Problem with CUDA PTX Register declarations vector type register are

Accelerated Computing CUDA CUDA Programming and Performance
1

I am trying to declare a vector register in PTX using the command as given below:

.reg v3 .f32 v;
or in general
.(space type) v1/v2/v3/v4 (type of register) regname;
here space type may be { .global , .local , .const , . reg etc }
type of register may be [ .f32,.f64, .s32 , .s64 , .u32 , .u64 etc }

i am getting error while bulding cubin file from ptx file
error : unimplemented version
my PTX version is 1.1 and other conditions are also fulfilled and i am able to build other type of registers too.

these syntax are used to declare a vector as per given in the Ptx_isa1.1.pdf document of NVIDIA but the commands are not working .

can anybody help me in this regard ?
pls reply soon.

Thanks

2

Section 10.5.3 in PTX ISA 1.1 document says -

“Vector declarations, initialization, and conversions are not supported.
Vector operands are not generally supported. The LD, ST, and TEX instructions do
support limited use of vector operands written using the tuple notation.”