Computer architecture-Diagram

PAE: Physical Address Extension:
32bit+4=64g

32-bit operating system: 1G kernel +3g user

Cache: Faster than original storage, but small, because expensive, program run locality

Cache substitution algorithm: LRU MRU

Program Operation locality: Spatial locality, temporal locality

1. Registers

2. First level instruction cache, one level data cache

3. Level Two cache

4. Level three cache (shared)

1. Direct mapping (low performance, RAM >> cache)

2.N Link Association

5.CPU deals directly with Level 1 cache, its level two cache, level three cache to 1 level cache

6. Write through: Pass write, CPU update cache, all caches are also updated at the same time

Write back: writeback, updating other caches when the CPU discards the data in the cache

7. Device Identification: IO port

Device on PC boot BIOS detect, register IO port

8. Interrupt Controller: Accept external device request and interrupt processing to CPU request

9. Linear Low address space 16M for DMA use

10.BIOS: Load the lowest end of linear address

The 11.CPU frequency is greater than the memory frequency and needs to read the memory data at a certain scale after the rising delay

12.process: A separate operating unit

System resources: CPU time, storage space

13.os:vm

Slicing the CPU time

14. Instruction Register: Point to the next command address

15. Process switch: Save the scene, restore the scene

16. Divide the memory into a fixed unit [4K ...]

Each page is called a page box

Process virtual Address Space page mapped to page box via MMU

MMU: With protection resources, mapping resources

Mapping relationship: Saved in a level directory, level two directory, level three directory

Process priority: the higher the priority, the more priority the scheduling, the more time

17. Process classification:

Every Linux process is all scheduled according to one of the following scheduling Classes:sched_fifo [1-99]first-in, first-out Real-time process Sched_rrround Robin real-time process sched_normal (100-139) Conventional, time-shared Proces s ather named by sched_otherfor Normal processes priority adjustment: Sched_fifochrt-f [1-99]/path/to/program arguments Sched_rrchrt -R [1-99]/path/to/program arguments sched_other (sched_normal) Nicerenice in Linux, process priority is dynamicthe Sche Duler keeps track of what processes is doing and adjusts their priorities periodicallyprocesses that has been denied the Use of a CPU for a long time interval is boosted by dynamically increasing their priorityprocesses running for a long ti Me is penalized by decreasing their

18.

preemption tick: Clock interrupt 100hz1000hz RHEL 6.4tick lessinterrupt-driven hard Interrupt soft interrupt deep sleep

Computer architecture-Diagram