ever heard of compression?)

Technical blogs are forthcoming, but until then, here's some quick answers:

yzi: If you have real 4.77MHz 8088 and RGB, the demo will run but without correct colors, so you can still enjoy it somewhat. The kefrens bars, moire/interference, and end credits parts look nearly identical to composite out.

jmph: No dithering, no pageflipping, real 1024 colors from the display over composite output. VileR's core discovery, refined by reenigne to have double the colors.

introspec: Have already done mega research on ZX Spectrum beeper but you can't use 1-bit multi-chan engines as background music; they take up too much CPU time. Besides, reenigne came up with something much better than ZX Spectrum 1-bit engines, which you can hear in the end credits: It is a 16KHz 4-channel protracker chiptune player. It takes up 99.99% of CPU which is why the end credits are fairly simple.

Baudserver: We plan to update our blogs in the next few days :) I'll link to mine here when it's done.

F-Cycles: No flicker; 1024c mode is stable.

drzeissler: Sorry, make your own :-) 286+EGA is too "powerful" for us.

There may or may not be a final version. There are visual bugs in the party version which you can spot viewing the video multiple times (they're all my fault, actually :-), but we're fairly tired and burnt out, as is to be expected.

Technical details are online. This is only my perspective, though; others may write more detailed parts of their various contributions.

coda did the end mod. Trixter or reenigne should have it. You could also check his website at coda.s3m.us or find him on IRC (#mod_shrine I think).

More writeups from reenigne:

http://www.reenigne.org/blog/1k-colours-on-cga-how-its-done/
http://www.reenigne.org/blog/8088-pc-speaker-mod-player-how-its-done/

He might have been talking about the actual "cgademo".

Scali's writeup
reenigne's 3rd post (effects)

Well, we're trying to get more people to make oldskool pc demos :-)

While there are a ton of limits for this platform, the most annoying one IMO is lack of a decent sound device. And it's not as simple as "just add an Adlib" because 1. that likely wouldn't qualify for a pure oldskool compo if we did, and 2. Adlib writes are quite slow; updating FM music @ 60Hz can take upwards of half the raster time :-O so that's right out. Simpler devices like the C/MS (game blaster) don't take as much time to control, but those are even harder to find than Adlib/SB.

BTW I'd just like to say it has been an honor getting thumbs from some of my long-term scene heroes in this and other threads. Makes teenage Trixter extremely happy :-D

Oh! Sure, one sec... Ok, check the github repo for SWINGIN1.MON (first tune, Phoenix) and NO_ROADS.MON (second looping tune, virt).

You may be tempted to listen to them in Adlib mode so you can hear all channels properly, however that wasn't how they were composed. virt did that and wrote me this:

Why, monotone of course.

And another excellent writeup, from the artist:

http://8088mph.blogspot.com/2015/04/cga-in-1024-colors-new-mode-illustrated.html

Another article, this time about sprites and scrolling.

And the last article from me, the polygon part.

Effect wasn't really for vector-lovers, but rather an exercise in how to show large polygons while dealing with the abysmally slow memory speed of CGA.

If animation-presented-as-effect equals thumb down from you, then I guess you thumb down Planet Rocklobster as well? Oh well, I think there were plenty of cool parts in both demos that had nothing to do with animation.

__sam__: do check out my writeup (linked to in phoenix's last comment) and reenigne's articles too.
A belated thanks to everyone who voted - both at Revision and here :)

Hornet, CRTC and DESiRE are pleased to announce a final version of 8088 MPH which can be downloaded at http://www.reenigne.org/8088MPH_final.ZIP . Also a capture of the final version can be seen and heard at https://www.youtube.com/watch?v=hNRO7lno_DM . This final version has some new graphics, minor bug fixes and compatibility improvements, but it still breaks all your emulators.

Thanks for all your kind words about our demo!

One-click convenience:

http://www.reenigne.org/8088MPH_final.ZIP
https://www.youtube.com/watch?v=hNRO7lno_DM

Article explaining what was changed and new in the final version: https://scalibq.wordpress.com/2015/08/02/8088-mph-the-final-version/

jmph, I'm afraid there is no easy way to fix the audio sync - basically the 1-channel audio is not synchronized at all with the effects (except for the start of the first tune music when the title screen rolls down). We just happened to get lucky with the party capture, but the exact changeover point depends on the exact speed of the floppy drive's spindle and stepper motors. I suppose we could wait for a specific point in the music but we'd have to limit how long to wait in case somebody tries to run it on a machine with a very slow drive and we miss the right spot in the music.

Well, there is the ghett0-fix we used at the party: adjust the timing of effects until it looks good on a specific system.
In this case, we could adjust the 'mugshot' to appear shorter, so that the endpart should start at about the same place as before.



Yes, but I think the biggest impact is that by adding the new CGA graphics, some parts are now larger than before, so they take longer to load/decrunch/preprocess. Also, this may have changed the position of the files on disk (there are now more files as well), which may have resulted in larger seek times for certain parts.

More graphics-related verbiage, about the final version and other gory details: http://8088mph.blogspot.com/2015/08/8088-mph-final-old-vs-new-cga-and-other.html

I would like to make a short comment on the vectors.
Clearly we make use of some precalc/shortcuts, as every 8-bit platform does for complex shaded vectors at reasonable framerate. However:
1) We felt like people would expect to see vectors like this, and would not be impressed with anything less... "Hey, it's a PC, and you have 4.77 MHz! It should be way better than a C64!". In reality, it is far worse than a C64 because of the limitations of CGA (You need to draw every pixel, no shortcuts with custom character sets or sprites or such. You have no backbuffer, so you have to think of something clever to avoid flicker/tearing/etc. You have these horrible waitstates on CGA memory writes. And the pixel addressing is not linear, with even and odd scanlines in different pages).
2) It is actually sorta realtime, just probably not how you think. Eg, the pyramid object is actually drawn from a timer interrupt routine, while the data for the other objects is prepared in the background (the rules state that the entire prod has to run from a single 360K floppy, so the budget for the vector part was about 70K-80K, and had to include 4 sets of border graphics as well, two palettes and old/new CGA). We had to fight both the size limit and the time limit here. We couldn't put too much pre-cooked data on the floppy, and we also couldn't take too long loading/decrunching/precalcing, because our whole demo had to fit into 8 minutes, so parts had to follow eachother quite rapidly.
3) The loop is 64 frames, not ~45.