Thanks everyone. At least we keep making progress, and can "apply" what we learn about Canoe for other bugs.
Sink or Swim -- yes! This suggests Canoe runs at same "slow cycles" speed as Geiger. But VBlank triggers sooner on ~224 and HBlank also (maybe 0.25-0.50 scanline early). Going to get writeup on this after F1 Grand Prix and move on. Helps explain Cacoma Canoe error.
ActRaiser 2 is another slowrom variant.
e3e0
0A42 00 0D42 01
0A42 00 0D42 00
80e2eb lda $80e302,x [80e3e0] A:4200 X:00de Y:4209 S:01fb D:0000 DB:81 nvmxdIzC V: 15 H: 166 F: 0
80e2ef bmi $e300 [80e300] A:420a X:00de Y:4209 S:01fb D:0000 DB:81 nvmxdIzC V: 15 H: 166 F: 0
80e2f1 tay A:420a X:00de Y:4209 S:01fb D:0000 DB:81 nvmxdIzC V: 15 H: 182 F: 0
80e2f2 sep #$20 A:420a X:00de Y:420a S:01fb D:0000 DB:81 nvmxdIzC V: 15 H: 196 F: 0
80e2f4 lda $80e304,x [80e3e2] A:420a X:00de Y:420a S:01fb D:0000 DB:81 nvMxdIzC V: 15 H: 258 F: 0
80e2f8 sta $0000,y [81420a] A:4200 X:00de Y:420a S:01fb D:0000 DB:81 nvMxdIZC V: 15 H: 258 F: 0
80e2fb inx A:4200 X:00de Y:420a S:01fb D:0000 DB:81 nvMxdIZC V: 15 H: 294 F: 0
80e2fc inx A:4200 X:00df Y:420a S:01fb D:0000 DB:81 nvMxdIzC V: 15 H: 308 F: 0
80e2fd inx A:4200 X:00e0 Y:420a S:01fb D:0000 DB:81 nvMxdIzC V: 15 H: 322 F: 0
80e2fe bra $e2e9 [80e2e9] A:4200 X:00e1 Y:420a S:01fb D:0000 DB:81 nvMxdIzC V: 15 H: 336 F: 0
Writes 420a = 00 (irq timer), 420d = 00 (slowrom). Still want to find specific "copier protection" algorithm.
Buffet is so big right now. And have some sidework to take care of too. Will manage somehow.
MMX2 slowdown ...... going to whip up quick FastRom patch later and see what it does. ^^
And if you can find emus that show Canoe errors, please mention which emu reproduces. Would require less nagging on me for testing.

Sink or Swim - dual patch
http://s000.tinyupload.com/index.php?file_id=89749534021598941660And writeup coming post 1.