Does the all curves have this vulnerability?

[zounathan]

I collect the cpu cycle and nonce's length for curve NID_sect283r1, NID_secp256k1 and NID_X9_62_prime256v1. But the result doesn't indicate that the shorter nonce has the less time to computer the signature. Only the curve NID_sect163r2 that used in the paper has the right relationship between time and nonce's length.