I am finally able ot run the "Loops of addition" benchmark on the 41CL. The code is the usual: LBL 00 + GTO 00, and runs for 60 seconds. Starting condition is X=0, Y=1, Z=1, T=1.
1X mode: 1,055
2X Turbo mode: 1,913
5X Turbo mode: 4,153
10X Turbo mode: 6,538
20X Turbo mode: 9,179
50X Turbo mode: 12,022
The results are not linear because the NEWT CPU automatically throttles back to 1X for those instructions that must run at this speed for compatibility. For a user program such as this, each program line samples the keyboard flag and the peripheral request flag, and both of these operations must be done at 1X. In addition, all LBL program lines move the "goose" in the display, and both the display select and the display update must be done at normal speed.
Even with these "speed bumps" the results are pretty good...
Monte
Loops of Addition benchmark for the 41CL
|
|
« Next Oldest | Next Newest »
|