If i just want to load 5.0 in st (or xmmn for example)
[
push 5.0
fld [esp]
]
seems clear but doesn't work
It was said (in topic about float calculations at compile time) that FASM treats floats as dwords so why can't i push it?

fasm doesn't just treat floats as double words, it chooses the right conversion depending on context - compare "dd 5.0" with "dq 5.0" and "dt 5.0".

You must have been using "push 5.0" in 16-bit mode, since in 32-bit it works. In 16-bit mode the default size for push is 16-bit and there is no 16-bit floating point format - thus the error. Try "pushd 5.0" or "push dword 5.0".

I've now got it. I'm writing Win32 program and include one ASM file into another. I usually try to compile this standalone file for syntax checking. I've quite forgotten about default use16 mode. The main program compiles right.

Sorry for wasting your time...