Who Discovered the cycle 73, cycle 74 Hmove?

Omegamatrix · March 23, 2013

The cycle 73/74 Hmove trick is used in 2600 games to position objects without having the notorious little black line appear on the left side of the screen, which is often referred to as the comb or a hmove bar. The trick has one drawback in that objects can only be shifted to the right. I always thought it was a homebrewer trick, that was unknown to the original 2600 programmers. A while back Batari mentioned a reference he found here of cycle 73/74 Hmove's being discovered back in the old era day.

I just found a reference that strongly suggests that an M-Network programmer figured out cycle 73/74 HMOVEs back in 1983.

Eric Del Sesto, when working on a game called "Monkey Business" that was never released, said he developed a technique he called "Hidden V-clocks" to suppress the HMOVE lines. When asked about his technique, he said, "The M Network group had an accepted workaround for this which would cause those lines to appear on every scanline, thereby forming a solid black border at the left edge of the display. My Monkey Business prototype demonstrated a method to eliminate those lines completely. The trick was to write to certain registers in exact synchronization with the horizontal retrace timing."

The programmer didn't specifically say he used cycle 73/74 HMOVEs, but he either did that or he hit RESPx only and accepted the coarse positioning of 3 pixels, which I doubt.

I filed that in the back of my head and promptly forgot about it. Last night completely unrelated I was digging around the Stella Mailing List trying to find out which homebrewer actually discovered the early Hmove trick. I came across this message from Bradford Mott. He mentions He-Man using a cycle 74 hmove, and gave the location in the code it occurs. I tested He-Man in Stella, and sure enough it does a cycle 74 Hmove. It is a very deliberate act too. There is a long delay before the Hmove is executed where cycles are wasted until they got exactly to 71, and then the STA HMOVE instruction is executed completing on cycle 74. The goal seems to be to have no Hmove bars on the title screen. It seems to execute the cycle 74 Hmove trick at least 3 times, all from the same routine:

Was it just me that didn't know this? The emulator authors surely must have known this, otherwise the support for He-Man would have been broken for a long time.

I thought it was pretty cool at least one old era game used this trick. Mattel also had a bunch or prototypes that never got released like Anteater, Loco-motion, Tarmin, etc... So I looked at Rocky and Bullwinkle, and sure enough it is using it too:

There could be more Mattel games that use it. I haven't checked. The unreleased prototypes would be strong candidates since it seems Del Sesto discovered this in mid '83 according to the article. Nonetheless this is pretty cool, yes?

+SpiceWare · March 23, 2013

Was it just me that didn't know this? The emulator authors surely must have known this, otherwise the support for He-Man would have been broken for a long time.

Brad created Stella. That post was from when he was researching why He-Man's title screen wasn't displaying correctly. I worked with him back then as I used to maintain the OS/2 port of Stella.

Omegamatrix · March 23, 2013

Brad created Stella. That post was from when he was researching why He-Man's title screen wasn't displaying correctly. I worked with him back then as I used to maintain the OS/2 port of Stella.

Yes that's right, I know I didn't mention that in my post, but I was also thinking of z26. Was Brad the first "Modern" developer to notice this than? I believe Andrew Towers TIA notes came after this, but I'm not 100% sure.