I think ATI KMS (kernel mode-setting) was merged or enabled in 2.6.33?
That would explain why this kernel; why the same issue happens
suspending from the console; why that graphics-quirk doesn't work the
same (I don't have s2ram installed since I use pm-suspend (via GNOME),
but I guess that's what you meant in regards to -f, -p or -m).
Suspend ought to work though (did you try testing 'echo mem >
/sys/power/state' too? I think that's mostly the same as pm-suspend
nowadays, maybe), but it's understandable since there are new
code-paths, if indeed it's graphics that are at issue. If you want to
debug I'd suggest try grabbing 2.6.34 kernel26 and -firmware from
testing (e.g. you can download the packages from online), as there's a
chance it's been fixed there (my Intel KMS-related issues were, for example)
Also Did you search? A quick search for
2.6.33 suspend ATI Radeon HD 3600
led me to e.g. a bug here that links to an archlinux thread that
contains a couple suggested workarounds to try (e.g. does disabling
networking before suspending fix your resume?)
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=573118#10
-- those may be completely unrelated issues or then again they may not.