"the second one can be also a SSE register or 128–bit memory location."

Also perhaps Fsave (now i am not sure) there was an instruction to store fpu registers and xmm registers even i think, and it would require 512 bits memory location. How to denote that on a label?

There is a table in the manual that lists all the size operators.

Thanks! Although it wasn't in the pdf coming with the windows package (i definitely cant remember 512 bites long datatype).

ok i just skipped that part.

So i write that i have skipped that part. But i actually read below another table with db, dw, dd, etc.

which did not contain all the size definition operator.
After your post i thought that this is a feature of fasm and it very well fits fasm this way not being a nasm styled for example (whatever way nasm is actually styled in)

so then i would have to define macros or labels or so for those size operators if i want a shorter form (so then some lables maybe dz or something or not so if i define it as a size operator).

Is that so?

If I understand the question the right way (“why are there no dX directives for 128+ bits?”), the answer seems to be quite easy: SSE and stuff, although they use large 128/256/512 bit registers don’t store single 128/256/512 bit values there: each such register is considered to be a set of 4/8/other number of 2/4/8 bytes-sized values, so there’s no need to provide new data definition directives that would be of no use anyway. (And, AFAIR, FASM uses 65 bits for its internal numeric values representation, so that might be another, but definitely not the main, reason.)

Where could you load 128 bit data to?

If we extend the GPRs to 128 bits, and also assume that the naming follows the qwerty pattern, then we get a new set of registers TAX, TDX, TCX, ..., T15.

In 128bits cup must be 32 registers !

Okay. Then we need to add a TEX prefix.

Now I wished the naming used S instead of T.

but fsave or something like this which saves various data of fpu or otherwise stores varried data. not simply repeats of 8 16 or 32 bits (or 64)