Have you tried adjusting cycles on the fly while the games are running?

Yes, although that method strikes me as inaccurate. I'd rather use figures based on something.

Using this method, "RoboCop" seems playable at around 200-700 cycles. And "Test Drive II" at around... I'd be damned, 3000 cycles!

I'd still like to hear other opinions, really. Even better if it's people who played the games on real hardware.

The problem is that with DOSBox not directly emulating bottlenecks such as the slowness of video memory, there is no direct translation of release year/MHz to DOSBox cycles. Finding a good cycle value has to be done empirically; there is no MHz-to-cycles formula that is "accurate".

I remember back in 1993 that Test Drive III played either too fast (with Turbo button on) or too slowly (with Turbo button off). I am pretty sure that there is no computer at which the "seconds" display really counts up at the speed of an actual second while the game still plays at an acceptable speed. The game is just badly-programmed. Not just in terms of speed, and the bionic-mosquito-like engine sound on Adlib, but also in terms of the game world being seemingly drawn at too small a scale, making you feel like you are driving Gulliver's car on the roads of Lilliput.

NewRisingSun wrote on 2021-01-03, 19:27:

The problem is that with DOSBox not directly emulating bottlenecks such as the slowness of video memory, there is no direct translation of release year/MHz to DOSBox cycles. Finding a good cycle value has to be done empirically; there is no MHz-to-cycles formula that is "accurate".

Yes, I can understand that. Still, an aproximated value can be figured out. If you've used DBGL, the author created a few templates based on actual computer models, and those can be very useful. They've saved my bacon more than once, believe me.

NewRisingSun wrote on 2021-01-03, 19:27:

I remember back in 1993 that Test Drive III played either too fast (with Turbo button on) or too slowly (with Turbo button off). I am pretty sure that there is no computer at which the "seconds" display really counts up at the speed of an actual second while the game still plays at an acceptable speed. The game is just badly-programmed. Not just in terms of speed, and the bionic-mosquito-like engine sound on Adlib, but also in terms of the game world being seemingly drawn at too small a scale, making you feel like you are driving Gulliver's car on the roads of Lilliput.

Very interesting. I hadn't contemplated the possibility that the game was just badly programmed. And it's not like I haven't seen a few examples over the years.

If you have not already done so, be sure to obtain the version 2.0 of the game Test Drive III, which allows you to select the sensitivity of the steering wheel by pressing F3. I have found that helps a lot at making the game at least somewhat enjoyable.

Neat! Didn't know about that.