Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
61 Cards in this Set
- Front
- Back
What is principal of locality |
Locality of space state that codes that are close together is likely to be accessed together Locality of time is recently used data is likely to be executed again |
|
3 type of Cache replacement policy |
Least recently used (LRU) Fifo(first in first out) Random replacement policy
|
|
Wad is LRU |
Least recently used data is being evicted, disadvantage is the LRU need to maintain history of the block which slows down the cache |
|
What is FIFO |
First in first out policy, the data that is in the cache the longest will be evicted first |
|
What is random replacement policy |
Random block is being replace by the new block thus it nv thrashes |
|
What is cache thrashing |
Cache blocks being replace frequently |
|
What is cache coherency |
Cache coherency occurs when the cpu writes into the cache but haven't update the main memory |
|
Types of write policy |
Write through policy Write back policy |
|
What is write back policy |
Write back policy only update the memory when the block is selected for replacement |
|
What is write through policy |
Write through policy update both cache n memory simultaneously on every write (no cache coherence issue ) |
|
What is Cache hit |
Data is found in the cache |
|
What is cache miss |
Data nt found in the cache |
|
What is cache hit rate |
The percentage of time data being found in the cache |
|
What is cache miss rate |
The percentage of time data not found in the cache |
|
Formula for cache effective access time |
EAT= Hit rate * cache access time + (1-hit rate) *(cache access time +main memory access time) |
|
What is virtual memory |
Is a technique used to map the virtual memory address to the physical memory address |
|
Advantage of virtual memory |
-Allows system to run programs tat are larger than the size of main memory |
|
What is translation lookaside buffer (TLB) |
Page table is located at the main memory thus access is slow, thus a page cache table (TLB) is used to store the recently page lookout value to speed up the page translation process |
|
Another approach to virtual memory |
Segmentation |
|
What is segmentation |
Instead of dividing the virtual memory into equal size, it divide the virtual memory to variable length segment and access the data through the segment table |
|
What is memory fragmentation |
Memory fragmentation occurs when there are small cluster of unusable memory address being created, paging n segmentation can cause fragmentation |
|
Wat subject to internal fragmentation. WHY |
Paging, a process may not necessary need the entire range of address for a page, thus creating unused fragment of memory |
|
What is seek time |
It is the time taken for the head to move to the correct track |
|
What is rotational delay (Tr) |
Is the time taken to rotate the head to the start position of the sector |
|
What is access time (Tr + Ts) |
Is the time taken from the request to the time the head is in position |
|
What is transfer time (Tt) |
Is the time taken to transfer the required data afterthe head is in position |
|
Formula for T(rav) |
0.5/RPS second |
|
Sample ques |
|
|
What is SSD make up off |
Is made up of NAND N NOR FLASH Is more expensive than Magnetic hard disk Limited erase cycle |
|
Advantage of HDD |
Lower cost per bit Almost infinite erasure cycle |
|
Disadvantage of HDD |
Consist of moving part,prone to data crashing if HDD is shaken or dropped Heavier and larger physical profile Slow transfer rate |
|
Disadvantage of virtual memory |
Take longer access time |
|
3 Advantage of SSD |
Fast transfer rate Lighter, occupy less space More robust to movement (immune to vibration) |
|
Disadvantage of SSD |
Finite erasure cycle Higher cost per bit |
|
Two addressing scheme for HDD |
Cylinder - head sector(CHS ) and logical block addressing(LBA) |
|
What are the technique use to extend the life of the SSD |
Wear LEVELLING Technique And use external ram as a buffer to minimise the number of write to the flash in SSD |
|
What is wear levelling technique |
It is to use to extend the SSD life by distributing the data and erase/write cycle evenly on the entire disk |
|
Formula for probability tat a unit survive |
|
|
Explain Raid level 0 |
It has zero redundancy and have the best performance. It need minimum 2 disk and data is strip as block across all disk But it has no fault tolerance thus lack of reliability |
|
Explain Raid level 1 |
It is a mirrored disks There's 100% redundancy, each data is written identically to each disk. Write through is slower as every time u write data it will update all the disk Need minimum 2 disk |
|
What is the advantage of raid level 1 |
Faster read operations Disk recovery is simple, swap fault disk and remirror |
|
What is the disadvantage of raid 1 |
Expensive Write through is slower as need to update every drive |
|
Explain Raid level 5 |
It is a block level stripping with single distributed parity, data on a single failed drive can be recover through the parity info and surviving data. Is commonly use in network server |
|
Explain Raid 6 |
Is a block level stripping with double distributedparity It can recover from multiple disk failure |
|
Advantage of raid 6 |
Higher data availability |
|
Disadvantage of raid 6 |
Write operation will be slower |
|
Explain Raid 10 |
Is a combination of raid 1 and 0, It will be more expensive as compared to raid 1 n 0 |
|
Why raid is used |
To improve access time and store the data and reserve redundancy so when one of the hard disk fail the data will not be lost |
|
3 type of pipelining conflict |
Resource conflict Data dependencies Branch statement |
|
What is common pitfall |
Common pitfall is to assume tat we can use each metric separately to measure performance |
|
Types of performance metric |
Clock frequency Millions of instruction per second (MIPS ) power consumption Execution time (seconds, millisecond |
|
What is the key advantage tat synthetic benchmarks& EEMBC benchmarks have over synthetic benchmark like linpack and whetstone |
There's third party validation of result |
|
What is standard performance evaluation spec(spec?) benchmarks |
It is formed to address the need for objective benchmarks Users run the benchmark and submit the result to spec for review. Spec will then publish the review result on the website |
|
What is embedded microprocessor benchmark consortium (EEMBC) |
It develop benchmarks for embedded system and organise their benchmark to suit various application domain like digital media, java |
|
What are the 3 advantage of polling |
Minimum hardware interface is used Programmer has complete control of the cpu Easy to test n debug
|
|
3 Disadvantage of polling |
Cpu cannot perform other task until data transfer is complete Inefficent use of computer resource Program execution is held up while waiting for I/o device to be ready |
|
3 Advantage of interrupt triggered technique |
Efficient use of cpu resource Cpu can perform other tasks while transferring data Allow prioritisation and pre emption |
|
2 Disadvantage of interrupt trigger |
More hardware interface is needed More complex and difficult to debug |
|
What is burst mode |
It is also noe as block transfer mode. It take control of the data bus and only return the control after it has finish the data transfer. It allows fast data transfer but cpu maybe suspended for a long period of time Suitable for HDD file transfer |
|
What is cycle stealing |
It will release the control of data bus transferring one word of data. Cpu maybe be suspended but only for a short period of time. It has slower transfer rate compare to burst mode but has the better cpu respond. It requires more complex hardware |
|
What is transparent mode |
It only take control of the data bus when cpu is not using. It has zero impact on cpu responds but it has the slowest transfer rate among the 3 mode. More complex data is needed to check when the cpu is not using |