data level parallelism and gpu architecture

MCQsQuestion.com has 11 Question/Answers about Topic data level parallelism and gpu architecture

Median vectorization getting improved from about 70% to about

Median vectorization getting improved from about 70% to about
  • A. 75%
  • B. 80%
  • C. 90%
  • D. 100%
  • Correct Answer: Option C

When vector gets longer than max length of its, then a technique called

When vector gets longer than max length of its, then a technique called
  • A. Buffer-grained multithreading
  • B. Control dependences
  • C. Strip cutting
  • D. Strip mining
  • Correct Answer: Option D

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

A register which can easily contain programs which naturally have very shorter vectors instead of max size a computer architecture can support, is known as
  • A. Gather-scatter operations
  • B. Scatter store
  • C. Fixed-length vector register
  • D. Variable-length vector register
  • Correct Answer: Option D

Cache benefits including Ray casting (RC) is only

Cache benefits including Ray casting (RC) is only
  • A. 0.6 times faster
  • B. 0.66 times faster
  • C. 1.4 times faster
  • D. 1.6 times faster
  • Correct Answer: Option D

Loops when gets vectored then they do not have dependences among iterations of a loop, which are called

Loops when gets vectored then they do not have dependences among iterations of a loop, which are called
  • A. Data dependences
  • B. Control dependences
  • C. Loop-control dependences
  • D. loop-carried dependences
  • Correct Answer: Option D

A solution of problem for representing parallelism in an algorithm is

A solution of problem for representing parallelism in an algorithm is
  • A. CDA
  • B. PTA
  • C. CUDA
  • D. CUD
  • Correct Answer: Option C

Horizontal and diagonal roofs when meeting offers, insight into computer is known as

Horizontal and diagonal roofs when meeting offers, insight into computer is known as
  • A. Ridge point
  • B. Gather operations
  • C. Scatter operations
  • D. Commit
  • Correct Answer: Option A

Level of detail getting dropping down, PTX assembler sets a marker naming as

Level of detail getting dropping down, PTX assembler sets a marker naming as
  • A. Data synchronization
  • B. Branch synchronization
  • C. Control synchronization
  • D. Thread synchronization
  • Correct Answer: Option B

NVIDIA thought that ‘unifying theme’ of every forms of parallelism is the

NVIDIA thought that ‘unifying theme’ of every forms of parallelism is the
  • A. CDA thread
  • B. PTA thread
  • C. CUDA thread
  • D. CUD thread
  • Correct Answer: Option C

An obvious and sufficient test for dependence’s absence is the

An obvious and sufficient test for dependence’s absence is the
  • A. Gross common divisor (GCD)
  • B. Greatest commit divisor (GCD)
  • C. Greatest common division (GCD)
  • D. Greatest common divisor (GCD)
  • Correct Answer: Option D

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

PTX instruction having format: opcode.type d, a, b, c; where d is the
  • A. Destination
  • B. Domain
  • C. Data
  • D. Double
  • Correct Answer: Option A