Golden Era Game of the Week 8/6/05: Robotron: 2084

Golden Era Game of the Week
http://www.gegotw.net/

Moderators: Marco Marocco, Weehawk

User avatar
Weehawk
MARPaholic
MARPaholic
Posts: 2554
Joined: Wed Jun 25, 2003 7:43 am
Location: Devil's Canyon
Contact:

Post by Weehawk »

Results:
1st: BBH
2nd: Don Hayes
3rd: TJT


No last minute surprises.

BBH takes 1st and all 6 bonus points for 16 total.

Don't forget about the further bounties.

http://www.gegotw.net/robotronresults.html

http://www.gegotw.net/news.html
John Cunningham (JTC)
Image
User avatar
Weehawk
MARPaholic
MARPaholic
Posts: 2554
Joined: Wed Jun 25, 2003 7:43 am
Location: Devil's Canyon
Contact:

Post by Weehawk »

As an added incentive (and just because I think this is a cool design):

Anyone bettering BBH's 609,250 before October 31, 2005 will receive this Robotron T-shirt:


Image
Your choice of size.

Requirements:

1) Only MARP members who have submitted at least one recording prior to this announcement are eligible.

2) Only one T-shirt will be awarded per member.

3) Inp and .wlf files must be submitted within 48 hours of recording and will be posted as attachments in this thread. (You may do it or I will do it for you).

4) You must, of course, be willing to give me your mailing address.
John Cunningham (JTC)
Image
User avatar
BBH
Editor
Posts: 1582
Joined: Wed Feb 06, 2002 7:06 am
Location: Portland, Oregon
Contact:

Post by BBH »

am I eligible if I beat that score myself? :P
User avatar
Weehawk
MARPaholic
MARPaholic
Posts: 2554
Joined: Wed Jun 25, 2003 7:43 am
Location: Devil's Canyon
Contact:

Post by Weehawk »

BBH wrote:am I eligible if I beat that score myself? :P
Certainly.
John Cunningham (JTC)
Image
User avatar
Weehawk
MARPaholic
MARPaholic
Posts: 2554
Joined: Wed Jun 25, 2003 7:43 am
Location: Devil's Canyon
Contact:

Post by Weehawk »

Just a reminder that there's a little over a week left on the further bounty here.

Also, if you use WolfMAME .101, your recording would be eligible to submit to Twin Galaxies, should you choose to do so.
John Cunningham (JTC)
Image
User avatar
Ahigh
Button Slapper
Button Slapper
Posts: 21
Joined: Thu Jul 28, 2005 6:22 am

Post by Ahigh »

I just hit a million at difficulty 10 on a real machine for the first time today. I couldn't find anywhere else on the 'net that would appreciate this feat, but I had to announce it somewhere as it's been a big quest of mine for several years now.

--
- Ahigh
BeeJay
MARPaltunnel Wrists
MARPaltunnel Wrists
Posts: 587
Joined: Fri Jun 17, 2005 1:40 am

Post by BeeJay »

Do you still find it easier on the real machine than even the latest Mame releases where they have some fixes for the difficulty problems?

Either I am total crap at this now, or it is still slightly harder than the actual machine was on the same settings.... either that or there is still no substitute for the actual arcade controls into an actual arcade machine versus going into a PC and being interpreted as keypresses.

Cheers,
BeeJay.
User avatar
BBH
Editor
Posts: 1582
Joined: Wed Feb 06, 2002 7:06 am
Location: Portland, Oregon
Contact:

Post by BBH »

I feel the same way, that even with the fixes in MAME it's still considerably harder compared to the arcade. I've never played an arcade machine on level 10 though, but on default difficulty it is much tougher for me to "marathon" it on MAME than it is on the actual machine.
User avatar
The TJT
MARPaholic
MARPaholic
Posts: 2479
Joined: Wed Mar 06, 2002 10:56 am
Location: 20 Grand Palace

Post by The TJT »

Aaron, It's now possible to submit at MARP using diff=10 with robotron.zip.
DBH and BBH have pulled off passing the big M here. Might be something to do with initials.
User avatar
Ahigh
Button Slapper
Button Slapper
Posts: 21
Joined: Thu Jul 28, 2005 6:22 am

Post by Ahigh »

I plan to Marp up what I got skill-wise soon, and I'll report back. I have a lot of stuff to say about all this, but I have to wait until I get all my ducks in a row.

