Didn't someone say that DOSBox is extremely single threaded? In which case, on the 64Bit processor, DOSBox would effectively be running on a single-core 1.6GHz CPU, correct?

"I just ran Daggerfall with 20k cycles and while it still ran, it would have given me a headache in a short period of time due to screen-tearing and such."

Correction, here. I haven't played Daggerfall in awhile and the last time I did was with dosbox .65, in which dynamic core would cause all sorts of graphical abnormalities in Daggerfall. It runs fine at 30K on a 3.2GHz system in version .72 with dynamic core set. ~24K was about as high as you could get it on normal core on the older versions (and probably still is).

In any case, 20K cycles on a dynamic core in recent versions (And 24K on normal core, I guess) was fine for everywhere except in buildings, where I did experience notable slowdown for some reason.

But yeah, the same basic principle still applies. 30K is great on the 3.2GHz system because it's faster. On the slower systems, 30K cycles is more than they can handle, I guess, so you actually see a performance increase with fewer cycles.

However you have to keep in mind that Daggerfall wasn't really coded that efficiently anyway.

The slowdown in buildings happens in the original game also, if playing on an older system. That problem is just Daggerfall not being optimized to the max, not DOSBox. As for 30k maxing them out, that is incorrect. I already checked that. Again, the 1.8GHz system maxes around 51~52k and the 1.6GHz AMD laptop maxes in the mid-40's, so 30k is fine on them. I can start Daggerfall and go windowed with it at 30k cycles and watch Task Manager report around 60%~70% usage on one core, so it isn't maxing out. I have also left it fullscreen and turned on logging in the MMC and never seen it go above 70%, so let's throw out the idea that the CPU is stressing or maxed. The problem lies within DOSBox itself because the simple fact is that 30k on this system is not 30k on the others, and none of them are maxing out.

TGS, would you care to define for me what exactly a cycle is, as I seem to be missing something here.

Certainly, and FYI this is where AMD had an edge on Intel chips for quite some time. Each processor tick, X cycles of data may be processed. Larger cycles means more data processed at the same speed. To put it in simpler terms, let's assume processor A has 1kb cycles at 2GHz, but processor B has 2kb cycles at 1GHz. Basically, each one has the same processing power even though the frequencies are generations apart. Again, this is how AMD could claim that their 1.4GHz chip was as fast as the Intel 1.8GHz chip years ago, and continued to be able to do so for quite some time.

My personal theory is that DOSBox passes a cycle of data directly to the physical CPU, thus limiting the size of the cycle tot hat of the CPU. This would explain why the AMD Turion is better with DOSBox than the old P4/1.8GHz system. I cannot prove this without reviewing the source, but it would explain my issues. I do not have two chips of the same generation to test this theory with. Your two chips (2.2GHz Centrino and 3.20GHz P4) are of the same generation and your configs work on both. I only buy a new system when it's time for an upgrade, so I have an 8086, 286, 486, P200, P2/233, etc etc up to this P4. Soon I will have a Core2Quad system.

Oh and I had to think about what TGS stood for, for about two minutes before I figured it out. You can call me Seph if you want to, my roommate did before I moved out.