Skip to content

02 FS UAE Configuration

Paul Bernicchi edited this page Jul 28, 2026 · 1 revision

FS-UAE configuration

Working config: config/AMIX-A3000UX.fs-uae.

The FS-UAE Launcher rewrites that file alphabetically and strips comments when it saves, which is why the reasoning lives here and not inline.

Settings

Setting Value Why
amiga_model A3000 ECS, A3000 chipset, built-in WD33C93 SCSI
kickstart_file amiga-os-204-a3000.rom KS 2.04 (A3000) 37.175
chip_memory 2048 2 MB chip
motherboard_ram 16384 Ceiling. AMIX recognises 4–16 MB; above that the kernel mis-maps the SCSI controller
uae_cpu_model / uae_mmu_model 68030 The MMU is mandatory — AMIX will not run without it
uae_fpu_model 68882 As per the real A3000UX
uae_cachesize 0 JIT off. JIT + MMU panics the kernel
uae_cpu_compatible etc. false "More compatible" causes kernel panics and sort: fatal: line too long during package install
hard_drive_0_controller scsi6 SCSI ID 6 is hard-coded in the installer
hard_drive_0_type rdb Without it the installer detects no disks
network_card a2065 The only NIC AMIX supports out of the box
uae_gfxcard_type A2410 Colour X; not exposed by FS-UAE's own graphics_card option
uae_gfxcard_size 4 Required. Defaults to 0, and the board is then silently never instantiated

Settings to avoid

uae_cpu_speed = max breaks booting. With the CPU unthrottled the WD33C93 never completes its handshake and the boot never leaves the SCSI probe: white screen, no MMU enabled in the log. Leave the option out entirely.

Do not put a second hardfile on the SCSI bus. Any additional hard_drive_N hangs the machine during the Kickstart bus scan, before AMIX gets control. Verified by A/B with an otherwise identical config: two drives gave no MMU enabled after 110 s; the boot disk alone reached it inside 70 s. Tried both with no RDB and with a valid AMIX-style RDB — the disk label is not the issue. A tape at ID 4 is fine (see Moving files in).

Both are findings from this build.

Checking a boot

68030 MMU enabled in the log is the reliable signal that the kernel took over:

grep -c "MMU enabled" ~/Documents/FS-UAE/Cache/Logs/fs-uae.log.txt

Do not diagnose with SCSI command buffer overflow!. It floods the log — gigabytes per session — during hung and healthy boots alike. It is noise.

Boot takes roughly 70 s, or 145 s with the tape attached; the extra time is the bus probe. Expect one CPU reset partway through, which is the normal two-stage bootstrap.

Clone this wiki locally