BTW that same loop (in a program) consistently takes 8 seconds on CloudT.

On Thu, Feb 12, 2026 at 12:41 AM B 9 <[email protected]> wrote:

> I’m testing only on Linux. I am surprised at the huge difference in time
> from leaving off the variable name in the NEXT. I posted a reference SPDCHK
> program (in the other thread) so we can all be talking about the same test.
> Essentially, it is doing:
>
> PRINT TIME$: FOR T=1 TO 3535: NEXT: PRINT TIME$
>
> Note: I was doing a lot of tests and got tired of waiting 30 seconds, so I
> shortened the loop to count to 3535, which takes almost precisely 10
> seconds on a stock Tandy 200.
>
> I don’t expect super accuracy, just something so that people can design
> games in BASIC on the Virtual T and know it’ll be roughly the same
> experience on the actual hardware.
>
> Speaking of emulators: Has anyone else tried out the MAME emulation of the
> Model T computers? It is still quite rough (for example, the screen looks
> terrible compared to the actual hardware) but its timing does seem to be
> accurate. Or rather, its “in game” version of the timing is right — if I
> use “PRINT TIME$” as my stopwatch, counting to 3535 the MAME emulated Tandy
> 200 thinks it takes ten seconds, however it seems to take longer on my wall
> clock.
>
> —b9
>
> On Wed, Feb 11, 2026 at 2:33 PM John R. Hogerhuis <[email protected]>
> wrote:
>
>> Yeah I agree. Between Linux and Windows difference it's likely a bug
>> unrelated to I/o waits
>>
>> -- John.
>>
>

Reply via email to