![]() |
|
Game Boy Advance Test Suite - Printable Version +- mGBA Forums (https://forums.mgba.io) +-- Forum: mGBA (https://forums.mgba.io/forumdisplay.php?fid=1) +--- Forum: Development (https://forums.mgba.io/forumdisplay.php?fid=11) +--- Thread: Game Boy Advance Test Suite (/showthread.php?tid=18) Pages:
1
2
|
RE: Game Boy Advance Test Suite - nitro2k01 - 08-02-2017 Has anyone tried these on NDS's GBA mode to see if the results differ from GBA hardware? RE: Game Boy Advance Test Suite - endrift - 08-02-2017 Yeah, they don't. RE: Game Boy Advance Test Suite - pepposole - 11-20-2017 Endrift-san! First of all, thank you for your incredible work you did so far! Sorry for the stupid question (i think i already know the answer) but... did you take reference values by running your Test Suite ROM on a real GBA? RE: Game Boy Advance Test Suite - endrift - 11-21-2017 Yes. RE: Game Boy Advance Test Suite - Arbee - 12-05-2017 So I know normal ARM CPU loads/stores follow the ARM standards for unaligned loads/stores on pre-v5 CPUs. Do DMAs? Do LDMIA/STMIA? RE: Game Boy Advance Test Suite - endrift - 12-05-2017 LDMIA/STMIA force alignment to 4 bytes IIRC. I haven't written the tests for that yet though. As for DMAs, I'm not 100% sure but I think they force align to at least 2 bytes. RE: Game Boy Advance Test Suite - Gamersmith - 09-11-2019 (11-07-2015, 06:01 AM)Spikestuff Wrote: correct, I apply it.. RE: Game Boy Advance Test Suite - Riven Skaye - 05-15-2026 Time to revive this. GBE+ is still under active development and aims to emulate a lot of peripherals and exotic features in addition to the normal games, with accuracy being a sub-goal to that. So, here's the results when ran on GBE+ version 1.10 compiled with fast instructions (I'll build and test without later) and test suite r0.76: Memory 754/1552 I/O Read 24/130 Timing 127/2020 Timer count-up 181/936 Timer IRQ 6/90 Shifter 140/140 Carry 71/93 Multiply long 52/72 BIOS math Crashed on a floating point exception DMA 932/1256 SIO register R/W 7/90 SIO timing Killed because it seemed to run endlessly Misc. edge case 1/12 Video - Basic 3 1:1 match Video - Basic 4 1:1 match Video - Degenerate OBJ transforms 1:1 match Video - Layer toggle Vertical height mismatch, see screenshots Video - Layer toggle 2 Large differences, see attached screenshots at the bottom Video - OAM Update Delay The green overlay mismatches vertical positions Video - Window offscreen reset Top half is fully white in the actual, instead of just the expected top-left quarter Attached screenshots are only the "actual" screens due to the limited amount on a post |