Thanks for the interest and responses. I'll report back soon.

--
- Ahigh
User avatar
Ahigh
Button Slapper
Button Slapper
Posts: 21
Joined: Thu Jul 28, 2005 6:22 am

First tidbits

Post by Ahigh »

On my laptop just now, robotron runs significantly faster under emulation than the real cabinet does. This causes the game to be more difficult under emulation.

It reminds me of a speedup chip for asteroids.

My understanding is that the real robotron slows down depending on the amount of activity, and that, in general, the tuning of how much it slows down (if it even slows down at all on my pc) is off.

It seems to me that the correct approach to get a 100% faithful emulation under mame you should create a 6809 interpreter that counts the number of instruction cycles and tries to hit real-time for the clock-speed of the 6809. I don't think cache is involved on a 6809, so it should be easy enough or at least doable.

The other thing that I don't know yet is what's the total time in milliseconds of the feedback loop on a real cabinet versus under emulation.

You can figure this out to the nearest 16 milliseconds using a video camera that can see both the controls and the video screen.

The smallest feedback loop would be if on the original hardware, there is drawing to the screen memory directly with no double buffer (likely as doubling the video ram would be expensive back then). In addition, if the custom input hardware on the original machine is set up to sync the polling on the controls to the time immediately before the movement of the players character from the game code's update player position routine, that could eliminate an additional 8.3 average milliseconds compared to using, for example, a USB connected controller with a regular 10ms polling period that is not synchronized to the 16ms update rate of the video display. IE: USB and other PC input devices poll at independent update rates of the video display which can cause beat frequencies obstructing optimal firing timing.

On level 10, it is critical that you can push in a direction at the exact moment in time to kill a spheroid that is trying to avoid your path of bullets. It's smart!

The beat frequency between 100 hz and 60hz would cause a variable lag in the input, where the actual lag value would vary from 0ms to to 16.5 milliseconds along the path of a sine wave at 40hz centered at the average lag of 8.3 milliseconds. IE: you would get sporadic lag in your inputs, with lag values alternating at a 40hz frequency.

I have no idea about the details of how Mame is implemented, and it is possible read usb input other ways, but it is less likely that Mame is doing it this way just wild-ass-guessing (for usb devices).

I don't know about PS/2 keyboard (EG: hotrod) re: typical poll rates or if it is typically read through interrupt versus polling and if polled what rate the polling is done.

But I do have a hotrod and a groovy game gear usb, so I should be able to tell if the hotrod is better or not.

in opengl, drawing to the frontbuffer is easy, but in directx, I think frontbuffer access is maybe more complicated at best. Assuming that Mame has a priority of keeping all games more similar, it might be using double buffering on the Mame version and single buffering on the real hardware(?)

But, absolutely, at least on my machine with whatever default settings I had on wolfmame32-0100 on my laptop, it's noticably faster, and thus more difficult. Even if Mame is doing single buffer, the faster speed would more than likely be more difficult than the slower speed. Unless, of course, you have a problem adapting to firing at a slower pace when the machine slows down as you do on a real machine (fire too fast, and you miss bullets .. normally I think there's at least 33ms between shots, but at half speed, you can actually wiggle the joystick between two directions so fast that a bullet doesn't come out).

These are my immediate thoughts.

I will hookup up hotrods and groovy game gears tomorrow and report back soon.

--
- Ahigh
User avatar
mahlemiut
Editor
Posts: 4183
Joined: Mon Feb 04, 2002 10:05 pm
Location: New Zealand
Contact:

Post by mahlemiut »

I'm pretty sure MAME does cycle accurate emulation of the 6809. In any game that slows down due to the speed of the emulated CPU, then MAME is likely to be running it just a touch faster than the actual arcade, as wait states aren't emulated.

As far as I know, no driver in MAME emulates wait states, largely due to there not being any easy way to implement it in MAME as yet.

