CPUs and architecture

What is the main reason GPUs are well suited to matrix operations and AI training and inference?

  1. ATheir clock frequency is over 100 times that of CPUs
  2. BThey have many simple arithmetic units and can process large amounts of data in parallel
  3. CThey have no memory
Show answer and explanation

Answer: B. They have many simple arithmetic units and can process large amounts of data in parallel

GPUs were originally designed to process many image pixels at once, with thousands of simple arithmetic units arranged in parallel. For workloads that repeat the same calculation on large amounts of data, such as matrix operations, they deliver high performance by exploiting this parallelism.

Practice this topic Back to topic

More CPUs and architecture questions

  1. If the cache access time is 10 ns, the main memory access time is 100 ns, and the cache hit rate is 90%, what is the average access time?
  2. In a CMOS circuit, if only the supply voltage is halved with everything else unchanged, what happens to dynamic power consumption?
  3. What do names like ARM, x86 and RISC-V mainly refer to?
  4. Which is a characteristic of RISC-V?
  5. What is the processor design called that issues and executes multiple instructions simultaneously in one clock cycle?