How does hypertreading affect project build time? With or without templates, is there a tangible acceleration?



The author posted a question in Programming

How does hypertreading affect project build time? With or without templates, is there a tangible acceleration? and got a better answer

Response from
1. There are different build systems: single-threaded and multi-threaded. Obviously, HT is useless for a single-threaded builder 2. There are different processors. A dual-core processor becomes like a 4-core processor with HT. And a 16-core processor becomes a 32-core processor. But it's not a fact that the builder will need 32 cores, maybe 8 is enough 3 There are different disks: SSD and HDD. If the drive is HDD, it will limit the speed of the build, not the processor It still depends on the number of modules in the project. Usually the programmer changes only one module, so you have to compile only one module, you can hardly parallelize anything. If a full build of a large project, all the modules can be built in parallel HT does not always give a speed gain, because all the cores are using a common CPU cache and shared memory access. If the threads don't have enough cache capacity, will HT get in the way?

Response from 0[+++++]
1. There are different build systems: single-threaded and multi-threaded. Obviously, HT is useless for a single-threaded builder 2. There are different processors. A dual-core processor becomes like a 4-core processor with HT. And a 16-core processor becomes a 32-core processor. But it's not a fact that the builder will need 32 cores, maybe 8 is enough 3 There are different disks: SSD and HDD. If the drive is HDD, it will limit the speed of the build, not the processor It still depends on the number of modules in the project. Usually the programmer changes only one module, so you have to compile only one module, you can hardly parallelize anything. If a full build of a large project, all the modules can be built in parallel HT does not always give a speed gain, because all the cores are using a common CPU cache and shared memory access. If threads don't have enough cache capacity, HT will get in the way.

Response from 0[+++++]
1 Multithreading is always faster than 2 HT in principle gives a boost of about 20 to the CPU performance.

 

To answer the question:

Name*

E-mail:*

Reply text:*
Verification code (enter 22):*