As for the controls, MAME only polls digital inputs once per frame. Typically, this is all most games need. MAME checks analog inputs only as the game requests the data, and can be (but isn't necessarily) numerous times per frame (for accuracy). There is certainly some inherent delay in control responsiveness, but this can't really be helped.
- Barry Rodewald
MARP Assistant Web Maintainer
Image
User avatar
Ahigh
Button Slapper
Button Slapper
Posts: 21
Joined: Thu Jul 28, 2005 6:22 am

Post by Ahigh »

mahlemiut wrote:I'm pretty sure MAME does cycle accurate emulation of the 6809. In any game that slows down due to the speed of the emulated CPU, then MAME is likely to be running it just a touch faster than the actual arcade, as wait states aren't emulated.

As far as I know, no driver in MAME emulates wait states, largely due to there not being any easy way to implement it in MAME as yet.

As for the controls, MAME only polls digital inputs once per frame. Typically, this is all most games need. MAME checks analog inputs only as the game requests the data, and can be (but isn't necessarily) numerous times per frame (for accuracy). There is certainly some inherent delay in control responsiveness, but this can't really be helped.
Well, I played, but I didn't do very well at all yet. I got 305K on my second game and it went downhill from there. I was using perfect 360's with slikstick stainless steel balltop replacement shafts on a groovy game gear usb keywiz 49-way hardware setup (switches appear as joystick buttons rather than as keyboard buttons). Part of the problem was the controls for sure. The throw is so long and the weight of the ball so heavy, you're fighting inertia of the stick itself. My real cabinet has happ competitions, and the clicky feedback helps a lot more than i realized. And my motor memory is tuned to those by now. I got the replacement happ comps from tornado terry out in texas, and I highly recommend them as the best robotron controllers at this point.

I have suzo joysticks and I think I will order another pair of tornado terry's red-balltop happ competitions and try both of those.

The game didn't seem as fast using joysticks as it did using the keyboard, so there may have been a psychological factor in not being as good at the keyboard and thinking it was just too fast because my motor memory isn't there.

I'm hopeful I can build some controls and get back in there, but so far, I can't translate my million score on a real cabinet into wolfmame. I'll keep the forum posted.

Also, I was playing on an LCD on a lapop, and I know there is resizing latency when you put that laptop into any non-native res, so running in a window I think is the way to go on that to avoid too much upscaling-related display lag from non-native res. Open to suggestions, and willing to run on a crt too, which i may do to eliminate as many variables as possible.

--
- Ahigh

The setup I played on
User avatar
mahlemiut
Editor
Posts: 4183
Joined: Mon Feb 04, 2002 10:05 pm
Location: New Zealand
Contact:

Post by mahlemiut »

Like I said, digital inputs (joysticks, buttons) are likely to be once per frame, and analogue inputs (trackballs, spinners, pedals) maybe once per frame, maybe more. Certainly relative analogue inputs would need more than one check per frame, to keep it as accurate as possible. Absolute inputs would probably only need one check every frame (or so), as whenever it's checked, the value returned is the exact position it is in.
- Barry Rodewald
MARP Assistant Web Maintainer
Image
User avatar
Ahigh
Button Slapper
Button Slapper
Posts: 21
Joined: Thu Jul 28, 2005 6:22 am

Post by Ahigh »

mahlemiut wrote:Like I said, digital inputs (joysticks, buttons) are likely to be once per frame, and analogue inputs (trackballs, spinners, pedals) maybe once per frame, maybe more. Certainly relative analogue inputs would need more than one check per frame, to keep it as accurate as possible. Absolute inputs would probably only need one check every frame (or so), as whenever it's checked, the value returned is the exact position it is in.
On the accuracy point, a zero that should be a one because of the progression of time is arguably more wrong than a 0.35 that should be a 0.37...

On my real cabinet, I scored 1.93 million April 27, 2007. I have also scored over a million on that cabinet 4 times by now (again diff 10, xtra 25k, 3 men to start). I haven't yet gotten my controls and displays all set up to make a better attempt at home, but I will be trying soon.

--
- Aaron
Post Reply