2

Is there any chance of SMT (simultaneous multithreading) having worse performance than normal CPU?

For example, is it possible for a normal CPU to run two programs one by one and have better performance than running these two programs simultaneously on a CPU that supports 2-way SMT?

Improve this question
2
  • 2
    Consider the cache.
    – Dan D.
    Jan 28, 2015 at 7:35
  • @Mansumen Can you clarify what you mean by a normal CPU?
    – Ramhound
    Jan 29, 2015 at 12:12

1 Answer 1

Reset to default
4

is it possible for normal CPU to two programs one by one to have better performance than running these two programs simultaneously on CPU

It would be the case when the programs fit in the CPU cache(s) individually, but not combined. They would require more RAM load/stores when running simultaneously. Remember that RAM accesses have a relatively high latency (in CPU cycles).

Improve this answer
1
  • 1
    Conflict misses are also a problem. DRAM accesses can also have a conflict effect increasing the amount of PRECHARGE and row ACTIVATE activity, reducing memory bandwidth. Branch predictors are also subject to conflict and capacity issues.
    – Paul A. Clayton
    Jan 28, 2015 at 12:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.