Q. Every dependence analysis algorithms working on the supposition that array indices are

View solution

Q. The instruction d = 0 - a; is the type of

View solution

Q. NVIDIA thought that 'unifying theme' of every forms of parallelism is the

View solution

Q. The aggregate register file has a size of

View solution

Q. Each bank register of fixed-length maintaining a single vector, is referred to as

View solution

Q. Threads being blocked altogether and being executed in the sets of 32 threads, called a

View solution

Q. A register which can easily contain programs which naturally have very shorter vectors instead of the max size a computer architecture can support, is known as

View solution

Q. An obvious and sufficient test for the dependence's absence is the

View solution

Q. Pipeline having stalls are required only single time per vector instruction, instead of once per

View solution

Q. The memory bandwidth of the GPU is,

View solution

Q. The two iteration indices, j and k, and these both are in the limits of the FOR loop, for-loop index variable which runs from m to n i.e

View solution

Q. Cache benefits including Ray casting (RC) is only

View solution

Q. The PTX instruction having format: opcode.type d, a, b, c; where d is the

View solution

Q. When every unit is fully pipelined, and a new operation can be started on each clock-cycle known as

View solution

Q. When elements are tends to be operated on in the dense form, then the sparse vector gets stored in an expanded form by a

View solution

Q. The vector instructions which can be potentially executed altogether, is known as

View solution

Q. If exploiting and finding the parallelism, across branches require scheduling code, a substantially very difficult algorithm is

View solution

Q. Having load before the store in a running program order, then interchanging this order, results in a

View solution

Q. When instruction i and instruction j are tends to write the same register or the memory location, it is called

View solution

Q. Branch-selected entries in a (2,2) predictor, which is having a total of 8K-bits in the prediction buffer are

View solution