Space Bunny Free
Every 8-pixel tile is drawn mirrored (nestest's menu reads "stset llA nuR"), which destroys the edges the video metric compares. A 2-hour run on a free model, kept as the first data point.
| Model | Space Bunny Free |
|---|---|
| Harness | opencode 1.18.35 |
| Thinking | medium |
| Time limit | 2 hours |
| Environment | NES Eval f74eee9 |
| Code | 3,538 lines of Rust: CPU, PPU, APU, mappers 0–4 |
| Tool calls | 78 oracle runs; reference dumps for 39 ROMs |
| Working time | 1h34m of 2h: a free-tier rate limit stalled opencode |
| Source | github.com/IlyaasK/nes-eval-attempt-space-bunny-free |
| Play | run it next to the reference in your browser |
Score
| Checkpoint | Overall | Procedural | Replay | Audio |
|---|---|---|---|---|
| 1h00m | 0.0578 | 0.0000 | 0.0072 | 0.2674 |
| final | 0.0594 | 0.0000 | 0.0098 | 0.2673 |
Replays
The strip under each game is the per-frame penalty: empty means the frame matched the reference, a full-height mark means it scored zero.
Falling 0.0057
5,798 of 5,800 frames differ, the first at frame 2 · audio 0.0234
Thwaite 0.0012
5,996 of 6,000 frames differ, the first at frame 3 · audio 0.0657
Double Action Blaster Guys 0.0202
5,386 of 5,400 frames differ, the first at frame 1 · audio 0.5935
Minekart Madness 0.0006
5,397 of 5,400 frames differ, the first at frame 3 · audio 0.1890
Nova the Squirrel 0.0310
5,384 of 5,400 frames differ, the first at frame 16 · audio 0.7063
Concentration Room 0.0003
5,999 of 6,000 frames differ, the first at frame 1 · audio 0.1556
Procedural
| Test ROM | Subsystem | Score |
|---|---|---|
nestest | cpu | 0.0000 |
instr-official | cpu | 0.0000 |
cpu-timing | cpu | 0.0000 |
instr-misc | cpu | 0.0000 |
nmi-and-brk | cpu | 0.0000 |
vbl-basics | ppu | 0.0000 |
nmi-timing | ppu | 0.0000 |
sprite0-basics | ppu | 0.0000 |
sprite0-timing | ppu | 0.0000 |
sprite-overflow | ppu | 0.0000 |
oam-read | ppu | 0.0000 |
apu-len-ctr | apu | 0.0000 |
apu-irq-flag | apu | 0.0000 |
apu-dmc-basics | apu | 0.0000 |
mmc3-clocking | mapper | 0.0000 |
holy-mapperel-mmc1 | mapper | 0.0000 |
holy-mapperel-uxrom | mapper | 0.0000 |
holy-mapperel-cnrom | mapper | 0.0000 |
holy-mapperel-mmc3 | mapper | 0.0000 |
apu-mixer-square | apu | 0.0000 |
apu-mixer-triangle | apu | 0.0000 |
apu-mixer-noise | apu | 0.0000 |
apu-mixer-dmc | apu | 0.0000 |
Failed verdict screens
Reference on the left, this candidate on the right. Showing 6 of 23; every score is in the table above.
nestest kevtris nestest: every official and unofficial 6502 opcode; Start runs the official-opcode tests and prints OK per group.
instr-official blargg instr_test-v5 official_only: results of all 151 official opcodes, including flags and addressing modes.
cpu-timing blargg cpu_timing_test6: cycle count of every instruction.
instr-misc blargg instr_misc: page-wrap, branch wrap and dummy reads.
nmi-and-brk blargg cpu_interrupts_v2/2: NMI hijacking a BRK.
vbl-basics blargg ppu_vbl_nmi/01: VBL flag set/clear and $2002 read behaviour.
