Playability testing-
(Document version 2.2, last updated 10/01/02) |
When I first started to use the two available Gameboy (gb) emulators- Liberty (commercial) and Phoinix (freeware)- I decided to compare their relative performance so that I could select just one and free some memory on my 2Mb Palm. After a few tests I realised that different games seem to run better on different emulators; neither emulator was consistently superior. Looking over the information I had collected, I thought it would make a useful resource for both users and developers, providing quantitative playability data for different games. Therefore I decided, with the help of Bodo 'Phoinix' Wenzel, to publish these Playability Datasheets. What began as a simple task to free some memory has now become a bit of an obsession and I still have both emulators permanently installed on my Palm!
The rest of this document describes the source and meaning of the information on the datasheets. I think it is pretty straight forward, so only bother reading it if you want to produce your own datasheets.
HJ (hjack@usa.com) - Please feel free to contact me with suggestions, mistakes, additions, etc.
Please note that only text-based emails will be read and all attachments will be deleted automatically.
A brief description first qualified by the type of game and subtype, then a sentence describing the actually game play. Often games combine elements of different styles, in which case the most dominant influence is taken. If you think I've got it wrong, contact me. The main genres currently include:
Action; BatNBall; Blocks; Board; Card; Collection; Demo; Combat; Maze; Misc; Music; Pinball; Platform; Puzzle; Racing; RPG; Special; Sport; Strategy; Test; Utility; Virtual pet
Any supported special features such as: link cable, printer, rumble, etc.
Nintendo cartridges have a reference number printed along the edge of the cartridge label. These vary by game and country and may be of interest to collectors. Most of these have been sourced from the 'Gameboy Cartridge List' produced by Steve Kiratz on the web. Many numbers are still unknown, so please supply any you might have.
To avoid possible arguments over the quality of a game, a very basic three star rating system is used:
* - a poor game, probably not worth playing
* * - an average to good game
* * * - a very good game, that is definitely worth playing
Supported formats from the rom header. Often older gb games have been re-released with enhanced cgb or sgb features- I try to avoid them due to the extra size. Purely cgb games will obviously not be featured- emulate to preserve the old, not to pirate the new ;-)
BW - original black and white Gameboy
CGB - colour Gameboy
SGB - super Gameboy
An estimation of a game's release year, based on the last copyright year displayed in the game or similar investigative guesswork.
Licensee details taken from the rom header. Should correspond to the game publishers. For commercial cartridges 'Com' is written before the licensee. The associated hex code has also been included in case it benefits developers. The author(s) of public domain (PD) games will be credited in a similar format, even if the information is not taken from the rom header.
The game title taken from the rom header. For newer colour Gameboy games the last four letters encode extra information and hence are written in brackets.
Game cartridge version number, taken from the rom header. This is normally zero, unless a bug was found and new cartridges produced.
Internal cartridge specification taken from the rom header. Probably of interest to developers.
Country code will either be Japan or Non-Japan, taken from the rom header
Available languages in the game. For translated (hacked) games, the translation language is written in brackets.
Detailed information regarding the game emulation on the palm. For the purpose of testing the gb roms as they should run, I use No$GMB by Martin Korth for the PC. This is a superb emulator and generally considered the best.
Size of the rom file in kilobytes. On the palm I have also taken into account the size of the emulator, which is the second number in the bracket.
Playability is rated primarily on the presence of any major bugs and secondly by the speed of emulation.
Yes - game operates without any major bugs
No - not playable, see Bug row.
Too fast - Runs <100% of gb time (e.g. 29 sec) it could happen ;-)
Great - Runs 100-125% of gb time (e.g. 30-37s)
Ok - Runs 126-150% of gb time (e.g. 38-45s)
Slow - Runs 151-200% of gb time (e.g. 46-60s)
V.Slow - Runs 201-300% of gb time (e.g. 61-90s)
Too slow - Runs >300% of gb time and is considered not playable (e.g. 91s).
Any bugs found either in the original or emulated game. These are classified as follows:
Feature - not necessarily a bug, but a difference between a real gb and emulator
Cosmetic - graphical glitches of minor significance
Graphics - error in graphics, often minor, but could impact game play
Minor - A fault that probably won't hinder game play
Major - A fault that will seriously affect game play
Fatal - Causes the emulator to crash or behave dangerously
Please note that due to time constraints, games are only basically tested for bugs, i.e. probably only the first level has been played. If you have found a bug please notify the appropriate developer.
The procedure followed to assess the speed of emulation. When ever possible this test is based on actual game play, whilst also ensuring that the test is repeatable and reproducible.
This is the average of the fastest and slowest overclocked times, which can be used to gauge a 'typical' speed. The percentage of the actual time taken by the gb is then calculated, from which the playability rating is derived.
The time taken for a test to run is measured both under normal (Nm) and overclocked (Oc) states and at the highest and lowest screen refresh rates, to provide a range of speeds under each condition. To overclock my M100 I use the excellent Afterburner 3 hack, (from Daniel Wee) with '0-ws Core', 'Fast Mpx' and 'Fast Page' always selected, setting the speed to 'Norm' for normal or to 37MHz for overclocking. I typically only have around 4 hacks running. Using Benchmark (v2.0) by Neal Bridges, I achieve 136% for normal or 247% with overclocking. I consider this performance to be fairly representative of most Palm users.
Quality of emulated graphics:
Good - Equal to an actual gb
Ok- Some minor problems
Bad- Any serious graphics bug
Sound quality is assessed by whether it contributes to the game, rather than if it actually sounds anything like the gb. Recommended settings are written in brackets, occasionally with an in game setting written before the emulator setting (such as turning the background music off). The rating is made during actual game play and does not take account of any opening title music, etc. Some games are simply marked as 'not tested'- I really don't enjoy having my Palm screech those noises at me, but I will eventually re-test those games.
Currently reserved, but will feature Game Genie codes, patches, cheats, etc. depending on whether this is requested.
The fastest running emulator basically determines this, unless any serious bugs were found.
When and if I come across an alternative, either for the gb or a native Palm game, I will try to suggest them. (If I miss any, please inform me!)
General comments on the game and the game's emulation.
Who tested the game (in case other people want to get involved), the date and software versions.