Reputation: 895

cuda and double-precision floating-point arithmetics

this is what i found: " Cores perform only single-precision floating-point arithmetics. There is 1 double-precision floating-point unit. "

is this true for all compute capabilities (versions) ?

Upvotes: 0

Answers (2)

Reputation: 895

section 5.4.1, table 5-1.

Upvotes: 0

Reputation: 15734

Single and double precision floating point accuracy and performance has continuously evolved and is different for each of the compute capabilities.

Upvotes: 2