In the emulator

Reading the binary tells you what the code does. Some things only tell you what the code is worth once it runs, and that is what openMSX is for. Everything on this page is measured, and the raw tables are in medidas/.

The house rule

The scripts in tools/omsx_*.tcl only read memory. The ROM is never touched: no patches, no changed bytes, nothing forced to execute.

The one exception is declared and is a measurement, not a modification: tools/omsx_prueba_n.tcl stops the machine at 0x422E —right after 0x4216 sets the round and the event to 1— and writes into 0xE016 the number of the event you want to see. 0xE016 is the game's own variable, and from there on the cartridge runs by itself, with its data and its code. What you save is having to qualify in the three previous events to reach the fourth.

The four events

With that, they were photographed one by one (medidas/pruebas.txt, captures in docs/imagenes/):

0xE016	label that comes up	record it paints
1	110 HURDLERS	12 SEC 93
2	JAVELIN THROW	96 M 72
3	HIGH JUMP	02 M 36
4	1500 METER	03 min 31

All four records match byte for byte the table at 0x517B, which INIT copies to 0xE040: 00 12 93 / 00 96 72 / 00 02 36 / 00 03 31. So what is on screen and what the binary says are the same thing, checked from both ends.

The capture confirms the typo, but it is not the proof

On the first event's screen you can read 110 HURDLERS, with an extra R. A capture on its own would prove nothing: it could be the emulator, the crop or the resolution.

The proof is in the data. The rotulo_110_vallas label starts at 0x64B1, and behind its header byte it carries the glyph string

0A 01 01 00 0A 12 1E 1B 0E 16 0F 1B 1C 0A

and in the cartridge's font (0x6197) those indices read 110 HURDLERS . The typo is written into the ROM. The capture only shows the result.

The clock was worked out for 60 Hz

SUMA_AL_RELOJ (0x56D7) adds to the BCD counter at 0xE0A9-0xE0AB once per frame, alternating 0x0167 and 0x0166. If the top byte is seconds and the middle one hundredths, that is 1.67 and 1.66 hundredths: an average of 1.665, which is one sixtieth of a second to four figures.

It is the same routine as the sibling cartridge's, and over there it is measured by running the event mashing the space bar and sampling the counter once per emulated second:

machine	Hz	game clock per real second
Philips VG-8020 (PAL)	50	0.8325 s — runs 16.75 % short
C-BIOS MSX1 JP (NTSC)	60	0.9990 s — spot on

Both figures agree with the arithmetic: 50 × 1.665 = 83.25 and 60 × 1.665 = 99.9.

The consequence: on a European MSX the times the cartridge shows are not seconds. A 110 hurdles timed at 13.00 by the game took 15.6 real seconds. The qualifying marks at 0x500A and the world records at 0x517B are meant for the Japanese machine.

And mind reading it backwards. That is a period, not a speed: the game does not run slower in Europe —the athlete moves across the screen at whatever speed 50 Hz allows— what happens is that the stopwatch counts short. Those are two different claims and only one of them is measured.

Four lines, the same as in the sibling cartridge: one player or two, and joystick or keyboard. Whichever you pick leaves a number from 1 to 4 in 0xE01B, and from that come the number of players (0xE010) and the controller (0xE004).

How to repeat it

PRUEBA=3 openmsx -machine Philips_VG_8020 -cart hyperolympic2.rom \
        -script tools/omsx_prueba_n.tcl
openmsx -machine Philips_VG_8020 -cart hyperolympic2.rom \
        -script tools/omsx_pruebas.tcl